qualcomm hsdpa

HSDPA Protocols and Physical Layer

UMTS University

WCDMA (UMTS) HSDPA: Protocols and Physical Layer


Student GuideBook 1

80-W0331-1 Rev B

Export of this technology may be controlled by the United States Government. Diversion contrary to U.S. law prohibited.

QUALCOMM is a registered trademark and registered service mark of QUALCOMM Incorporated. gpsOne, QCTest, repeaterOne, and Retriever are trademarks of QUALCOMM Incorporated.

cdma2000® is a registered certification mark of the Telecommunications Industry Association. Used under license. All other trademarks and registered trademarks are the property of their respective owners.

Material Use RestrictionsThese written materials are to be used only in conjunction with the associated instructor-led class. They are not intended to be used solely as reference material.

No part of these written materials may be used or reproduced in any manner whatsoever without the written permission of QUALCOMM Incorporated.

Copyright © 2005 QUALCOMM Incorporated. All rights reserved.

QUALCOMM Incorporated5775 Morehouse DriveSan Diego, CA 92121U.S.A.

© 2005 QUALCOMM Incorporated

Table of Contents – Book 1WCDMA (UMTS) HSDPA: Protocols and Physical Layer 80-W0331-1 Rev B


UMTS UniversityUMTS.HELP

[email protected]

Email hotline resource to assist our UMTS customers worldwide.

Experienced UMTS engineers in our Engineering Services Group will answer your technical questions on topics including:

–Industry Standards–Infrastructure Design–Voice Quality–System Design

–Network Planning–Network Optimization–Test Engineering–Training

iii



This page left blank intentionally.

iv



v

Table of Contents Section 1: Introduction .......................................................................... 1-1 HSDPA Protocols and Physical Layer – Course Overview..................... 1-2 Section Learning Objectives .................................................................... 1-3 High Speed Downlink Packet Access (HSDPA) ..................................... 1-4 WCDMA Evolution................................................................................. 1-5 UMTS Standards...................................................................................... 1-6 Release 99 Packet Data ............................................................................ 1-7 Release 99 Downlink Limitations............................................................ 1-8 High Speed Downlink Packet Access (HSDPA) ..................................... 1-9 HSDPA Basic Concepts......................................................................... 1-10 Comparison Summary ........................................................................... 1-11 Introduction – What We Learned .......................................................... 1-12 Section 2: HSDPA Concepts ................................................................. 2-1 Section Learning Objectives .................................................................... 2-2 References .......................................................................................... 2-3 UMTS Network Architecture .................................................................. 2-4 UMTS Network Architecture with HSDPA ............................................ 2-5 UMTS Protocol Stack.............................................................................. 2-6 HSDPA Protocol Stack ............................................................................ 2-7 Release 99 Channels ................................................................................ 2-8 HSDPA Channels..................................................................................... 2-9 HSDPA Channel Timing ....................................................................... 2-11 HS-DPCCH ........................................................................................ 2-12 HS-SCCH ........................................................................................ 2-13 HS-PDSCH ........................................................................................ 2-14 Data Rate Quiz 1.................................................................................... 2-15 Data Rate Quiz 1 – Answer ................................................................... 2-16 Theoretical HSDPA Maximum Data Rate............................................. 2-17 Multi-code Transmission ....................................................................... 2-18 Consecutive Assignments ...................................................................... 2-19 Hybrid Automatic Repeat Request (HARQ) ......................................... 2-20 Lower Coding Gain................................................................................ 2-21 16-QAM ........................................................................................ 2-23 Theoretical HSDPA Maximum Data Rate............................................. 2-24 Data Rate Quiz 2.................................................................................... 2-25 Answer ....................................................................................... 2-26 More Data Rate Factors ......................................................................... 2-27 Inter-TTI Interval ................................................................................... 2-28 Retransmissions ..................................................................................... 2-29 ACK/NAK Repetitions .......................................................................... 2-30 Node B Considerations .......................................................................... 2-31 OVSF Allocation ................................................................................... 2-32



vi

Node B Transmit Power Allocation....................................................... 2-33 CQI Report Processing .......................................................................... 2-34 Node B Scheduler .................................................................................. 2-35 HSDPA Cell Re-pointing Procedure ..................................................... 2-36 Review Quiz ........................................................................................ 2-37 Answers ..................................................................................... 2-38 HSDPA Concepts – What We Learned ................................................. 2-41 Section 3: Physical Layer Channels ..................................................... 3-1 Section Learning Objectives .................................................................... 3-2 References .......................................................................................... 3-3 HSDPA Physical Layer Model Downlink...................................................................................... 3-4 Uplink .......................................................................................... 3-5 Physical Layer Frame Timing.................................................................. 3-6 Downlink HS-PDSCH ............................................................................. 3-7 HS-DSCH Channel Coding and Physical Channel Mapping .................. 3-8 HS-DSCH Channel Coding ..................................................................... 3-9 Physical Layer HARQ Functionality......................................... 3-12 HARQ Combining Schemes ...................................................... 3-13 Segmentation and Interleaving .................................................. 3-14 16-QAM Constellation Rearrangement ..................................... 3-15 HS-DSCH Physical Channel Mapping .................................................. 3-16 Downlink HS-SCCH.............................................................................. 3-17 HS-SCCH Channel Coding.................................................................... 3-18 HS-PDSCH and HS-SCCH Spreading and Modulation........................ 3-19 HS-PDSCH and HS-SCCH Timing....................................................... 3-20 Uplink HS-DPCCH ............................................................................... 3-21 HS-DPCCH Channel Coding................................................................. 3-22 HS-DPCCH Spreading and Modulation ................................................ 3-23 HS-DPCCH Timing............................................................................... 3-24 HARQ Transmission Quiz..................................................................... 3-25 HARQ Transmission Quiz – Answers................................................... 3-26 Review Quiz ........................................................................................ 3-27 Answers...................................................................................... 3-29 Physical Layer Channels – What We Learned ...................................... 3-31 Section 4: UE Physical Layer Processing............................................. 4-1 Section Learning Objectives .................................................................... 4-2 UE Physical Channel Processing ............................................................. 4-3 UE HS-SCCH Monitoring ....................................................................... 4-4 UE HS-DSCH Decoding.......................................................................... 4-6 HARQ Processing.................................................................................... 4-7 UL Feedback Signaling............................................................................ 4-8 Channel Quality Indicator (CQI) Measurement ...................................... 4-9 CQI Reporting ........................................................................................ 4-10



vii

CQI Mapping Table ............................................................................... 4-11 UE Categories ........................................................................................ 4-12 DL Transmit Diversity........................................................................... 4-13 Review Quiz ........................................................................................ 4-14 Answers...................................................................................... 4-16 UE Physical Layer Processing – What We Learned.............................. 4-18 Section 5: Layer 2 Protocols.................................................................. 5-1 Section Learning Objectives .................................................................... 5-2 References .......................................................................................... 5-3 HSDPA Protocol Stack ............................................................................ 5-4 UTRAN MAC Architecture..................................................................... 5-5 UTRAN MAC-hs Architecture................................................................ 5-6 UE MAC-hs Architecture ........................................................................ 5-7 Data Flow Example.................................................................................. 5-8 RNC MAC-d PDU to Node B Priority Queue............................. 5-9 Node B MAC-hs PDU Assembly .............................................. 5-10 Node B HARQ Process ............................................................. 5-11 UE HARQ Process..................................................................... 5-12 UE Re-ordering Queue .............................................................. 5-13 UE MAC-hs PDU Disassembly................................................. 5-14 MAC-hs Quiz ........................................................................................ 5-15 Answer ....................................................................................... 5-16 MAC Multiplexing................................................................................. 5-17 HARQ Protocol...................................................................................... 5-18 UE HARQ Process Flowchart ............................................................... 5-19 HARQ Protocol Signaling on HS-SCCH .............................................. 5-20 HARQ Protocol Errors .......................................................................... 5-21 HARQ Protocol Error Impact ................................................................ 5-23 Re-ordering Protocol.............................................................................. 5-24 MAC-hs Header ......................................................................... 5-25 In-sequence Delivery of MAC-hs PDUs ................................... 5-26 Re-ordering Protocol Quiz..................................................................... 5-27 Answer ....................................................................................... 5-28 Re-ordering Protocol – Transmit Window ............................................ 5-29 Flushing the Re-ordering Queue Window Method ........................................................................ 5-30 Timer Method ............................................................................ 5-32 Summary of Re-ordering Queue Flushing Methods.............................. 5-33 RLC Considerations............................................................................... 5-34 HSDPA Cell Re-pointing Procedure ......................................... 5-35 Sequence Number Space and Data Rates .................................. 5-36 Review Quiz ........................................................................................ 5-37 Answers ..................................................................................... 5-38 Layer 2 Protocols – What We Learned.................................................. 5-40



viii

Section 6: Layer 3 Protocols and Procedures...................................... 6-1 Section Learning Objectives .................................................................... 6-2 References .......................................................................................... 6-3 HSDPA RRC Functions........................................................................... 6-4 Mobile Originated PS Data Call Setup .................................................... 6-5 Establish HSDPA Operation from Cell_DCH......................................... 6-6 UE Measurements Change of Best Cell (Event 1d) ................................................... 6-7 Measurement Control Message.................................................... 6-8 Event 1d Parameters .................................................................... 6-9 HSDPA Configuration Messages .................................................................................... 6-10 Radio Bearer Reconfiguration Message .................................... 6-11 Layer 1 Information Elements ................................................... 6-12 Layer 2 Information Elements ................................................... 6-13 UE State Transitions .............................................................................. 6-14 Stopping HS-DSCH in Cell_DCH............................................. 6-15 Cell_DCH to Cell_FACH.......................................................... 6-16 Cell_FACH to Cell_DCH with HS-DSCH................................ 6-17 HSDPA Cell Re-pointing Procedure Overview ................................................................................... 6-18 Messages and Information Elements ......................................... 6-19 Synchronized vs. Unsynchronized............................................. 6-20 HSDPA Cell Repointing Procedure Synchronized Inter-Node B ....................................................... 6-21 Unsynchronized Inter-Node B ................................................... 6-22 Problem Areas in Release 5 Signaling................................................... 6-23 Review Quiz ........................................................................................ 6-24 Answers...................................................................................... 6-26 Layer 3 Protocols and Procedures – What We Learned ........................ 6-28 Section 7: Summary............................................................................... 7-1 High Speed Downlink Packet Access (HSDPA) ..................................... 7-2 HSDPA Protocol Stack ............................................................................ 7-3 Theoretical HSDPA Maximum Data Rate............................................... 7-4 HS-DSCH Channel Coding and Physical Channel Mapping .................. 7-5 UE MAC-hs Architecture ........................................................................ 7-6 Re-ordering Protocol – In-sequence Delivery of MAC-hs PDUs ........... 7-7 HSDPA Cell Re-pointing Procedure – Overview ................................... 7-8



Acronyms and Abbreviations 16-QAM 16-Quadrature Amplitude Modulation 3GPP 3rd Generation Partnership Project ACK ACKnowledge ACLR Adjacent Channel Leakage Ratio AICH Acquisition Indicator Channel AM Acknowledged Mode AMC Adaptive Modulation and Coding AMR Adaptive Multi-Rate ARQ Automatic Repeat Request AS Access Stratum AuC Authentication Center BCCH Broadcast Control Channel BCH Broadcast Channel BLER Block Error Rate BTFD Blind Transport Format Detection CC Call Control CCCH Common Control Channel CLTD Closed Loop Transmit Diversity CM Connection Management CPICH Common Pilot Channel CQI Channel Quality Indicator CS Circuit Switched CTCH Common Traffic Channel DCCH Dedicated Control Channel DCH Dedicated Channel DPCCH Dedicated Physical Control Channel DPDCH Dedicated Physical Data Channel DTCH Dedicated Traffic Channel DTX Discontinuous Transmit EDGE Enhanced Data rates for GSM Evolution EUL Enhanced Uplink EVM Error Vector Magnitude FACH Forward Access Channel FEC Forward Error Correction GGSN GPRS Gateway Support Node GMM GPRS Mobility Management GMSC Gateway Mobile Switching Center GPRS General Packet Radio Service GSM Global System for Mobiles GTF Generalized Transport Format H-RNTI High Speed Radio Network Temporary Identity HARQ Hybrid Automatic Repeat Request HLR Home Location Register

ix



HS-DPCCH High Speed Dedicated Physical Control Channel HS-DSCH High Speed Downlink Shared Channel HS-PDSCH High Speed Physical Downlink Shared Channel HS-SCCH High Speed Shared Control Channel HSDPA High Speed Downlink Packet Access IR Incremental Redundancy ISDN Integrated Services Digital Network kbps Kilobits Per Second L1 Physical Layer MAC Medium Access Control MAPL Maximum Allowable Path Loss Mbps Megabits Per Second Mcps Megachips per second MM Mobility Management ms Millisecond MSC Mobile Switching Center NACK Negative Acknowledgment NAK Negative Acknowledgment NAS Non-Access Stratum NDI New Data Indicator OVSF Orthogonal Variable Spreading Factor PA Power Amplifier PAR Peak-to-Average-Power Ratio PCCH Paging Control Channel PCCPCH Primary Common Control Physical Channel PCH Paging Channel PDU Protocol Data Unit PICH Paging Indicator Channel PRACH Physical Random Access Channel PS Packet Switched PSTN Public Switched Telephone Network QID Queue Identifier QoS Quality Of Service QPSK Quadrature Phase Shift Keying RAB Radio Access Bearer RACH Random Access Channel RLC Radio Link Control RM Reed-Muller coding RNC Radio Network Controller ROT Rise Over Thermal RRC Radio Resource Control RSCP Received Signal Code Power RV Redundancy Version SAW Stop And Wait SCCPCH Secondary Common Control Physical Channel SCH Synchronization Channel

x



SF Spreading FactorSGSN Serving GPRS Support Node SID Size Index Identifier SIR Signal-to-Interference Ratio SM Session Management SMS Short Message Service SPER Sub-Packet Error Rate SRB Signal Radio Bearer SS Supplementary Services STTD Space Time Transmit Diversity TFCI Transport Format Combination Indicator TFRC Transport Format Resource Combination TFRI Transport Format Resource Indicator TM Transparent Mode TPC Transmit Power Control TSN Transmission Sequence Number TTI Transmission Time Interval UE User Equipment UM Unacknowledged Mode UMTS Universal Mobile Telecommunications Systems USIM Universal Subscriber Identity Module UTRAN Universal Terrestrial Radio Access Network VF Version Flag VLR Visitor Location Register WCDMA Wideband Code Division Multiple Access

xi



This page left blank intentionally.

xii


1-1

Section 1: Introduction80-W0331-1 Rev B



Section 1-1UMTS University

Section 1:Introduction

1SECTION

Introduction

Notes


1-2





HSDPA Protocols and Physical Layer –Course Overview

1. Introduction2. HSDPA Concepts3. Physical Layer Channels4. UE Physical Layer Processing5. Layer 2 Protocols6. Layer 3 Protocols and Procedures7. Summary

Notes


1-3





Review WCDMA and HSDPA evolution and standards.Review Release 99 packet data service methods.Define HSDPA data transfer model.Describe how HSDPA addresses the Release 99 limitations for packet data service.

Section Learning Objectives

Notes


1-4




Section 1-4UMTS UniversityHigh Speed Downlink Packet Access (HSDPA)

• Data Rate– Demand for high data rate

multimedia services– Demand for higher peak

data rates

• Throughput– Cost per megabyte

• Capacity– Improved Link Adaptation

dependent on Radio Conditions

What are the drivers and motivations for migrating to HSDPA?

Data Services and High Speed Downlink Packet Access (HSDPA)

Data Services are expected to grow significantly within the next few years. Current 2.5G and 3G operators are already reporting that a significant proportion of usage is now due to data, implying an increasing demand for high-data-rate, content-rich multimedia services. Although current Release 99 WCDMA systems offer a maximum practical data rate of 384 kbps, the 3rd Generation Partnership Project (3GPP) have included in Release 5 of the specifications a new high-speed, low-delay feature referred to as High Speed Downlink Packet Access (HSDPA).

HSDPA provides significant enhancements to the Downlink compared to WCDMA Release 99 in terms of peak data rate, cell throughput, and round trip delay. This is achieved through the implementation of a fast channel control and allocation mechanism that employs such features as Adaptive Modulation and Coding and fast Hybrid Automatic Repeat Request (HARQ). Shorter Physical Layer frames are also employed.


1-5




Section 1-5UMTS UniversityWCDMA Evolution

Downlink Peak Data Rate (Typical Deployment)

Downlink Peak Data Rate (Theoretical Maximum)

GSM 9.6 kbps 9.6 kbpsGPRS 40kbps 171 kbpsEDGE 120 kbps 473 kbps

WCDMA Release 99 384 kbps 2.0 MbpsHSDPA 10.0 Mbps 14.4 Mbps

WCDMA Evolution

WCDMA evolved from GSM/GPRS, inheriting much of the upper layer functionality directly from those systems. The first commercial deployments of WCDMA are based on a version of the standards called Release 99.

Enhanced Data rates for GSM Evolution (EDGE) is another system in the GSM/GPRS family that some operators have deployed as an intermediate step before deploying WCDMA.

HSDPA was introduced in WCDMA Release 5 to offer higher speed Downlink data services.

Release 6 introduces the Enhanced Uplink (EUL) that will provide faster data services for the Uplink.


1-6




Section 1-6UMTS UniversityUMTS Standards

Specifications are segmented by layers and are available on the ftp site: ftp://ftp.3gpp.org/Specs/See Rel-5 folder for HSDPA versions.

For a complete list of Release 99 specifications, see 21.101.

For a list of acronyms,see 21.905.

Topic Specifications Series Number

RF Performance 25.1xx

Physical Layer 25.2xx

Layer 2 and Layer 3 25.3xx

UTRAN 25.4xx

NAS Layer (CC, SS, SMS, MM) 22.xxx, 23.xxx, 24.xxx

Packet Switched Data Service 22.060, 23.060

Circuit Switched Data Service 23.910

Voice Service 26.xxx

USIM 31.xxx

UE Conformance 34.xxx

UMTS Standards

The 3rd Generation Partnership Project (3GPP) is responsible for writing and maintaining the UMTS specifications.

Revisions of the specifications are published every three months, as contributing members of 3GPP suggest enhancements and corrections. When a set of features is deemed completed, the current revision is designated as the next release and the specifications are frozen. Corrections are allowed after the release is frozen, but no enhancements may be added.

HSDPA features are described in Release 5.


1-7




Section 1-7UMTS UniversityRelease 99 Packet Data

How is Packet Data handled in Release 99 (FDD)?• DCH (Dedicated Channel)

– Spreading codes assigned per user– Closed loop power control– Macro diversity (soft handover)

• FACH (Common Channel)– Common spreading code– User ID detected by MAC layer– No closed loop power control– No soft handover

Release 99 Packet Data

There are three different techniques defined in the Release 99 specification to enable Downlink packet data. Most commonly, data transmission is supported using either the Dedicated Channel (DCH) or the Forward Access Channel (FACH).

The DCH is the primary means of supporting packet data services. Each user is assigned a unique Orthogonal Variable Spreading Factor (OVSF) code dependent on the required data rate. Fast closed loop Power Control is employed to ensure that a target Signal to Interference Ratio (SIR) is maintained in order to control the block error rate (BLER). Macro Diversity is supported using soft handover.

Data transfer can also be supported on the FACH. This common channel employs a fixed OVSF code. As it needs to be received by all UEs, higher data rates are generally not supported. Macro Diversity is also not supported and the channel operates with a fixed (or slow changing) power allocation. Each data block contains a unique UE identifier that allows a given UE to keep its own data and discard that belonging to other UEs.


1-8




Section 1-8UMTS UniversityRelease 99 Downlink Limitations

Dedicated Channel Features (DCH)• Maximum implemented Downlink of 384 kbps• OVSF Code limitation for high speed users• Rate switching response to bursty throughput is slow• Outer loop power control responds slowly to channel

conditions

Common Channel Features (FACH)• Good for bursty data applications• Only low data rates supported• Fixed transmit power

Release 99 Downlink Limitations

Although WCDMA Release 99 standard allows for maximum data rates of up to 2.0 Mbps, it has only been widely implemented with a maximum data rate of 384 kbps. This data rate is achieved by allocating a dedicated channel to each user. The use of dedicated resources can be a limitation, especially for data applications with bursty characteristics.

Each dedicated channel uses an OVSF code. Shorter codes are used for higher data rates and longer codes for lower data rates. When an OVSF of a particular length is used, all longer OVSF codes derived from that code become unavailable. This limits the number of simultaneous high speed data users in a given cell. The Release 99 standards provide support for a Secondary Scrambling Code, which eases this limitation, but it has not been widely implemented in commercial systems and will likely be removed from future versions of the specification.

The data rate of a dedicated channel can be adjusted to accommodate varying requirements of a data service application, but the procedure for doing so is slow and thus inefficient.

Capacity is controlled both by the maximum amount of PA power that is available and by the power requirement of each data service. In dedicated mode, fast power control is used so that a target Eb/No is achieved on the Downlink. However, the required Eb/No setpoint changes at a much slower rate. This can result in wasted resources whereby a better than required Eb/No is achieved for the required BLER.


1-9





• Set of high speed channels• Channels are shared by multiple users• Each user may be assigned all or part of the total

bandwidth every 2 ms.

High Speed Downlink Packet Access (HSDPA)

In HSDPA, the Node B allocates a set of high speed channels. These channels are assigned to a user using a fast scheduling algorithm that allocates the channels every 2 ms. All or part of the channels may be assigned to a given user during any 2 ms period.

The rapid scheduling of HSDPA is well-suited to the bursty nature of packet data. During periods of high activity, a given user may get a larger percentage of the channel bandwidth, while it gets little or no bandwidth during periods of low activity.


1-10




Section 1-10UMTS UniversityHSDPA Basic Concepts

How will HSDPA address the limitations of Release 99?

• Adaptive modulation and coding– Fast feedback of channel condition– QPSK and 16-QAM– Coding from R=1/3 to R=1

• Multi-Code operation– Multiple codes allocated per user– Fixed spreading factor

• Node B scheduling– Physical Layer HARQ

HSDPA Basic Concepts

In HSDPA a common channel with fixed power is employed for data transfer. Users are separated in both the time and code domains. A fixed spreading factor is employed but multi-code operation is possible for increased data rates. Adaptive Modulation and Coding (AMC) replaces the role of power control so that the modulation and coding rate are changed depending on the channel condition. This is accomplished by locating the scheduling algorithm for channel allocation at the Node B instead of the RNC in Release 99.


1-11




Section 1-11UMTS UniversityComparison Summary

Mode DCH FACH HSDPAChannel Type Dedicated Common Common

Power ControlClosed Inner Loop at 1500 Hz - Slow

Outer LoopNone

Fixed Power with link

adaptationSoft Handover Supported Not Supported Not Supported

Suitability for Bursty Data Poor Good Good

Data Rate Medium Low High

Comparison Summary

DCH and FACH are the two Release 99 channels typically used for packet switched data in practice. The advantages and disadvantages of each approach are apparent. Whereas DCH is suited for medium high data rates (with a maximum rate of 384 kbps), rate switching is slow, making it unsuitable and inefficient for bursty data such as a Web browsing application. By contrast, FACH provides good support for bursty data but is a common channel without power control or other mechanism to account for channel conditions. This makes it unsuitable for higher data rates. Switching from DCH to FACH is slow and inefficient, due in part to the typical timer values used to detect inactivity (may be as much as 5 seconds).

HSDPA is suitable to high date rates for a bursty application, though we will see that the absence of soft handover makes it more suitable for stationary or low-mobility users than for highly mobile users. HSDPA typically operates at a fixed power, but feedback from the UE can instruct the Node B to use lower power when the UE is in good channel conditions. Link adaptation is used to adjust data rate, coding, and modulation to quickly respond to changing channel conditions.


1-12





Introduction –What We Learned

WCDMA and HSDPA evolution and standards. Release 99 data service review.HSDPA data transfer model.HSDPA solutions to Release 99 limitations.

Notes


2-1

Section 2: HSDPA Concepts80-W0331-1 Rev B




Section 2:HSDPA Concepts

2SECTION

HSDPA Concepts

Notes


2-2




Section 2-2UMTS UniversitySection Learning Objectives

Review UMTS network architecture and protocols.Define HSDPA protocol stack.Review UMTS Release 99 channels.Define HSDPA channels.Illustrate theoretical maximum HSDPA data rate.Show how theoretical data rate is reduced to a practical data rate in a real world scenario.Describe Node B enhancements to support HSDPA.

Notes


2-3




Section 2-3UMTS UniversityReferences

3GPP Release 5 Specification References

25.308 HSDPA overall description stage 225.858 HSDPA physical layer aspects

Notes


2-4




Section 2-4UMTS UniversityUMTS Network Architecture

UMTS Network Architecture

A UMTS system consists of three major subsystems:

• User Equipment (UE) – May be a mobile, a fixed station, a data terminal, etc. Includes a Universal Subscriber Identity Module (USIM), which contains all of a user’s subscription information.

• Universal Terrestrial Radio Access Network (UTRAN) – Includes all of the radio equipment necessary for accessing the network. Consists of Node Bs that provide radio links to the UEs, and Radio Network Controllers (RNC) that control the radio resources and interface to the Core Networks.

• Core Network – Includes all of the switching and routing capability for connecting to either the PSTN (circuit-switched calls) or to a Packet Data Network (packet-switched calls), for mobility and subscriber location management, and for authentication services.


2-5




Section 2-5UMTS UniversityUMTS Network Architecture with HSDPA

Core Network

UserEquipment

UTRAN

Mobile EquipmentUSIM

Node B

Node B

Node B

RNC

RNC

HLR/AuC

Node B

Node B

Node B

GMSCPSTN/ISDN

SGSN GGSN Internet

MSC/VLR

Node B

Node B

Uu Iub

Iub

Iups

IucsHardware and Software Changes

Software Changes

Iur

UMTS Network Architecture with HSDPA

Adding HSDPA to an existing UMTS network requires no new network entities, but hardware and/or software changes may be required for each entity. The changes are concentrated in the UE, Node B, and RNC. Interface changes are concentrated on the Uu interface between UE and Node B and on the Iub interface between Node B and RNC.

• UE and Node B – Require hardware and software changes to support the new channels and functionality of HSDPA.

• RNC – Requires software changes to support the new signaling messages used to configure and manage HSDPA channels.

• Uu Interface – Requires new signaling messages exchanged over existing signaling channels and new transport and physical channels to support high-speed operation.

• Iub Interface – Requires a new frame protocol for sending high-speed user data from the RNC to the Node B.

No functional changes to the Iups are required, although there may be bandwidth issues to supporthigher data rates to multiple users.


2-6




Section 2-6UMTS UniversityUMTS Protocol Stack

Mobility Management (MM)

Radio Resources Control (RRC)

Supplementary Services (SS)

Short Message Services (SMS)

Layer 2

Physical Layer (L1)

Non-Access Stratum

Access Stratum

GPRS Mobility Management (GMM)

Session Management (SM)

Radio Link Control (RLC)

Medium Access Control (MAC)

Connection Management (CM)

Call Control (CC)

Short Message Services (SMS)

Circuit Switched Packet Switched

UMTS Protocol Stack

The UMTS signaling protocol stack is divided into Access Stratum (AS) and Non-Access Stratum (NAS). The Non-Access Stratum architecture evolved from the GSM/GPRS upper layers and is divided into Circuit Switched (CS) and Packet Switched (PS) protocols.

The Access Stratum consists of three layers:

• Layer 3 – The Radio Resource Control (RRC) layer handles establishment, release, and configuration of radio resources.

• Layer 2 – Consists of two sublayers. The Radio Link Control (RLC) sublayer provides segmentation, re-assembly, duplicate detection, and other traditional Layer 2 functions. The Medium Access Control (MAC) sublayer multiplexes data and signaling onto the appropriate channels and controls access to the Physical Layer.

• Layer 1 – The Physical Layer transfers data over the radio link.


2-7




Section 2-7UMTS UniversityHSDPA Protocol Stack

HSDPA Protocol Stack

In a Release 99 PS network, the NAS layer protocols are terminated at the SGSN. The RRC, RLC, and MAC protocols are terminated at the RNC. The Physical Layer protocol is terminated at the Node B.

The Release 5 specifications define a new sublayer of MAC called MAC-hs, which implements the MAC protocols and procedures for HSDPA. This sublayer operates at the Node B and the UE.

The location of MAC-hs in Node B has an important implication for HSDPA operation. In Release 99, a UE may be in soft handover with multiple Node Bs. Transport channel frames are constructed by the MAC sublayer in the RNC and sent over the Iub interface to each Node B with which the UE is in soft handover. The UE receives identical Transport channel frames from each Node B.

HSDPA requires fast scheduling of the shared channels, and allocates the channels in 2 ms intervals called subframes. To meet this requirement, the Transport channel frames are constructed by the MAC-hs sublayer operating in the Node B. By design, the HSDPA channels cannot operate in soft handover because the MAC-hs sublayer of each Node B operates independently.


2-8




Section 2-8UMTS UniversityRelease 99 Channels

Release 99 ChannelsThis diagram shows possible mappings of logical, transport, and physical channels in the control and user planes for UMTS Release 99. Some channels exist only in Physical Layer (CPICH, SCH, DPCCH, AICH, PICH). These channels carry no upper layer signaling or user data.Transport channels carry the following types of information:

Broadcast Control Channel (BCH) – Broadcast information that defines overall system configuration.Paging Channel (PCH) – Paging notification messages. A Paging Indicator Channel (PICH) is associated with a PCH to allow a UE to quickly determine whether it needs to read the PCH during its assigned paging occasion.Forward Access Channel (FACH) – Common Downlink signaling messages. Also carries dedicated Downlink signaling and user information to a UE operating in Cell_FACH state. An Acquisition Indicator Channel (AICH) is associated with a FACH channel.Random Access Channel (RACH) – Common Uplink signaling messages. Also carries dedicated Uplink signaling and user information to a UE operating in Cell_FACH state.Dedicated Channel (DCH) – Dedicated signaling and user information for a UE operating in the Cell_DCH state. DCH is mapped to a Dedicated Physical Data Channel (DPDCH). An associated Dedicated Physical Control Channel (DPCCH) carries Physical Layer control information, such as power control commands.


2-9




Section 2-9UMTS UniversityHSDPA Channels

New HSDPA Channels• High Speed Downlink Shared Channel (HS-DSCH)

– Downlink Transport Channel• High Speed Shared Control Channel (HS-SCCH)

– Downlink Control Channel• High Speed Physical Downlink Shared Channel

(HS-PDSCH)– Downlink Physical Channel

• High Speed Dedicated Physical Control Channel (HS-DPCCH)– Uplink Control Channel

HSDPA Channels

HSDPA introduces three new Downlink channels and one new Uplink channel:

• High Speed Downlink Shared Channel (HS-DSCH) – A Downlink transport channel shared by several UEs. The HS-DSCH is associated with one or several Shared Control Channels (HS-SCCH). It operates on a 2 ms Transmission Time Interval (TTI).High Speed Shared Control Channel (HS-SCCH) – A Downlink physical channel used to carry Downlink control information related to HS-DSCH transmission. The UE monitors this channel continuously to determine when to read its data from the HS-DSCH, and the modulation scheme used on the assigned physical channel.

• High Speed Physical Downlink Shared Channel (HS-PDSCH) – A Downlink physical channel shared by several UEs. It supports Quadrature Phase Shift Keying (QPSK) and 16-Quadrature Amplitude Modulation (16-QAM) and multi-code transmission. It is allocated to a user at 2 ms intervals.

• High Speed Dedicated Physical Control Channel (HS-DPCCH) – An Uplink physical channel that carries feedback from the UE to assist the Node B’s scheduling algorithm. The feedback includes a Channel Quality Indicator (CQI) and a positive or negative acknowledgement (ACK/NAK) of a previous HS-DSCH transmission.


2-10




Section 2-10UMTS UniversityHSDPA Channels (continued)

HSDPA Channels (continued)

Only dedicated logical user data channels may be mapped to HS-DSCH. When DTCH is mapped to HS-DSCH, only Unacknowledged Mode (UM) and Acknowledged Mode (AM) channels may be used.

A UE operating in HSDPA mode also has at least one Release 99 dedicated channel (DCH/DPDCH) allocated, to ensure that RRC and NAS signaling can always be sent, even if the UE is not able to receive the high speed channels.

The HS-DPCCH is a Physical Layer control channel. It carries no upper layer information, and therefore has no logical or transport channel mapping.


2-11




Section 2-11UMTS UniversityHSDPA Channel Timing

HSPDA Channel Timing

HSDPD channel timing is based on a time interval of 2 ms, or 3 slots.

This slide illustrates a single HSDPA channel assignment. Consecutive assignments to a single UE allow the theoretical maximum HSDPA data rate to be achieved.

1. The UE measures the Downlink channel quality and sends a CQI report on the HS-DPCCH. An ACK or NAK from a previously received block may also be included in this transmission.

2. If the Node B decides to send data to the UE, it will send information on the HS-SCCH to assign the physical channel and give the UE information about how the data was encoded. The earliest that this assignment can be made is in the subframe following the end of CQI report.

3. During the next 2 ms HS-DSCH transmission time, one or more HS-PDSCHs carry the UE’s data. The HS-SCCH transmission overlaps the HS-PDSCH transmission.

4. After the UE decodes the data, it sends an ACK or NAK on the HS-DPCCH. The UE must send the ACK or NAK 5 ms after the end of the HS-DSCH transmission. If the UE sends a NAK, the Node B may send the data again during a later time slot, or may choose not to retransmit the data. A CQI report may also be included in this transmission.


2-12




Section 2-12UMTS UniversityHS-DPCCH

High Speed Dedicated Physical Control Channel (HS-PDCCH)

• 1st slot carries ACK or NAK for received HS-DSCH blocks• 2nd and 3rd slots carry Channel Quality Indicator (CQI)

– UE measures Downlink CPICH channel quality– CQI indicates the highest data rate for error rate < 10%– Frequency of CQI reports configured by UTRAN

• DTX during ACK/NAK and CQI slots if nothing to send• Uses Spreading Factor = 256

HS-DPCCH

Whenever the UE is operating in HSDPA mode, it uses the HS-DPCCH to give feedback to the serving Node B. This feedback consist of two parts:

ACK/NAK – The UE sends a positive or negative acknowledgement for each HS-DSCH assignment. UTRAN may configure the UE to repeat the ACK/NAK, up to a maximum of 4 transmissions. The first ACK/NAK for a given HS-DSCH assignment is sent 5 ms (7.5 slots) after the end of the HS-DSCH transmission.Channel Quality Indicator (CQI) – The UE measures the channel quality of the Downlink CPICH and computes a CQI value. The value is an index into a table, and corresponds to the maximum data rate that the UE can decode with an error rate of less than 10%, assuming the channel conditions don’t change. UTRAN may configure the UE to repeat the CQI, up to a maximum of 4 transmissions. UTRAN may also configure the periodicity of CQI reporting, ranging from 2 ms to 160 ms.


2-13




Section 2-13UMTS UniversityHS-SCCH

High Speed Shared Control Channel (HS-SCCH)• 1st part carries modulation information

– OVSF code assignment – Modulation scheme

• 2nd part carries transport block size, Hybrid ARQ parameters• UE Identity encoded over each part

– UE decodes each part independently• UE assigned up to 4 HS-SCCHs to monitor• Uses Spreading Factor = 128

HS-SCCH

Whenever the UE is operating in HSDPA mode, it continuously monitors up to four HS-SCCHs. Each HS-SCCH transmission carries scheduling information about the next HS-DSCH assignment and the Physical Layer parameters of the associated HS-PDSCH.

OVSF Code Assignment – The HS-SCCH indicates which of the OVSF codes allocated to the HS-PDSCHs will be used. HS-PDSCH uses multi-code transmission, which means that multiple OVSF codes may be assigned to one UE at the same timeModulation Scheme – HS-PDSCH uses either QPSK or 16-QAM modulation. This can change from one assignment to the next, and HS-SCCH indicates which method will be used.Transport Block Size – The HS-SCCH indicates how much data will be sent during the next assignmentHybrid ARQ (HARQ) Parameters – The HARQ protocol supports retransmissions and incremental redundancy. These parameters allow the UE to differentiate new transmissions from retransmissions.UE Identity – Multiple UEs may be monitoring the same set of HS-SCCHs. Each UE has an assigned identity called the H-RNTI. The first part of the HS-SCCH is scrambled using the H-RNTI so that an UE can determine whether the corresponding HS-DSCH assignment carries its data or data belonging to another UE. The second part contains additional information to allow the UE to decode the block, and a CRC masked with the H-RNTI.


2-14




Section 2-14UMTS UniversityHS-PDSCH

High Speed Physical Downlink Shared Channel (HS-PDSCH)• Carries UE data• Up to 15 HS-PDSCH may be assigned simultaneously

– UE capability indicates maximum number of codes it supports• Uses Spreading Factor = 16

HS-PDSCH

When the UE decodes the HS-SCCH and determines that there is an HS-DSCH assignment in the next TTI, it decodes the assigned HS-PDSCHs. Each HS-PDSCH uses an OVSF of length 16. If multiple HS-PDSCHs are assigned simultaneously to one UE, they must use consecutive OVSF codes. The HS-SCCH indicates the first OVSF code and the number of codes for each assignment.

A UE is a member of one of 12 categories, as a function of its hardware capabilities. Each category represents different values of the following parameters:

Number of simultaneous HS-PDSCH codes (5, 10, or 15)Maximum transport block sizeInter-TTI interval – minimum time between consecutive assignments.Incremental redundancy buffer size – used to soft-combine symbols from retransmissions.


2-15




Section 2-15UMTS UniversityData Rate Quiz 1

Question:Assuming a transport block size of 320 bits, what HSDPA data rate can be achieved by a single UE using the channel allocation timing shown above?

Data Rate Quiz 1

Determine the HSDPA data rate achieved by a single UE, assuming the following parameters:

1. Each HSDPA assignment is for a single HS-PDSCH (no multi-code).2. Each HS-PDSCH block carries 320 bits.3. Each transport block is successfully decoded after the first transmission (the UE always

sends an ACK and the Node B never retransmits any block).4. The Node B schedules an assignment as early as possible following the ACK

transmission, as shown in the above timing diagram.


2-16




Section 2-16UMTS UniversityData Rate Quiz 1 – Answer

Answer:320 bits are transmitted every 10 ms, so the maximum data rate is 32 kbps.

Data Rate Quiz 1 – Answer

Using the allocation scheme and block size given in this example, the UE achieves only 32 kbps! Obviously, this is not the whole story of HSDPA.


2-17




Section 2-17UMTS UniversityTheoretical HSDPA Maximum Data Rate

Theoretical Maximum is 14.4 Mbps!

How do we get from 32 kbps to 14.4 Mbps?• Multi-code transmission• Consecutive assignments using multiple Hybrid

Automatic Repeat Request (HARQ) processes• Lower coding gain• 16-QAM

Theoretical HSDPA Maximum Data Rate

The theoretical maximum data rate is 14.4 Mbps. The following techniques are used to achieve this data rate:

Multi-code transmission – Up to 15 HS-PDSCH channels may be assigned to a single UE during one 2 ms TTI.Consecutive assignments – The HARQ procedure allows the Node B to send back-to-back assignments at 2 ms intervals.Lower Coding Gain – The block size of 320 bits was chosen assuming a turbo code rate of 1/3. Higher data rates can be achieved by puncturing more bits for a higher effective code rate (and thus lower coding gain).16-QAM – This modulation scheme increases the data rate over QPSK by a factor of 2.


2-18




Section 2-18UMTS UniversityMulti-code Transmission

Data Rate with 15-code Multi-code32 kbps X 15 = 480 kbps

Multi-code Transmission

HSDPA allows up to 15-code multi-code. Each HS-PDSCH uses an OVSF of length 16. The Node B signals the number of codes to the UE in the HS-SCCH.

The number of codes supported by the UE is one factor in determining the UE’s HSDPA category. The allowed choices are 5, 10, or 15 codes.

For a UE capable of the maximum number of codes, the data rate in the above example is 15 times greater than the single code assignment, or 480 kbps.


2-19




Section 2-19UMTS UniversityConsecutive Assignments

Data Rate with Consecutive Assignments480 kbps X 5 = 2.4 Mbps

Consecutive Assignments

HSDPA allows the channels to be assigned in consecutive TTIs to the same UE. In the UE, up to six simultaneous HARQ processes operate in parallel to decode consecutive assignments. Each HARQ process is responsible for decoding one assignment, and transmitting the associated ACK or NAK 5 ms after the end of that assignment.

The UE can achieve a maximum data rate that is five times greater than in the previous example, or 2.4 Mbps, if all of the following conditions are met:

The UE supports 15-code multi-code.The Node B assigns all 15 OVSF codes every TTI.Every data block is correctly decoded (the UE always sends an ACK).


2-20




Section 2-20UMTS UniversityHybrid Automatic Repeat Request (HARQ)

Hybrid Automatic Repeat Request (HARQ)• Each HSDPA assignment is handled by a HARQ process

– HARQ Processes run in Node B and UE– Up to 8 HARQ processes per UE – Number configured by Node B when HSDPA operations begin

• The UE HARQ process is responsible for:– Attempting to decode the data– Deciding whether to send ACK or NAK– Soft-combining of retransmitted data

• The Node B HARQ process is responsible for:– Selecting the correct bits to send according to the selected

retransmission scheme and UE capability

Hybrid Automatic Repeat Request (HARQ)

To support consecutive assignments, HSDPA defines a Hybrid Automatic Repeat Request (HARQ) protocol. This protocol is implemented in both the Node B and the UE, and consists of procedures implemented in both the MAC-hs sublayer and the Physical Layer.

When the Node B assigns an HSDPA subframe to a UE, it also assigns a HARQ process to handle the data transfer. The UE HARQ process is responsible for

Decoding the initial transmissionSending an ACK or NAKSoft-combining retransmissions of the data packet until it is successfully decoded or until Node B aborts the packet.

Up to 8 HARQ processes may run simultaneously. At least 6 simultaneous processes are required to sustain consecutive HSDPA assignments. Depending on its implementation, the Node B scheduler algorithm may require more than 6 HARQ processes to sustain consecutive assignments. When HSDPA operations begin, the RNC configures the number of HARQ processes in a signaling message to the UE.

The maximum number of HARQ processes that a UE supports is a function of its HSDPA category. The minimum number of HARQ processes supported by any UE is 2, which corresponds to a UE that uses an inter-TTI interval of 3.


2-21




Section 2-21UMTS UniversityLower Coding Gain

R=1/3 Turbo Coding and QPSK Modulation

Lower Coding Gain

All examples so far have assumed a turbo code rate of 1/3 and QPSK modulation. If we assume a single HS-PDSCH and a transport block containing 320 data bits, rate 1/3 turbo coding produces 960 symbols. QPSK modulation maps two symbols onto one modulation symbol, which then gets spread by the OVSF of length 16. This results in 7680 chips sent every 2 ms, corresponding to the fundamental WCDMA chip rate of 3.84 Mcps.

If the transport block is not exactly 320 data bits, the rate matching step adjusts the number of symbols after turbo coding to produce 960 symbols.

If multiple HS-PDSCHs are used, the rate matching step produces an integer multiple of 960 symbols, and blocks of 960 symbols are mapped to each HS-PDSCH.


2-22




Section 2-22UMTS UniversityLower Coding Gain (continued)

Data Rate with Rate 1 Turbo Coding and QPSK Modulation2.4 Mbps X 3 = 7.2 Mbps

Lower Coding Gain (continued)

HSDPA allows the initial transmission of a data block to contain no parity bits, only systematic bits. Systematic bits are the original data bits that are input into the turbo encoder. Sending only systematic bits produces an effective code rate of 1, resulting in a data rate 3 times the previous example, or 7.2 Mbps.

The HARQ procedure provides a mechanism for sending the parity bits in a later assignment if the UE is not able to decode the block using only systematic bits; however this will reduce the UE’s overall throughput.


2-23




Section 2-23UMTS University16-QAM

Data Rate with 16-QAM7.2 Mbps X 2 = 14.4 Mbps

16-QAM

HSDPA supports higher order modulation on HS-PDSCH. Where QPSK maps 2 bits onto one of 4 modulation symbols, 16-QAM maps 4 bits onto one of 16 modulation symbols. The resultingdata rate is two times that of QPSK, or 14.4 Mbps.

16-QAM is more sensitive to both channel conditions and interference than QPSK, and therefore is only useful in very good channel conditions (e.g., close to the cell site, low speed).

Support for 16-QAM is one factor in determining the UE’s HSDPA category.


2-24




Section 2-24UMTS UniversityTheoretical HSDPA Maximum Data Rate

Review: How do we get to 14.4 Mbps?• Multi-code transmission

– Node B must allocate all 15 OVSF codes of length 16 to one UE

• Consecutive assignments– Node B must allocate all time slots to one UE– UE must decode all transmissions correctly on the first transmission

• Lower Coding Gain– Effective code rate = 1

– Requires very good channel conditions to decode

• 16-QAM– Requires very good channel conditions


The following assumptions are needed to achieve the theoretical maximum data rate of 14.4 Mbps:

Multi-code transmission – All 15 HS-PDSCH channels must be assigned to a single UE during one 2 ms TTI. This uses up a significant portion of the OVSF tree, leaving very few codes for non-HSDPA users and overhead channels.Consecutive assignments – The Node B must send back-to-back assignments to a single UE, and the UE must be able to correctly decode every block without requiring retransmission.Lower Coding Gain – Using an effective code rate of 1 increases the data rate, but the channel conditions must be very good for the UE to correctly decode every data block on the first transmission.16-QAM – This modulation scheme works well only in very good channel conditions.

In a practical scenario, the practical maximum data rate will be considerably less than 14.4 Mbps, due to less than ideal channel conditions, the need for retransmission, and the need to share the channel with other HSDPA users and Release 99 users.

Other factors that reduce the practical maximum data rate will be discussed in subsequent slides.


2-25




Section 2-25UMTS UniversityData Rate Quiz 2

Calculate the data rate for one UE assuming:• 5 OVSF codes• Consecutive assignments• QPSK modulation• Turbo Code R = 1/3• Retransmission statistics:

– 75% of blocks decoded on first transmission– 25% of blocks decoded on second transmission

Data Rate Quiz 2

Calculate the data rate achieved by one UE using the assumptions given above.


2-26




Section 2-26UMTS UniversityData Rate Quiz 2 – Answer

Start from 14.4 Mbps and decrease• 5 code Multi-code

– 14.4 Mbps / 3 = 4.8 Mbps• QPSK modulation

– 4.8 Mbps / 2 = 2.4 Mbps• Turbo Code R = 1/3

– 2.4 Mbps / 3 = 800 kbps• Retransmission statistics

– 800 kbps * 0.8 = 640 kbps

Data Rate Quiz 2 – Answer

Starting with the 14.4 Mbps figure, decrease the data rate for each of the stated assumptions:

5-code multi-code – Using 5 codes instead of 15 reduces the data rate by a factor of 3, to 4.8 Mbps.QPSK Modulation – Using QPSK instead of 16-QAM reduces the data rate by a factor of 2, to 2.4 Mbps.Turbo Code R = 1/3 – Using R = 1/3 decreases the data rate by a factor of 3, to 800 kbps.Retransmission Statistics – If every 4th block needs to be retransmitted, the data rate is reduced by 20%, to 640 kbps.


2-27




Section 2-27UMTS UniversityMore Data Rate Factors

More Factors that Affect Data Rate• Inter-TTI Interval• Retransmissions• ACK/NAK Repetition

More Data Rate Factors

Other factors that influence the maximum data rate are:

Inter-TTI Interval – The interval between consecutive assignments is called the inter-TTI interval. If the UE supports an inter-TTI interval of 1, then it is capable of receiving a new HSDPA assignment every 2 ms. Allowed values of the inter-TTI interval are 1, 2, and 3.Retransmissions – If the UE NAKs a transmission, the Node B may retransmit that data in a subsequent assignment. The retransmission may consist of identical symbols that were sent previously, or may be a different redundancy version of the turbo coded output symbols.ACK/NAK Repetition – The Node B may configure the UE to send the ACK/NAK transmission up to four times.


2-28




Section 2-28UMTS UniversityInter-TTI Interval

Inter-TTI Interval = 2

Inter-TTI Interval

One parameter of the UE’s HSDPA capability is its inter-TTI interval. This parameter determines the interval between consecutive assignments that the UE is capable of decoding. Allowed values are 1, 2, and 3.

The diagram above illustrates inter-TTI interval equal to 2. This reduces by half the maximum data rate achieved by the UE, all other parameters being equal.

The UE’s signals its HSDPA capability to the Node B before beginning HSDPA operation, to allow the Node B to correctly schedule HSDPA assignments.


2-29




Section 2-29UMTS UniversityRetransmissions

Retransmissions

If the UE is unable to decode an HSDPA data block, it sends a NAK 5 ms after the end of the received block. The Node B may choose to retransmit the data as early as the next HS-SCCH assignment following the NAK. The earliest a retransmitted block may be sent is 10 ms after the beginning subframe boundary of the previous transmission.

The retransmitted block may be identical to the previous transmission, or it may be a different redundancy version. This means that a different combination of systematic and parity bits are sent. In either case, the UE retains the symbols from the first transmission and uses either Chase combining or incremental redundancy to increase the probability that the data will be decoded correctly on the 2nd attempt.

Retransmissions decrease the data rate, as the retransmitted data occupies an interval that would otherwise be used to transmit new data.


2-30




Section 2-30UMTS UniversityACK/NAK Repetitions

ACK/NAK Repetitions

The Node B may configure the UE to transmit the ACK/NAK up to four times, to increase the reliability of decoding the ACK/NAK. Using an ACK/NAK repetition greater than one has the same effect on data rate as the UE’s inter-TTI interval.

In the example shown above, the Node B cannot transmit a data block to the same UE in subframe 2, because the ACK/NAK slot for that subframe is occupied by the repetition of the ACK/NAK corresponding to subframe 1. It may, of course, send data to a different UE in subframe 2.

The Node B signals the ACK/NAK repetition rate to the UE before the UE begins HSDPA operation.


2-31




Section 2-31UMTS UniversityNode B Considerations

Node B Considerations• OVSF Code Allocation• Power Allocation• CQI Report Processing• Scheduler• HSDPA Cell Re-pointing Procedure• Compressed Mode

Node B Considerations

Most of the changes to support HSDPA on network side are implemented in the Node B. Things to consider are:

OVSF Code Allocation – HSDPA uses OVSF codes of length 16. The number of HS-PDSCH codes allocated affects the number of other users that can be supported for Release 99 operations (including voice).Power Allocation – HSDPA channels may be allocated all the remaining transmit power on a 2 ms basis.CQI Report Processing – Node B uses the CQI reports from the UE to determine when to schedule the HSDPA channels and what data rate to use.Scheduler – The scheduler in the Node B must allocate the channels as a function of the number of HSDPA users in the cell, the channel conditions reported by each user, and available transmit power.HSDPA Cell Re-pointing Procedure – HSDPA channels do not operate in soft handover, but there is a mechanism to re-point the serving Node B to support the UE’s mobility.Compressed Mode – The Node B should not schedule an HSDPA assignment during a UE’s compressed mode gaps.


2-32




Section 2-32UMTS UniversityOVSF Allocation

OVSF Allocation

Each HS-PDSCH uses an OVSF of length 16, which blocks all codes above and below it in the OVSF code tree. Each HS-SCCH uses an OVSF of length 128.

The illustration above shows a possible OVSF allocation if 15 HS-PDSCH codes are used and only 1 HS-SCCH.

If only one HS-SCCH is used, then only one UE can operate in HSDPA during each 2 ms TTI.The overhead channels CPICH, PICH, AICH, and PCCPCH require codes of length 256. SCCPCH spreading factor is configurable, but SF = 128 is typical. Each HSDPA user requires a DPCH in addition to its high speed channel. The spreading factor of this channel is configurable.If voice users are supported in the same cell, they typically use codes with SF = 128.

The conclusion to be drawn is that using 15 HS-PDSCH codes is not practical unless the cell is dedicated to HSDPA users.


2-33




Section 2-33UMTS UniversityNode B Transmit Power Allocation

Node B Transmit Power Allocation

The Node B transmit power allocation algorithm is not specified by the standard, but two possible schemes are likely:

Static – A fixed amount of power is allocated to the HS-PDSCHs and HS-SCCHs. Remaining power is distributed among common channels and power controlled dedicated channels. The overall transmit power fluctuates as a function of the power controlled channels.Dynamic – HS-PDSCH and HS-SCCH power is allocated dynamically as a function of the remaining available power, which fluctuates due to the power controlled dedicated channels. The overall transmit power of the cell remains constant.

The above diagram does not consider the Node B’s power margin, whereby the Node B’s power fluctuates. The Node B power doesn’t really remain constant, due to the peak-to-average ratio of transmit power.


2-34




Section 2-34UMTS UniversityCQI Report Processing

• UE measures CPICH strength– Measurement reference period is 3 slots, ending 1 slot before

CQI is sent• UE reports index into CQI Table

– Highest data rate for which UE can guarantee error rate < 10%• Node B may filter CQI reports

– Varying CQI means UE is in a fast changing environment– Steady CQI means UE is in a stable environment

CQI Report Processing

The Node B may use the UE’s CQI reports in its scheduling algorithm. The details of this isimplementation dependent.

When the UE is required to perform CQI reporting, the measurement reference period consists of 3 slots ending 1 slot before the CQI is sent. The value reported is an index into a table, where each row of the table maps to a combination of:

Transport block sizeNumber of HS-PDSCH codesModulation Scheme (QPSK or 16-QAM)Reference power adjustment

The CQI reported corresponds to the highest data rate that the UE can decode with an error rate less than 10%, assuming the channel conditions and transmit power stay at the same level as in the reference period.

The reference power adjustment maps to a negative value when the channel conditions are so good that the UE can decode the highest data rate at a lower power level than is currently being used.


2-35




Section 2-35UMTS UniversityNode B Scheduler

Node B Scheduler

The Node B scheduler is responsible for deciding how to allocate the available HSDPA channels and transmit power among users. The standard puts no requirements on this algorithm, leaving it entirely implementation dependent.

Some possible schemes:

Round Robin – Each user is allocated the channel in a fixed rotation. The scheme could be simple, or modified to account for CQI and/or user priorities.Proportional Fair – Each user sees a throughput proportional to the peak rate that its link can sustain.CQI Based – Channel is allocated to the user in the best radio condition. This scheme provides the highest cell throughput, though at the cost of not serving users in located in areas of poor coverage.

Scheduling algorithms for systems such as HSDPA are the subject of much research and analysis in the wireless industry.


2-36




Section 2-36UMTS UniversityHSDPA Cell Re-pointing Procedure

HSDPA Cell Re-pointing Procedure

HSDPA channels do not operate in soft handover. For a given UE, the Node B from which it receives the HSDPA channels is called the Serving Node B.

The UE may be in soft handover on the associated DPCH.

If the radio conditions change such that there is a better cell on another Node B for HSDPA operations, the HSDPA Cell Re-pointing Procedure is performed. This procedure occurs independently from the Active Set update procedure.


2-37




Section 2-37UMTS UniversityReview Quiz

1. List the HSDPA channels and describe the purpose of each.

2. What is the fundamental time unit of HSDPA channels?

3. List five assumptions necessary to achieve the theoretical maximum data rate of 14.4 Mbps.

4. What effect does an inter-TTI interval of 3 have on the data rate?

5. What effect does an ACK/NAK repetition factor of 4 have on data rate?

6. What is the minimum interval between first transmission and retransmission of the data block?

7. List three parameters that may be used by the Node B scheduler to allocate the channel.

8. How many HS-SCCHs are needed if HSDPA channels are allocated to three users simultaneously?

Notes


2-38




Section 2-38UMTS UniversityReview Quiz – Answers

1. List the HSDPA channels and describe the purpose of each.• HS-DPCCH Uplink channel carries ACK/NAK and CQI report• HS-SCCH Downlink channel assigns HSDPA channel to a user• HS-DSCH Downlink transport channel that carries HSDPA data• HS-PDSCH Downlink physical channel that carries HSDPA data

2. What is the fundamental time unit of HSDPA channels?• 2 ms

3. List five assumptions necessary to achieve the theoretical maximum data rate of 14.4 Mbps.

• 15 OVSF codes allocated• Consecutive assignment and inter-TTI interval = 1• Turbo code effective rate = 1• 16-QAM• All data decoded correctly on first transmission

Notes


2-39





4. What effect does an inter-TTI interval of 3 have on the data rate?• Reduces by a factor of 3

5. What effect does an ACK/NAK repetition factor of 4 have on data rate?

• Reduces by a factor of 4

6. What is the minimum interval between first transmission and retransmission of the a data block?

• 10 ms

Notes


2-40





7. List three parameters that may be used by the Node B scheduler to allocate the channel.

• CQI Report• Code Tree Utilization• User priority

8. How many HS-SCCHs are needed if HSDPA channels are allocated to three users simultaneously?

• 3

Notes


2-41





HSDPA Concepts –What We Learned

UMTS network architecture and protocols.HSDPA protocol stack.UMTS Release 99 channels.HSDPA channels.Theoretical maximum HSDPA data rate.Practical maximum HSDPA data rate.Node B enhancements to support HSDPA.

Notes


2-42



Comments/Notes


3-1

Section 3: Physical Layer Channels80-W0331-1 Rev B




Section 3:Physical Layer Channels

3SECTION

Physical Layer Channels

Notes


3-2





Identify HSDPA Physical Layer channels.Describe channel coding for individual channels.Understand the timing relation of HSDPA channels.

Notes


3-3






25.211 Physical channels and mapping of transport channels onto physical channels (FDD)

25.212 Multiplexing and channel coding (FDD)25.213 Spreading and modulation (FDD)25.214 Physical layer procedures (FDD)25.306 UE Radio Access Capabilities25.308 HSDPA overall description stage 225.321 Medium Access Control (MAC) protocol specification25.331 Radio Resource Control (RRC) protocol specification25.858 HSDPA physical layer aspects

References

Notes


3-4




Section 3-4UMTS UniversityHSDPA Physical Layer Model – Downlink

Node B PHY

HS-DSCH coding and modulation

HS-DSCH associated control signaling

HS-PDSCH HS-PDSCH

…

HS-DSCH

HS-SCCH

UE PHY

HS-DSCH decoding and demodulation

HS-DSCH associated control information

HS-PDSCH HS-PDSCH

…

HS-DSCH

HS-SCCH

HS-PDSCH – carries actual information payload from HS-DSCHHS-SCCH – carries Physical Layer control information including HARQ parameters, channelization codes, and UE ID

HSDPA Physical Layer Model – Downlink

In 3GPP Release 5, two new Downlink physical channels have been introduced to enable HSDPA. In addition, the existing R99 channels are also required for HSDPA operation.

HS-PDSCH – Transmitted by Node B to send HS-DSCH data to UEs in the HSDPA serving cell. Unlike a dedicated channel, this shared channel is assigned to a user for a 2 ms period and may be assigned to another user in the next 2 ms period. This fast user switching rate is well suited for the bursty packet data and helps increase the capacity of a cell. There can be multiple (up to 15) HS-PDSCHs in a serving cell, which enables use of both time division and code division multiple access methods. HS-PDSCH carries user data and has a transport channel HS-DSCH mapped on it.

HS-SCCH – Transmitted by Node B to signal control information to the users in the HSDPA serving cell. This channel is shared by multiple users and the control information sent on it is masked with a UE ID. The mask allows a UE to identify if there is HS-DSCH data for it in the upcoming HS-PDSCH subframe and the control information tells how to decode that data. The control information is transmitted in two parts:

Part 1 consists of HS-PDSCH channelization codes and modulation scheme

Part 2 consists of HARQ parameters and HS-DSCH transport block size

HS-SCCH does not have a transport channel mapped on it.


3-5




Section 3-5UMTS UniversityHSDPA Physical Layer Model – Uplink

UE PHY

DCH coding and modulation

HARQ and channelquality feedback

signaling

UL DPCH

DCH

HS-DPCCH

Node B PHY

DCH decoding and demodulation

HARQ and channel quality feedback

information

UL DPCH

DCH

HS-DPCCH

HS-DPCCH – carries feedback signaling consisting of HARQ acknowledgement and channel quality indicator (CQI)

HSDPA Physical Layer Model – Uplink

In 3GPP Release 5, there is one new Uplink physical channel. The existing R99 channels are required for the HSDPA operation.

HS-DPCCH – Transmitted by the UE to signal feedback information to Node B. The feedback information consists of:

acknowledgement of data received by the UE on HS-PDSCH

Downlink channel quality indicator (CQI)

Node B uses this feedback information to send retransmissions and to schedule HS-PDSCH transmissions to UEs. HS-DPCCH doesn’t carry any higher layer control or traffic and doesn’t have a transport channel mapped on it.


3-6




Section 3-6UMTS UniversityPhysical Layer Frame Timing

Frame Timing• 10 ms radio frame, 15 slots• 2 ms HSDPA subframe, 3 slots

– 1 HS-DSCH Transmission Time Interval (TTI)

Slot Timing• 2560 chips per slot, 0.67 ms

– 7680 chips per HSDPA subframe

Symbol Timing• QPSK: 2 bits per symbol• 16-QAM: 4 bits per symbol• OVSF spreads symbol to chips

Physical Layer Frame Timing

A basic WCDMA radio frame is 10 ms long and has 15 slots. HSDPA introduces the notion of subframes within a WCDMA radio frame. An HSDPA subframe is 2 ms (3 slots) long and all the HS-channels use this subframe timing. The subframe allows fast user switching where the shared channel can potentially be assigned to a different user every subframe. As the HSDPA subframe is only 2 ms long, it alleviates the need for power control. HS-DSCH has a fixed TTI of 2 ms. Each HS-DSCH transport block is mapped to an HS-PDSCH subframe. HS-SCCH and HS-DPCCH also use the 2 ms subframe to transmit control and feedback respectively.

Each HSDPA subframe has 3 slots and each slot is comprised of symbols. The number of symbols in a slot depends on the spreading factor used for that channel. HS-PDSCH, HS-SCCH, and HS-DPCCH use SF 16, 128, and 256 respectively, giving number of symbols per slot as 160 (HS-PDSCH), 20 (HS-SCCH), and 10 (HS-DPCCH).

A symbol is made up of 1 or more bits and each bit is spread using SF to an equivalent number of chips. A QPSK symbol consists of two consecutive bits, one bit each mapped onto the I and Q branch. A 16-QAM symbol, on the other hand, has four consecutive bits with two bits on each branch.


3-7




Section 3-7UMTS UniversityDownlink HS-PDSCH

High Speed Physical Downlink Shared Channel (HS-PDSCH)• Fixed spreading factor SF 16 with 2 slot formats• Up to 15 HS-PDSCHs under a cell• May use QPSK or 16-QAM modulation scheme• Node B responsible for transmitting HS-PDSCH at reasonable power

DL HS-PDSCH – High Speed Physical Downlink Shared Channel

An HS-PDSCH channel carries the actual user payload to the UE. One HS-PDSCH subframe contains one TTI (2 ms) of HS-DSCH transport channel payload. There is no transport channel multiplexing in HSDPA so the information contained in HS-PDSCH subframe is from a single HS-DSCH transport channel.

An HS-DSCH serving cell can have as many as 15 channelization codes assigned to HS-PDCH. The HS-PDSCH channels are shared among different users by using time division, code division, or a combination of the two multiple access methods. The number of HS-PDSCHs that can be simultaneously decoded by a UE depends on the HS-DSCH UE Category.

The HS-PDSCH power control depends on the Node B’s implementation and is not specified by standards. The UE may assume that the power is kept constant during the corresponding HS-PDSCH subframe. If multiple HS-PDSCHs are allocated to one UE, all the HS-PDSCHs intended for that UE shall be transmitted with equal power.

The phase reference used for demodulating HS-PDSCH is the same as for the associated DL DPCH. By default, P-CPICH is used as the phase reference. UE is informed through higher layer signaling if S-CPICH or dedicated Pilot is to be used as the phase reference.


3-8





HS-DSCH Channel Coding and Physical Channel Mapping

Release 99 Channel Release 5 HS-DSCHNew Physical Layer blocks in Release 5:

• Bit Scrambling

• Physical Layer HARQ functionality

• HS-DSCH Interleaving

• Constellation re-arrangement for 16-QAM


Channel coding is done to add robustness to the information bits. Usually, channel coding is performed by adding redundant bits determined by an FEC coding scheme such as CRC, convolutional coding, turbo coding, etc. The HS-DSCH channel coding involves a number of other functions performed by the Node B’s Physical Layer. The main reason for this additional processing is the dynamic size of the transport block transmitted in an HS-DSCH TTI. Other reasons include large HS-DSCH payload size and the possible use of 16-QAM modulation for HS-PDSCH. Comparing the coding chain for the Release 99 channel with the Release 5 HS-DSCH channel, some blocks have been removed and some new blocks have been added.

HS-DSCH coding chain does not require:1. Concatenation, because there is always only one transport block per HS-DSCH TTI. The

transport block size, however, varies from 137 bits to 27952 bits. In case of retransmission, the transport block size remains the same as of the original transmission.

2. First DTX insertion, because HS-DSCH doesn’t support fixed position transport channel and thus Blind Transport Format Detection (BTFD).

3. Second DTX insertion, because there is just one transport channel mapped on to HS-PDSCH.

4. Radio frame segmentation, because HS-DSCH has a fixed TTI of 2 ms, which is equal to the HS-PDSCH subframe duration.

5. Transport channel multiplexing, because there is just one transport channel mapped on to HS-PDSCH.


3-9




Section 3-9UMTS UniversityHS-DSCH Channel Coding

MAC delivers 1 HS-DSCH Transport Block per TTI to Physical Layer

CRC Attachment• 24-bit CRC is added per Transport

BlockBit Scrambling

• Facilitates uniform distribution of 16-QAM symbols at receiver

Code Block Segmentation• FEC encoder has a fixed maximum

code block size of 5114 bits• If bit scrambled data is more than 5114

bits, need to segment into equal code blocks

HS-DSCH Channel Coding

Node B’s MAC-hs delivers the HS-DSCH transport channel data to the Physical Layer in Node B. The Physical Layer then performs a number of functions on the HS-DSCH TTI data before the data is finally mapped to one or more HS-PDSCH physical channels.

CRC Attachment – A fixed 24-bit Cyclic Redundancy Check (CRC) is attached to HS-DSCH TTI data. There is only one transport block per HS-DSCH TTI.

Bit Scrambling – Done to avoid non-uniform symbol distribution over 16-QAM constellation at the receiver. A uniform symbol distribution helps the UE efficiently decode the received HS-DSCH bits. Typically, the received symbols are uniformly distributed over the entire constellation. However, certain degenerate HS-DSCH bit sequences (e.g., the all-ones or all-zeroes sequences) could violate this condition, leading to an asymmetric HS-DSCH bit distribution (over {0,1}) and hence a non-uniform 16-QAM symbol distribution at the receiver input. This is true regardless of the use of turbo-encoding on the HS-DSCH, due to the possibility of transmitting turbo-codewords comprised predominantly of systematic bits. The estimated performance loss due to the non-uniform distribution in such very unlikely cases is between 1.0-1.5 dB.

Code Block Segmentation – It is done if the number of bits output from the bit scrambler is more than the maximum input code block size of the FEC encoder. The maximum encoder code block size in case of HS-DSCH is 5114 bits. If segmentation is performed, all the resulting segments are of equal size and may require adding some filler bits to the beginning of 1st code block. The filler bits are all 0s and are transmitted along with data.


3-10





HS-DSCH Channel Coding (continued)

FEC Coding• Rate 1/3 Turbo Coder used for Forward Error Correction (FEC)• Tail bits are padded to perform trellis termination• Effective code rate changes after HARQ• Code blocks from the same transport block are concatenated after encoding.

HARQ

Turbo Coder(r = 1/3)

Code Blk

Serialconcatenation ofencoded blocks

Systematic bits Parity 2 bitsParity 1 bits


FEC Coding – Rate 1/3 turbo coder is used for encoding HS-DSCH bits. FEC coding is done on one or more code blocks, where code blocks are formed by segmenting bit scrambled HS-DSCH data (if more than 5114 bits). The minimum HS-DSCH transport channel data rate is 68.5 kbps, which makes turbo coding a better choice than convolutional coding. The turbo coder uses two parallel concatenated convolutional coders, each with constraint length K = 4. The output from turbo coder consists of Systematic bits (original input data bits) and Parity bits. For each input bit, there is 1 Systematic bit and 2 Parity bits. Twelve tail bits are added per block after encoding for the trellis termination. The encoded blocks, when more than one, are serially concatenated and fed to the HARQ block. The code rate after turbo encoding is 1/3 but the effective coder rate after HARQ rate matching may be different. An effective code rate of close to 1 is required to achieve peak throughput of 14.4 kbps.


3-11






Hybrid ARQ (HARQ)• Combines ARQ with adaptive coding (FEC)• Transmitter sends new set of parity bits if

the previous transmission failed (NAK’d)• Receiver buffers the failed decodes for soft

combining with future retransmissions• Soft combining is done before each FEC

decoding attempt

HARQ

Serialconcatenation ofencoded blocks

To Physical ChannelSegmentation


Hybrid ARQ (HARQ) – HARQ is a technique combining FEC and ARQ methods that save information from previous failed decode attempts to be used in the future decoding. There are two different HARQ schemes, Chase and IR, depending on which bits are chosen to be sent over the air to UE. The redundancy version (RV) parameters, r and s, indicate to the UE the HARQ scheme used for the current transmission.

Both HARQ combining schemes soft combine bits from the previous failed decodes with the currently received retransmission. Soft combining helps minimize the number of retransmissions. For a retransmission, HARQ uses the same transport block size and consequently the same number of HS-DSCH bits that were used in the initial transmission. However, it may use a different modulation scheme, channelization code set (including the size of the channelization code set), or transmission power. Thus, HARQ implicitly provides a link adaptation technique called Adaptive Modulation and Coding (AMC). The number of physical channel bits available for a retransmission may differ from that of the previous transmission.


3-12





HS-DSCH Channel Coding –Physical Layer HARQ Functionality

Physical Layer HARQ consists of two rate matching stages and a virtual buffer• 1st stage: matches number of input bits to the virtual IR buffer size

– IR buffer size is determined by UE’s soft memory capability– Puncturing is done if inputs bits exceed the virtual IR buffer size

• 2nd stage: matches numbers of bits to the number of HS-PDSCH bits in the given TTI– Redundancy Version (RV) parameters control the output from 2nd stage– Repetition or puncturing is done to perform 2nd stage rate matching

RV Parameters

HS-DSCH Channel Coding – Physical Layer HARQ Functionality

The Physical Layer HARQ functionality in Node B matches the number of bits at the output of the channel coder to the total number of HS-PDSCH physical channel bits in a subframe. It consists of two rate matching stages and a virtual IR buffer.

1st rate matching stage matches the number of input bits to the virtual Incremental Redundancy (IR) buffer size, information about which is provided by the higher layers. The systematic bits remain untouched in this stage but some parity bits (P1 and P2) may be punctured if the total number of input bits (including S1, P1, P2) is more than the virtual IR buffer size.

2nd rate matching stage matches the number of bits after the 1st rate matching stage to the number of available HS-PDSCH bits. Given a fixed number of input and output bits for the 2nd

rate matching stage, the exact set of output bits depends on the RV parameters s and r. Either puncturing or repetition is performed to accomplish the 2nd stage rate matching.


3-13





HS-DSCH Channel Coding –HARQ Combining Schemes

First Transmission• Always self-decodable, RV parameters s = 1

Chase Combining• Each retransmission is self decodable, RV parameter s = 1

– Systematic bits are prioritized• Same coded data packet may be sent in each retransmission

– Using the same RV parameter r in each retransmission• Retransmission with a different r value implies different set of

punctured bits• Receiver attempts to decode by soft combining multiple copies

Incremental Redundancy (IR)• Retransmissions are not self decodable, RV parameter s = 0

– Parity bits are prioritized• Redundant information is incrementally transmitted if initial decoding

fails• Each retransmission provides additional redundant bits to the receiver

– RV parameter r is different for different set of redundancy bits• Receiver attempts to decode based on accumulated bits

HS-DSCH Channel Coding – HARQ Combining Schemes

HARQ combining refers to the combining of the HS-DSCH soft bits in the receiver (UE). If an HS-DSCH subframe transmission is not correctly decoded (CRC failure) by the UE’s Physical Layer the soft bits from this failed decode are buffered in the IR buffer to be combined with the soft bits from the future retransmissions. This type of combining changes the effective received code rate with each retransmission and helps in minimizing the number of retransmissions. There are different types of HARQ combining schemes:

Chase combining (also called HARQ Type III) requires each retransmission to be self-decodable. The transmitter may retransmit the same coded data packet in which case the decoder at the receiver combines multiple copies of the same transmitted packet weighted by the received SNR. Time diversity gain is thus obtained. Using a different redundancy version parameter r, a different set of puncture bits can be used in each retransmission.

Incremental Redundancy (IR) (also called H-ARQ Type II) is another implementation of the HARQ technique where retransmissions are not self decodable, i.e., they may have a very low proportion (or none) of the systematic bits. Additional redundant information, prioritizing the parity bits, is incrementally transmitted if the decoding fails on the prior attempt. Retransmitted subframes are soft combined with the buffered soft bits to achieve additional coding gain, which helps the UE to successfully decode the subframe.

The Node B’s proprietary algorithm in MAC-hs determines which HARQ scheme to use to transmit an HS-DSCH subframe. The RV parameter signaled to the UE indicates the HARQ scheme used, allowing the UE to use the same scheme for HARQ combining.


3-14





HS-DSCH Channel Coding – Segmentation and Interleaving

Physical Channel Segmentation• Segments data equally into P segments• P is number of HS-PDSCHs allocated

to UE– Up to a maximum of 15

• Total HS-PDSCH bits per TTI equals:P * (Number of bits per HS-PDSCH channel)

HS-DSCH Interleaving• Block interleaving using 32x30 matrix

– Write in rows, read out columns• Done separately on each HS-PDSCH• Bits input = Bits per TTI per HS-PDSCH

– 960 bits (QPSK), 1920 bits (16-QAM)

HS-DSCH Channel Coding – Segmentation and Interleaving

Physical Channel Segmentation – When more than one HS-PDSCH is allocated to a UE, physical channel segmentation divides the post-HARQ HS-DSCH bits among the different HS-PDSCH physical channels. The number of HS-PDSCHs is denoted by P. If R is the number of bits input to the physical channel segmentation block, the number of bits in one subframe for each HS-PDSCH is R/P.

H-DSCH Interleaving – Interleaving combats burst errors. For HS-DSCH, the basic interleaver is the same as the 2nd interleaver for the Release 99 channels. The block interleaver is of fixed size: 32 rows and 30 columns. The interleaving operation is done independently for each HS-PDSCH bit stream. Bits are written in rows and read out in columns.

For 16-QAM, there are two identical 32x30 interleavers. A set of four consecutive output bits from the physical channel segmentation is divided between the interleavers such that bits up,k and up,k+1 go to the first interleaver and bits up,k+2 and up,k+3 go to the second interleaver. Bits are collected in the same order from the two interleavers after interleaving is done.


3-15





HS-DSCH Channel Coding –16-QAM Constellation Rearrangement

Bit reliability . . . • changes with bit position within a symbol• is different for 0 and 1 in case of i2 and q2

HS-DSCH Channel Coding – 16-QAM Constellation Rearrangement

An optional 16-QAM modulation scheme has been introduced for HS-PDSCH to achieve high data rates. Constellation rearrangement is required in the case of 16-QAM modulation because two of the four bits in a 16-QAM symbol have a higher probability of error than the other twobits. The rearrangement occurs during retransmission and disperses the error probability equally among all the bits when averaged over retransmissions.

The reliabilities of the bits mapped to the 16-QAM symbols vary from the most significant bits (i1, q1) to the least significant bits (i2, q2). These variations reduce the performance of the turbo decoder with respect to having equal bit reliabilities. By rearranging the signal constellation during retransmissions, the same bit gets placed at different positions within a symbol across different retransmissions and the bit reliabilities are averaged out over the retransmissions. For both Chase combining and IR, the decoder performance increases with the constellation rearrangement due to a more homogeneous input of log-likelihood values to the turbo decoder.

The bits output from the HS-DSCH interleaver are taken in groups of four consecutive bits (i1q1i2q2) and then rearranged based on the value of constellation version parameter b. Node B signals this parameter to the UE on HS-SCCH channel so that the UE can undo this bit rearrangement. In case of QPSK modulation, the constellation rearrangement block is transparent.


3-16




Section 3-16UMTS UniversityHS-DSCH Physical Channel Mapping

• Bits are mapped to one or more HS-PDSCH• Same number of bits/subframe on each HS-PDSCH

– 960 bits (QPSK) or 1920 bits (16-QAM)

HS-DSCH Physical Channel Mapping

After constellation rearrangement (for 16-QAM) or HS-DSCH interleaving (for QPSK), the HS-DSCH bits are finally mapped to one or more HS-PDSCH channels. This is called Physical Channel Mapping. A UE may be assigned one or more HS-PDSCH codes depending on the UE capability, QoS requirement, and the Node B’s radio resource availability. In case of more than one HS-PDSCH channel assigned to a UE, the number of bits in the given subframe on each assigned HS-PDSCH channel is the same. In an HS-PDSCH subframe, there are 960 bits in case of QPSK modulation and 1920 bits in case of 16-QAM modulation.


3-17




Section 3-17UMTS UniversityDownlink HS-SCCH

High Speed Shared Control Channel (HS-SCCH)• Fixed rate 60 kbps (SF 128) channel with one slot format• UE may need to simultaneously monitor up to four

HS-SCCHs• More than four HS-SCCHs possible under one cell• Channel power depends on Node B implementation• QPSK only

DL HS-SCCH – High Speed Shared Control Channel

The Node B transmits control information required for detecting and decoding HS-PDSCH subframes to UEs on HS-SCCH channel. UEs are signaled to monitor a set of HS-SCCH channels containing up to a maximum of four HS-SCCHs. At any time, only one of the four HS-SCCHs contains information for a given UE. There may be more than four active HS-SCCHsunder a cell. Multiple users are assigned to the same HS-SCCH (or set of HS-SCCHs) and thus a UE can successfully decode the information on this channel only when the information is intended for that UE. The HS-SCCH information is scrambled with the UE ID, which enables the desired UE to successfully decode HS-SCCH. The reason for having multiple HS-SCCHs is to enable Node B to address multiple UEs in the same subframe.

Each HS-SCCH is spread with SF 128 (channelization code is not fixed) and has a single slot format. The HS-SCCH channel power is controlled by the Node B’s proprietary algorithm and is not specified by standards. A possible implementation is that HS-SCCH follows the power control commands sent by UE on the UL DPCCH.

The phase reference used for demodulating HS-SCCH is the same as for the associated DL DPCH. By default, P-CPICH is used as the phase reference. UE is informed through higher layer signaling if S-PICH or dedicated Pilot is to be used as the phase reference.


3-18




Section 3-18UMTS UniversityHS-SCCH Channel Coding

CRC computed over both

Part 1 and Part 2

HS-SCCH Channel CodingConvolutional coding and CRC coding are used as the main channel coding schemes by Node B for the HS-SCCH channel. Part 1 and Part 2 of an HS-SCCH subframe are individually coded and mapped to the allocated slots in a subframe. Both Part 1 and Part 2 are scrambled with the UE ID. The UE ID used for scrambling HS-SCCH is a 16-bit HS-DSCH Radio Network Temporary Identity (H-RNTI). Part 1 consists of the following information:

Channelization Code Set – Contains the number of in-sequence HS-PDSCH codes assigned to a UE and the offset of the first code. Modulation Scheme – HS-PDSCH modulation scheme where 0 = QPSK and 1 = 16-QAM.

Part 2 consists of the following information:Transport Block Size – The transport block size used for the corresponding HS-PDSCH subframe is signaled as a 6-bit Transport Format Resource Indicator (TFRI). The actual transport block size in bits is derived from TFRI and depends on the modulation scheme and the number of HS-PDSCH channelization codes signaled on HS-SCCH.HARQ Process ID – Contains the HARQ process ID for the corresponding HS-PDSCH subframe. There may be one to eight simultaneous HARQ processes running in a UE.Redundancy & Constellation Version – Contains RV parameters r and s that are used by the Physical Layer HARQ functionality. If 16-QAM modulation is used, this field also contains the constellation version parameter b that indicates the rearranged version of 16-QAM constellation used for the corresponding HS-PDSCH subframe transmission.New Data Indicator – Contains 1-bit indicator that toggles every time the Node B sends new HS-DSCH data. The indicator is not toggled in case of retransmissions.


3-19





HS-PDSCH and HS-SCCH Spreading and Modulation

• HS-PDSCH is spread with SF 16, scrambled with Primary Scrambling Code.

• HS-SCCH is spread with SF 128, scrambled with same code as HS-PDSCH.

HS-PDSCH and HS-SCCH Spreading and Modulation

The Downlink physical channels (except SCH) are spread to the chip rate with individual channelization codes and then scrambled with the same scrambling code. All such channels use QPSK modulation except HS-PDSCH, which can use either QPSK or 16-QAM.

The Downlink physical channels HS-SCCH and HS-PDSCH consist of a sequence of binary symbols. In the case of QPSK modulation, each pair of two consecutive symbols is first serial-to-parallel converted and then mapped to the I and Q branches. The QPSK modulation mapper maps the even and odd numbered symbols to the I and Q branch respectively. In the case of 16-QAM modulation, a set of four consecutive binary symbols nk, nk+1, nk+2, nk+3 (with k mod 4 = 0) is serial-to-parallel converted to two consecutive binary symbols (i1= nk, i2= nk+2) on the I branch and two consecutive binary symbols (q1= nk+1, q2= nk+3) on the Q branch and then mapped to 16-QAM constellation by the modulation mapper. The modulation mapper converts the binary symbols into the real-valued symbols.

The I and Q branches, containing real-valued symbols, are then both spread to the chip rate by the same real-valued channelization code Cch,SF,m. The real-valued chip sequences on the I and Q branch are then treated as a single complex-valued sequence of chips which is scrambled (complex chip-wise multiplication) by a complex-valued scrambling code Sdl,n. After spreading and scrambling, the complex-valued chip sequence from each channel is separately weighted bya weight factor Gi. All Downlink physical channels are then combined using complex addition and then sent to modulation.


3-20




Section 3-20UMTS UniversityHS-PDSCH and HS-SCCH Timing

• Start of HS-SCCH subframe # 0 is aligned with the start of PCCPCH frames.

• HS-PDSCH subframe is transmitted two slots after the associated HS-SCCH subframe.– Helps UE identify if upcoming HS-PDSCH subframe has data

for that UE.

HS-PDSCH and HS-SCCH Timing

An HS-PDSCH subframe starts two slots (5120 chips) after the start of the associated HS-SCCH subframe. Because Part 1 of HS-SCCH is contained in the 1st slot of the HS-SCCH subframe, this offset between HS-SCCH and HS-PDSCH subframe lets the addressed UE start preparing early for demodulating the upcoming HS-PDSCH subframe.

There are five HS-PDSCH and five HS-SCCH subframes in one 10 ms WCDMA radio frame. For both HS-PDSCH and HS-SCCH, the time alignment is the same for all active channels.


3-21




Section 3-21UMTS UniversityUplink HS-DPCCH

High Speed Dedicated Physical Control Channel (HS-DPCCH)

• Exists on Uplink with non-HS dedicated channels (UL DPCCH/DPDCH)

• One HS-DPCCH per radio link • 1 slot format, SF 256, OVSF code – Cch, 256, 64

• Fixed power offsets (∆ACK , ∆NAK , ∆CQI) relative to UL DPCCH

2560 chips

Subframe # 0 Subframe # 2

ACK/NAK CQI

One radio frame (10 ms)

One HS- DPCCH subframe (2 ms)

5120 chips

Subframe # 1 Subframe # 3 Subframe # 4

UL HS-DPCCH – High Speed Dedicated Physical Control Channel

Each UE operating in the HSDPA mode has an active Uplink HS-DPCCH along with the dedicated UL DPCCH. The UE uses UL DPCCH as reference for adjusting the HS-DPCCH channel power. UE transmits HS-DPCCH at a fixed power offset relative to UL DPCCH but the offset is different for ACK, NAK, and CQI fields. These power offsets are signaled to UE by UTRAN and are used by UE’s Physical Layer to calculate the HS-DPCCH gain factor(βhs). As the HS-DPCCH power is adjusted relative to UL DPCCH, the Uplink power control is indirectly adjusting the HS-DPCCH power.

Each subframe (2 ms) of HS-DPCCH has one slot for HARQ ACK/NAK and two slots for Channel Quality Indicator (CQI) field. UTRAN may configure the UE to repeat each ACK/NAK and/or CQI report up to three more times in the consecutive subframes. If there is nothing to acknowledge, i.e., no data received on HS-PDSCH or CRC error on HS-SCCH, then DTX bits are sent in the ACK/NAK field.

UTRAN configures CQI reporting by signaling CQI feedback cycle parameter to UE. Based on the feedback cycle parameter, UE may be asked to not send CQI at all or send CQI at periodic intervals ranging from 2 ms to 160 ms. For example, if the CQI feedback cycle is 4 ms, the UE reports CQI in every other subframe. Those subframes not scheduled to report CQI have DTX bits in place of CQI.


3-22




Section 3-22UMTS UniversityHS-DPCCH Channel Coding

• 1-bit ACK/NAK is coded as 10 bits.• 5-bit CQI is coded as 20 bits.• Subframe repetition of ACK/NAK and CQI adds more

reliability.

HS-DPCCH Channel Coding

Channel coding is done by UE’s Physical Layer to add robustness to the HS-DPCCH information. In general, there are different methods of doing channel coding such as repetition, convolutional coding, turbo coding, Reed-Muller (RM) coding, etc., but the basic strategy is to add some redundant bits to the original bit(s). This redundancy helps the receiver correctly decode the original bits which may have been impaired due to bad RF channel conditions. The 1-bit ACK/NAK information is coded into 10 bits by repeating the original bit. The 5-bit CQI information is coded into 20 bits by using RM coding.

UTRAN may configure UE to repeat each ACK/NAK and/or CQI report up to three more times in the consecutive subframes. This repetition also provides robustness to the HS-DPCCH information, especially in the soft-handover scenarios. When a UE is in soft-handover, the UL power control follows OR-of-Downs rule. If an active set cell, but a non-HSDPA serving cell, is sending more down commands than up commands, then the UE’s UL DPCCH power, and consequently the HS-DPCCH power, may not be reliably received at the HSDPA serving cell. Repeating the HS-DPCCH information in such a scenario would help the HSDPA serving cell reliably decode ACK/NAK and CQI.


3-23




Section 3-23UMTS UniversityHS-DPCCH Spreading and Modulation

• Unique OVSF code Cch, 256, 64

• Gain factor βhs is derived from power offsets (∆ACK , ∆NAK , ∆CQI)

• Multiplexed on Q branch• Same scrambling code as on UL

DPCH• Increases PA Peak-to-Average

Ratio– HPSK alleviates this problem

• QPSK modulation

HS-DPCCH is code-multiplexed with UL DPCH

HS-DPCCH Spreading and Modulation

The HS-DPCCH channel is I/Q code multiplexed with UL DPCH. Depending on whether the number of active UL DPDCHs is even or odd, HS-DPCCH is mapped on to I or Q branch, respectively. The SF used for HS-DPCCH is 256 with OVSF code number Cch, 256, 64 when there is only one active UL DPDCH. The power offsets ∆ACK , ∆NAK , and ∆CQI are signaled to UE by UTRAN through higher layer signaling.

On Uplink, DPCCH, DPDCH, and HS-DPCCH are I/Q multiplexed. These channels can be at different power levels resulting in different amplitudes for I and Q branch and can thus produce strange constellations. As the number of simultaneous Uplink channels increase, the distribution of power between the I and Q axis becomes more unsymmetrical. Transitions from any point to any other point in the final constellation are possible. All this increases the peak-to-average power ratio of the Uplink signal. Power Amplifiers (PA) are typically most efficient when they operate close to their saturation level. With a high peak-to-average power ratio, the efficiency of the UE’s PA is reduced and adversely affects the battery life. WCDMA employs complex scrambling (HPSK) on Uplink to fix the unequal distribution of power by continuously rotating the constellation and thereby distributing the power evenly between I and Q paths. Since HPSK eliminates zero-crossings and 0° phase-shifts only for pairs of consecutive chip points, transitions across pairs may still go through zero.


3-24




Section 3-24UMTS UniversityHS-DPCCH Timing

HS-DPCCH Subframe• Starts 7.5 slots + (0 to 255) chips after the received

HS-PDSCH subframe.– UE gets ~ 5ms to decode the HS-DSCH data and be

ready with ACK/NAK.• May not be aligned to UL DPCH frame or slot boundary.

HS-DPCCH Timing

UE transmits the HS-DPCCH subframe 7.5 slots + (0 to 255) chips after the end of thecorresponding received HS-PDSCH subframe. This gives UE time to decode the HS-DSCH data received in the HS-PDSCH subframe and compute ACK/NAK to be signalled on HS-DPCCH. The variable time (0 to 255) chips is to account for UE slewing and to help preserve orthogonality between different UL channels. The resultant HS-DPCCH subframe boundary may not be aligned to the UL DPCH frame or slot boundary.


3-25




Section 3-25UMTS UniversityHARQ Transmission Quiz

HARQ Transmission Example• Transport Block Size: 2400 bits• Virtual IR Buffer: 7000 bits per

HARQ process• Total number of HS-PDSCH

bits: 3840 bits (QPSK, 4 codes)

Questions:How many systematic and parity bits after:

• 1st rate matching?• 2nd rate matching for 1st

transmission?• 2nd rate matching for

retransmission using Chase? Using IR?

HARQ Transmission Quiz

Hints:

1. HARQ matches the number of input coded bits to the number of HS-PDSCH physical channel bits.

2. Redundancy version parameter s = 1 prioritizes systematic bits and s = 0 prioritizes parity bits.

3. Change in the redundancy version parameter r indicates a change in the bit positions that need to be punctured or repeated.

Consider the following assumptions:

1. All systematic bits are needed for a transmission to be self-decodable.2. Any puncturing or repetition, if required, is done equally on parity 1 and parity 2 bits.3. The retransmission can be done using either Chase combining or Incremental

Redundancy combining scheme.


3-26





HARQ Transmission Example• Transport Block Size: 2400 bits• Virtual IR Buffer: 7000 bits per

HARQ process• Total number of HS-PDSCH

bits: 3840 bits (QPSK, 4 codes)

Answers:• 1st rate matching punctures 284

parity bits to match 7000 bits.• 2nd rate matching punctures 3160

parity bits resulting in total 3840 bits.

• 2nd rate matching outputs the same number of bits as 1st transmission for case of Chase retransmission. For IR retransmission, 3840 parity bits are sent with 0 systematic bits.

HARQ Transmission Quiz – Answers

HARQ Transmission Quiz – Answers


3-27





1. How many HS-DSCH TTIs can be accommodated in 1 HS-PDSCH subframe?

2. How many active Uplink channels are required per UE for the HSDPA operation?

3. How much time does the UE get to decode HS-DSCH subframe before sending acknowledgement?

4. What is the purpose of the HS-SCCH channel?

5. How does a UE identify information addressed to it on the HS-SCCH channel?

Review Quiz

Notes


3-28





6. What is the time offset between the HS-SCCH and the corresponding HS-PDSCH subframe?

7. Name the two different HARQ techniques.

8. What is the function of the 1st rate matching stage in Node B's HARQ functionality?

9. Why is the constellation rearrangement required for 16-QAM modulation?

Review Quiz

Notes


3-29





1. How many HS-DSCH TTIs can be accommodated in 1 HS-PDSCH subframe?• 1 HS-PDSCH subframe accommodates 1 HS-DSCH TTI (2 ms)

2. How many active Uplink channels are required per UE for the HSDPA operation?

• 3 channels (UL DPCCH, UL DPDCH, HS-DPCCH)3. How much time does a UE get to decode HS-DSCH subframe before sending

acknowledgement?• 7.5 slots + 0…255 chips

4. What is the purpose of the HS-SCCH channel?• Carries the control information required by UEs to decode HS-PDSCHs

5. How does a UE identify information addressed to it on the HS-SCCH channel?• Uses H-RNTI for decoding HS-SCCH information

Notes


3-30





6. What is the time offset between the HS-SCCH and the corresponding HS-PDSCH subframe?

• HS-SCCH subframe starts two slots prior to HS-PDSCH subframe7. Name the two different HARQ techniques.

• Chase Combining scheme• Incremental Redundancy scheme

8. What is the function of the 1st rate matching stage in Node B's HARQ functionality?

• Match the number of input bits to the IR buffer size9. Why is the constellation rearrangement required for 16-QAM modulation?

• To reduce the effect of bit reliability variations

Notes


3-31





Physical Layer Channels –What We Learned

Identified the HSDPA Physical channels.Described coding of HSDPA channels.HSDPA channel timing relation.

Notes


3-32



Comments/Notes


4-1

Section 4: UE Physical Layer Processing80-W0331-1 Rev B




Section 4:UE Physical Layer Processing

4SECTION

UE Physical Layer Processing

Notes


4-2





Explain the Physical Channel processing done in UE to accomplish HSDPA operation.List Transmit Diversity schemes used by HS channels.Identify HS-DSCH UE categories.

Notes


4-3




Section 4-3UMTS UniversityUE Physical Channel Processing

1. UE attempts to decode HS-SCCH subframe.

2. UE decodes HS-DSCH TTI data from HS-PDSCHs.

3. UE reports ACK/NAK and CQI on UL HS-DPCCH.

UE Physical Channel Processing

The HSDPA physical channel processing done by UE to accomplish HSDPA operation involves three main steps:

1. HS-SCCH Monitoring – UE continuously monitors the HS-SCCH channels to identify any HS-PDSCH subframes addressed to it on the set of HS-PDSCH channels.

2. HS-DSCH Decoding – Upon receiving an HS-PDSCH subframe, the UE’s Physical Layer decodes the HS-DSCH TTI to pass the information bits up to MAC-hs.

3. HS-DPCCH Signaling – UE sends feedback on UL HS-DPCCH containing information regarding the DL channel quality and the HARQ acknowledgements.


4-4




Section 4-4UMTS UniversityUE HS-SCCH Monitoring

1. UE attempts to decode HS-SCCH subframe.1.1 Successful decode of Part 1 (1st slot) provides UE:

Channelization Code Set and Modulation SchemeReasonable confidence that next HS-PDSCH subframe has data

1.2 UE prepares for demodulating upcoming HS-PDSCH subframe

UE HS-SCCH Monitoring

UTRAN sends information to UE regarding what HS-SCCH channels to monitor in one of the dedicated control messages such as radio bearer setup, radio bearer reconfiguration, transport/physical channel reconfiguration, etc. While monitoring the HS-SCCH channels, a UE attempts to decode the Part 1 and Part 2 of the HS-SCCH subframe using its H-RNTI. If a UE successfully decodes Part 1 of one of the monitored HS-SCCHs, the UE starts receiving the HS-PDSCHs indicated by the Part 1 control information. The UE considers the Part 1 decoding to be successful only if the decoded channelization code set is less than or equal to the maximum number of HS-PDSCH codes supported by its HS-DSCH UE category, and if the decoded modulation scheme is valid for its UE category.

In general, a UE monitors up to four HS-SCCHs. Once an HS-SCCH is successfully decoded, the UE continues to monitor it. When decoding fails, the UE again starts monitoring all the assigned HS-SCCHs.


4-5





UE HS-SCCH Monitoring (continued)

1.3Successful decode of Part 2 (2nd and 3rd slot) provides UE:

HARQ Parameters (s, r, b, nd), HARQ ID, and Transport Block SizeConfirmation that next HS-PDSCH subframe has data for this UE

1.4UE starts demodulating HS-PDSCH subframeHS-PDSCH subframe starts two slots after the start of HS-SCCH subframe

UE HS-SCCH Monitoring (continued)

Where Part 1 provides UE with the HS-PDSCH channelization codes and modulation scheme, Part 2 provides HARQ parameters (s, r, b, nd), HARQ process ID, and TFRI. All this information is required by UE’s Physical Layer to correctly decode the HS-DSCH TTI data from HS-PDSCH subframe. The information obtained from HARQ parameters is:

s identifies the HARQ combining scheme (Chase or IR) r identifies the redundancy version b identifies the 16-QAM constellation version nd identifies the new HS-DSCH data

A successful decode (CRC pass) of Part 2 confirms that the HS-SCCH subframe in intended for this UE. In case the decoding of Part 2 results in a CRC failure, the UE discards the information received on HS-SCCH subframe and the corresponding HS-PDSCH subframe.


4-6




Section 4-6UMTS UniversityUE HS-DSCH Decoding

• De-scramble, de-spread, and demodulate

• De-interleave

• HARQ processing

• Turbo decoding

• Bit de-scrambling

• CRC check

2. UE decodes HS-DSCH TTI data from HS-PDSCHs.

UE HS-DSCH Decoding

Once a UE successfully decodes HS-SCCH, it gets all the control information required to decode the HS-DSCH TTI data. The sequence of functions performed by the UE’s Physical Layer to decode HS-DSCH is basically the inverse of the coding chain followed by the Node B’s Physical Layer to code HS-DSCH data. The different functions performed in the coding chain have been explained earlier.

After the CRC check, the UE’s Physical Layer conveys the CRC result to MAC-hs. If CRC passes, the Physical Layer also delivers the decoded HS-DSCH bits to MAC-hs. If it fails, the Physical Layer buffers the soft bits in the UE’s IR buffer for soft-combining with future transmissions.


4-7




Section 4-7UMTS UniversityHARQ Processing

HARQ Processing• Up to 8 parallel HARQ

processes • 2 stage de-rate matching

per HARQ process• Soft-bits are combined

with buffered soft samples

• Retransmitted bits are buffered after combining

To Turbo Decoder

New Data Indicator (nd ),HARQ Process Id, Redundancy Version (s, r, b)

HARQ process sw itch

L1 HARQ Entity

bit separation

1st de-rate m atching

HARQ com bining

bit collection

2nd de-rate matching

IR buffer

HARQ process

Transport Format Resource Indicator (TFRI)

HARQ process sw itch

From Physical Channel Multiplexing

HARQ Processing

The UE’s Physical Layer HARQ processing is the inverse of the Node B’s Physical Layer HARQ functionality, with the additional function of buffering and soft combining. A UE buffers the failed HS-DSCH decodes to be soft combined with the future retransmissions. The total buffering capacity of a UE is determined by its IR buffer size, which is different for different HS-DSCH UE categories. The UE’s IR buffer size may be split evenly or unevenly among the multiple HARQ processes running simultaneously.

The received physical channel bits from all the allocated HS-PDSCHs are multiplexed into a bit stream, which is fed to the HARQ entity. The UE’s HARQ entity identifies the HARQ process ID from the HS-SCCH Part 2 information. The HARQ process uses the redundancy version parameters r and s to determine the HARQ combining scheme that needs to be used for combining soft bits. Soft combining happens only if the IR buffer has previously stored samples and the new data indicator nd bit remains the same as in the previous transmission. In case the ndbit toggles as compared to its value in the previous transmission, the UE buffer contents, if any, are flushed and restored with the new data. If the new nd bit is the same as previous but the UE’sbuffer is empty, it means UE has already successfully decoded the HS-DSCH subframe but ACK didn’t reach Node B. In this case the UE discards the new HS-PDSCH subframe and resends the ACK to Node B.


4-8




Section 4-8UMTS UniversityUL Feedback Signaling

3. UE reports ACK/NAK and CQI on UL HS-DPCCH.• UE’s MAC-hs decides ACK/NAK based on CRC result• UE’s L1 determines CQI based on CPICH strength

UL Feedback Signaling

To accomplish the HARQ functionality, the UE needs to send the MAC-hs acknowledgements to Node B so that Node B either retransmits the subframe or advances it transmit/receive window. The UE also needs to send some feedback on UL to accomplish the link adaptation process. Link adaptation involves adaptively changing the channelization code set, the modulation scheme, and the HS-PDSCH transmit power based on the DL channel conditions as perceived by UE. Link adaptation may also be used by the Node B scheduler to determine which UEs should be assigned HS-PDSCHs at a given time. For example, the UEs experiencing good DL RF channel may be given preference over the UEs experiencing bad channel conditions.

The UE reports ACK/NAK and CQI to Node B on HS-DPCCH. The UE’s MAC-hs provides an ACK or NAK to the Physical Layer depending on the CRC result evaluated over the recently received HS-DSCH subframe. This ACK/NAK field fills the 1st slot of HS-DPCCH subframe. For every received HS-PDSCH subframe, the UE reports ACK/NAK to Node B in an HS-DPCCH subframe starting 7.5 slots after the end of the corresponding HS-PDSCH subframe. No ACK/NAK is sent if nothing is received on HS-PDSCH or if the HS-SCCH CRC fails.

The UE’s Physical Layer measures the DL Pilot strength and computes CQI. CQI is reported in the 2nd and 3rd slot of HS-DPCCH subframe. The feedback cycle of CQI is a network parameterand is defined in 2 ms steps from 2 ms to 160 ms. An active HS-DPCCH may have slots in which no CQI information is transmitted.


4-9




Section 4-9UMTS UniversityChannel Quality Indicator (CQI) Measurement

CQI is:• a measure of Downlink channel quality as seen by UE• based on CPICH strength measured by UE • associated with a three slot reference period ending one slot

before CQI is sent– CQI indicates maximum data that could be reliably received during

reference period• used by Node B scheduler

HS-SCCH

HS-PDSCH 1

HS-PDSCH P

CQI report

Time

DL

ACK/NAKPreferencePeriod

6 ms

Channel Quality Indicator (CQI) Measurement

CQI is a metric that reflects the quality of the DL channel as seen by the UE. Depending on the UE’s implementation and its receiver architecture, it may perform better or worse than another UE under the same channel conditions. This justifies why CQI is not just a C/I or SNR measurement of CPICH but rather an indication of the highest data rate that can be reliably received by a UE under the current channel conditions. Transport Format Resource Combination (TFRC) points to the combination of number of HS-PDSCH channelization codes, modulation scheme, and the HS-DSCH transport block size.

3GPP describes CQI:

“Based on an unrestricted observation interval, the UE shall report the highest tabulated CQI value for which a single HS-DSCH sub-frame formatted with the transport block size, number of HS-PDSCH codes and modulation corresponding to the reported or lower CQI value could be received in a 3-slot reference period ending 1 slot before the start of the first slot in which the reported CQI value is transmitted and for which the transport block error probability would not exceed 0.1.”


4-10




Section 4-10UMTS UniversityCQI Reporting

• CQI is reported as an index into a 30-entry CQI table– Each entry corresponds to a different Transport Format Resource

Combination (TFRC)

• CQI indicates the highest TFRC that could be received during reference period– With transport block error rate ≤ 10 %

• CQI can also indicate a negative power offset for HS-PDSCH power– When UE can achieve peak data rate with lower than current power

• CQI reported by different UEs under same channel may be different– Depends on receiver architecture

• Mapping between CQI index and TFRC varies with UE HS-DSCH category

CQI Reporting

The CQI reported by the UE is an index into a table containing all possible TFRC combinations. The TFRC combinations are different for UEs with different HS-DSCH UE categories because of the differences in their capabilities. Along with TFRC, CQI may also indicate a power offset relative to the current HS-PDSCH power. A negative power offset may be signaled if the current HS-PDSCH power is more than what’s required by the UE to achieve peak data rate as per its UE category. The UE assumes a total received HS-PDSCH power of PHS-PDSCH = PCPICH + Γ+∆(dB), where the total received power is evenly distributed among the HS-PDSCH codes, the measurement power offset Γ is signaled by higher layers, and the reference power adjustment ∆is the power offset indicated by CQI.


4-11




Section 4-11UMTS UniversityCQI Mapping Table

Example for UE supporting up to five HS-PDSCH codes

CQI TransportBlock Size

Number of HS-PDSCHs

Modulation Scheme

Reference Power Adjustment?

NIR XRV

0 Not Valid1 137 1 QPSK 02 173 1 QPSK 0...

16 3565 5 16-QAM 0...

22 7168 5 16-QAM 023 7168 5 16-QAM -1...

30 7168 5 16-QAM -8

9600 0

CQI Mapping Table

The CQI table consists of 30 entries, where each entry indicates a different TFRC. Transport Format Resource Combination (TFRC) points to the combination of number of HS-PDSCH channelization codes, modulation scheme, and the HS-DSCH transport block size. The 5-bit CQI reported by a UE is an index into this table containing all possible TFRC combinations for that UE category. The TFRC combinations are different for UEs with different HS-DSCH UE categories because of the differences in the UE capabilities. Along with TFRC, CQI may also indicate a power offset relative to the current HS-PDSCH power.

The CQI table shown in the slide is for UE categories supporting up to five HS-PDSCH codes.


4-12




Section 4-12UMTS UniversityUE Categories

1.8 Mbps2880036301512*

0.9 Mbps1440036302511*

14.0 Mbps1728002795211510

10.1 Mbps172800202511159

7.2 Mbps134400144111108

7.2 Mbps115200144111107

3.6 Mbps672007298156

3.6 Mbps576007298155

1.8 Mbps384007298254

1.8 Mbps288007298253

1.2 Mbps288007298352

1.2 Mbps192007298351

Peak Data Rate

UE IR Buffer Size

Max. TB SizeMin. Inter-TTI Interval

HS-PDSCH Codes

HS-DSCH Category

* 16-QAM modulation not supported

UE Categories

HSDPA is advertised with data rates up to 14 Mbps. However, the actual HS-DSCH peak data rate depends on the UE’s HS-DSCH category. As shown in the table, only a category 10 UE can achieve the maximum HSDPA throughput of 14 Mbps when using all 15 HS-PDSCHssimultaneously.

Factors that decide the UE’s HS-DSCH category are:

1. HS-PDSCH codes – Determines the number of simultaneous HS-PDSCH channels that can be decoded by a UE.

2. Inter-TTI interval – Determines the minimum interval (in terms of HS-DSCH TTI) between two successive HS-PDSCH assignments. The more HARQ processes a UE supports, the shorter the inter-TTI interval. A minimum inter-TTI of 1 requires at least 6 simultaneous HARQ processes.

3. Transport Block size – Determines the maximum size of transport block that can be sent on HS-DSCH in a TTI. It is dependent on the number of HS-PDSCH codes and the modulation scheme.

4. IR buffer size – Determines the maximum number of soft bits that can be buffered by a UE across all simultaneously running HARQ processes.


4-13




Section 4-13UMTS UniversityDL Transmit Diversity

DL Transmit Diversity CombinationChannel

STTD

STTD

STTD

STTDNoneHS-SCCH

CLTD Mode 1NoneHS-PDSCH

CLTD Mode 1None

Associated DL DPCH

DL Transmit Diversity

HS-SCCH and HS-PDSCH are transmitted using diversity only if the associated DL DPCH is transmitted using one of the DL transmit diversity schemes. DPCH can be transmitted using: Space Time Transmit Diversity (STTD) or Closed Loop Transmit Diversity (CLTD) Mode 1. The transmit diversity scheme used for HS-PDSCH is the same as the transmit diversity scheme used for the associated DPCH. The transmit diversity scheme used for HS-SCCH is always STTD.


4-14





QUIZ

Notes


4-15





1. When does a UE monitor HS-SCCH channel?

2. Can a UE start decoding HS-DSCH before it fully receives the Part 2 of HS-SCCH?

3. What is the key difference between Node B's and UE's physical layer HARQ functionality?

4. What is the purpose of new data indicator flag in HARQ process?

5. What is CQI and how is it signaled to Node B?

6. Is it mandatory for an HSDPA capable UE to support 16-QAM modulation scheme?

Review Quiz

Notes


4-16





1. When does a UE monitor HS-SCCH channel?• The UE monitors the HS-SCCH channel all the time while it's operating in

the HSDPA mode.2. Can a UE start decoding HS-DSCH before it fully receives the Part 2 of

HS-SCCH?• No. The UE's Physical Layer needs Part 2 information to decode HS-DSCH

bits.3. What is the key difference between Node B's and UE's Physical Layer

HARQ functionality?• The UE's Physical Layer HARQ performs soft-combining, which is not done

on Node B side.4. What is the purpose of new data indicator flag in HARQ process?

• Indicates if the current HS-PDSCH subframe contains a retransmission or a new transmission.

Notes


4-17





5. What is CQI and how is it signaled to Node B?• CQI stands for Channel Quality Indicator• Signaled as a 5-bit value on HS-DPCCH• Signaled to Node B as an index into a TFRC table

6. Is it mandatory for an HSDPA capable UE to support 16-QAM modulation scheme?

• No. It depends on the UE HS-DSCH Category. The UE HS-DSCH categories 11 and 12 do not support 16-QAM modulation.

Notes


4-18





UE Physical Layer Processing –What We Learned

UE Physical Layer HSDPA processing.DL transmit diversity schemes.HS-DSCH UE categories.

Notes


5-1

Section 5: Layer 2 Protocols80-W0331-1 Rev B




Section 5:Layer 2 Protocols

5SECTION

Layer 2 Protocols

Notes


5-2





Review the HSDPA network and UE protocol architecture.Illustrate data flow from the RNC RLC layer to the UE RLC layer.Explain how MAC-d multiplexing occurs and how MAC-d flows are mapped to priority queues.Define the HARQ protocol and explain how various error scenarios affect the protocol.Define the re-ordering protocol and illustrate two mechanisms for flushing date from the re-ordering queues.Examine the impact of HSDPA on the RLC protocol.

Notes


5-3






25.308 HSDPA overall description stage 225.321 Medium Access Control (MAC) Protocol Specification25.322 Radio Link Control (RLC) Protocol Specification

References

Notes


5-4






In a Release 99 PS network, the NAS layer protocols are terminated at the SGSN. RRC, RLC, and MAC protocols are terminated at the RNC. The Physical Layer protocol is terminated at the Node B.


UTRAN MAC-hs is responsible for fast scheduling of the HS-PDSCHs. The scheduler determines:

To which UEs the channels are assigned.How much data to send.Which modulation scheme to use.Whether to send new data or retransmitted data.Which redundancy version to send.

UE MAC-hs is responsible for:

Sending ACK or NAK after decoding a block.Re-ordering data blocks before submitting to upper layers, if retransmissions caused data to be received out of order.


5-5




Section 5-5UMTS UniversityUTRAN MAC Architecture

UTRAN MAC Architecture

The UTRAN MAC protocol consists of three entities:

MAC-hs – Responsible for the high speed HSDPA channels and the only entity of MAC that resides in the Node B. When a UE operates in HSDPA mode, MAC-hs maps user data and signaling from DCCH and DTCH onto the shared HS-DSCH transport channels.MAC-c/sh – Responsible for common and shared logical (PCCH, BCCH, CCCH, and CTCH) and transport (PCH, BCH, RACH, FACH) channels. MAC-c/sh resides in the RNC, and there is one MAC-c/sh entity per RNC. When a UE operates in Cell_FACHstate, MAC-c/sh maps user data and signaling from its DCCH and DTCH onto the common FACH and RACH transport channels.MAC-d – Responsible for mapping data from dedicated logical channels (DCCH and DTCH) onto dedicated transport channels (DCH). MAC-d resides in the RNC, and there is one MAC-d entity for each UE to which dedicated logical channels have been assigned.


5-6




Section 5-6UMTS UniversityUTRAN MAC-hs Architecture

UTRAN MAC-hs Architecture

Data enters the UTRAN MAC-hs from a set of MAC-d flows. The data is routed to a set of priority queues with the following properties:

Up to 8 priority queues and 8 MAC-d flows are allowed per UE.The queue distribution entity maps each MAC-d flow onto one or more priority queues. The mapping is configured when the HSDPA operation begins. Each priority queue is mapped to only one MAC-d flow.

When data is removed from a priority queue for transmission, it is assigned to a HARQ process. There are a minimum of 6 and a maximum of 8 HARQ processes per UE. The HARQ process tracks the ACK/NAK signaling for the data block and determines when retransmission is necessary.

In response to CQI and ACK/NAK signaling on HS-DPCCH, the scheduler decides:

To which UEs the HSDPA channels will be assigned.For each scheduled UE, whether to send new data from a priority queue or a retransmission from a HARQ process.

Signaling on HS-SCCH indicates the scheduling decision to the UEs operating in HSDPA mode.


5-7




Section 5-7UMTS UniversityUE MAC-hs Architecture

Dis-assembly

MAC-d Flows

HS-DSCH Signaling on HS-SCCHSignaling on HS-DPCCH

Re-ordering queue distribution

MAC-d Flow Mapping

HARQ processes

Re-ordering Queue

Dis-assembly

Re-ordering Queue

...

...

...

ACK/NAK HARQ Process ID, New Data Indicator

User Data

Signaling

MAC-hs

UE MAC-hs Architecture

When the UE Physical Layer decodes a data block addressed to it, the associated HARQ process determines whether to ACK or NAK the block. If an ACK is sent, the data block is passed to the assigned re-ordering queue.

Re-ordering of MAC-hs PDUs is necessary because up to 8 HARQ processes can be operating onsequentially transmitted data. MAC-hs PDUs can be received out of order when a HARQ process sends a NAK.

The re-ordering queue passes the block up to the disassembly entity when it receives consecutive data blocks. The disassembly entity takes apart the MAC-hs PDU into its constituent MAC-d PDUs and passes them up to the appropriate MAC-d flow for processing by the MAC-d layer.


5-8




Section 5-8UMTS UniversityData Flow Example

Data Flow• Transmit

– RNC RLC PDU to Node B priority queue– Node B MAC-hs PDU assembly– Node B HARQ process

• Receive– UE HARQ process– UE re-ordering queue– UE MAC-hs PDU disassembly

Data Flow Example

In the next set of slides, we will trace the data flow of data blocks from the RNC RLC layer to the UE RLC layer.


5-9





Data Flow Example –RNC MAC-d PDU to Node B Priority Queue

Data Flow Example – RNC MAC-d PDU to Node B Priority Queue

In this example, two logical channels, DTCH 1 and DTCH 2, are mapped to one MAC-d flow. The MAC-d entity in the RNC constructs MAC-d PDUs by prepending a header to each RLC PDU. The MAC-d header contains a C/T field that identifies the DTCH from which the data came. The priority DTCH 1 is higher than DTCH 2, so MAC-d selects all the PDUs from DTCH 1, and then all the PDUs from DTCH 2.

The MAC-d flow is mapped to a MAC-hs priority queue. The RNC transfers the data across the Iub interface to the Node B, where the MAC-hs entity stores the MAC-d PDUs in the priority queue, preserving the order of the PDUs as sent across the Iub interface.


5-10





Data Flow Example –Node B MAC-hs PDU Assembly

Data Flow Example – Node B MAC-hs PDU Assembly

When the scheduler in the Node B MAC-hs decides to send data from a given priority queue, it constructs a MAC-hs PDU. The scheduler determines the size of the MAC-hs PDU as a function of the UE’s CQI report, number of HS-PDSCHs, available transmit power, and other proprietary parameters.

MAC-d PDUs are packed into the MAC-hs PDU sequentially. The MAC-hs PDU is then sent to the Physical Layer as the HS-DSCH transport block. The MAC-hs PDU header consists of the following fields:

Version Flag (VF) – Always set to 0 for this release.Queue Identifier (QID) – Identifies the priority queue in the Node B from which the data came, and the re-ordering queue in the UE to which the data is being sent.Transmission Sequence Number (TSN) – Used by the re-ordering protocol to ensure in-order delivery of MAC-d PDUs when retransmissions occur.Size Index Identifier (SID) – When HSDPA operations begin, the RNC sends a signaling message to the UE that maps valid MAC-d PDU sizes to a set of up to 7 SIDs.Number (N) – Indicates the number of consecutive MAC-d PDUs of the size given by the previous SID. The maximum number of MAC-d PDUs in a MAC-hs PDU is 70.Flag (F) – One-bit flag field to indicate the end of the MAC-hs header.Padding – MAC-hs adds padding as needed to fill the MAC-hs PDU size (transport block size) chosen by the scheduler.


5-11





Data Flow Example –Node B HARQ Process

Data Flow Example – Node B HARQ Process

The scheduler chooses a HARQ process from which to send the PDU. The Node B supports up to 8 HARQ processes for each UE.

The Node B transmits the HARQ process ID in the second part of the HS-SCCH. A one-bit indicator, the New Data Indicator (NDI), in the second part of HS-SCCH is toggled whenever a new PDU is transmitted.

The Physical Layer uses the UE’s H-RNTI to scramble the HS-SCCH. When the UE monitors the HS-SCCH, it looks for subframes scrambled with its H-RNTI, ignoring those that don’t match and processing those that do.

The Node B sends the MAC-hs PDU to the Physical Layer on the HS-DSCH transport channel. The Physical Layer processes the data and maps it onto one or more HS-PDSCHs.


5-12





Data Flow Example –UE HARQ Process

Data Flow Example – UE HARQ Process

Each UE HARQ process performs operations within the Physical Layer and within the MAC-hslayer.

Physical Layer HARQ Process Operations

When the UE decodes its H-RNTI on the HS-SCCH, it prepares to decode the next HS-DSCH TTI. The HS-SCCH includes a HARQ process ID. In the Physical Layer, the HARQ process decodes the associated HS-PDSCHs. If the data is decoded correctly, the data is routed to the MAC-hs part of the HARQ process.

MAC-hs Layer HARQ Process Operations

The MAC-hs HARQ process generates either an ACK or a NAK to be sent in the subframenumbered 5 in the diagram above. If the UE sends an ACK and the Node B decodes the ACK correctly, the earliest that HARQ process 1 can be used for a new data block is the subframenumbered 8 above. If other data blocks are sent to the UE during the intervening subframes, they must be assigned to other HARQ processes.

If the UE supports an inter-TTI interval of 1 (consecutive assignments), then it must support at least 6 HARQ processes. If the Node B scheduler runs slower than shown, up to 8 HARQ processes may be required. No more than 8 may be supported, as the field in the HS-SCCH that identifies the HARQ process is three bits long. The RNC configures the number of HARQ processes for a given UE when HSDPA operation is established and sends this configuration to the UE in a signaling message.


5-13





Data Flow Example –UE Re-ordering Queue

Data Flow Example – UE Re-ordering Queue

When the UE’s HARQ process ACKs the data block, it routes the MAC-hs PDU to a re-ordering queue, according to the QID given in the MAC-hs header. The re-ordering queue uses the TSN in the MAC-hs header to put the PDUs in the correct order. The re-ordering queue routes consecutively received PDUs to the disassembly entity.

If a HARQ process sends a NAK, this can create a hole in the re-ordering queue. The re-ordering queue buffers subsequent PDUs until either the missing PDU is successfully received, or the re-ordering protocol stops waiting for that PDU. Two mechanisms, timer-based and window-based, are used for stall avoidance. These are examined in detail in later slides.

This example illustrates a simple case in which consecutive assignments originate from the same Node B priority queue and thus are all routed to the same re-ordering queue. In a more complicated example, data from multiple priority queues can be interleaved according to the Node B MAC-hs scheduling decisions.


5-14





Data Flow Example –UE MAC-hs PDU Disassembly

Data Flow Example – UE MAC-hs PDU Disassembly

The UE MAC-hs entity disassembles the MAC-hs PDU, using the information in the MAC-hsheader to separate the PDUs. It passes the MAC-d PDUs to the MAC-d entity, which then routes the PDUs to the DTCH logical channels, using the C/T field to differentiate channels.


5-15




Section 5-15UMTS UniversityMAC-hs Quiz

Fill in the fields of the MAC-hs header for the given MAC-d PDUs, using the mapping from MAC-d PDU size to SID.

MAC-hs Quiz

Construct a MAC-hs header for the four MAC-d PDUs, using the mapping from MAC-d PDU size to SID.


5-16




Section 5-16UMTS UniversityMAC-hs Quiz – Answer

Fill in the fields of the MAC-hs header for the given MAC-d PDUs, using the mapping from MAC-d PDU size to SID.

Notes


5-17




Section 5-17UMTS UniversityMAC Multiplexing

MAC Multiplexing

Release 99 allows logical channels to be multiplexed onto a single transport channel, using the C/T field of the MAC header to identify the channel. In HSDPA, logical channels that are multiplexed together by MAC-d are called a MAC-d flow. Each MAC-d flow is then mapped to one or more priority queues. The rules that govern this mapping are:

The maximum number of priority queues is 8, because the QID field in the MAC-hsheader is 3 bits long.A MAC-d flow may be mapped to one or more priority queues, up to a maximum of 8.A priority queue may be mapped to only one MAC-d flow.The maximum number of MAC-d flows is 8, if one-to-one mapping is used between MAC-d flows and priority queues.

The mapping from MAC-d flow to priority queues is implementation dependent and signaled to the UE when HSDPA operation begins.

The mapping from logical channels within one MAC-d flow to priority queues is not strict, and may be established dynamically across the Iub interface. One reason for mapping a MAC-d flow to multiple priority queues would be to send RLC retransmissions at a higher priority than new data.


5-18




Section 5-18UMTS UniversityHARQ Protocol

Features of the HARQ Protocol• Soft combining of multiple transmissions (Layer 1)• N channel Stop and Wait (SAW) protocol

– Up to 8 simultaneous HARQ processes• Synchronous ACK/NAK response• Asynchronous retransmission

– Earliest that Node B can retransmit is 10 ms after first transmission

– Node B scheduling algorithm likely will require 12 ms retransmittime

Could occur later than 12 ms, depending on user and queue priority• Priority pre-emption

– Defer a lower priority retransmission by using another HARQ process

– Flush previous block and send a new block (toggle New Data Indicator)

HARQ Protocol

The HARQ protocol supports the following features:

Soft combining – If the UE NAKs a data block, the Node B may retransmit the data. The Physical Layer performs soft combining of the retransmitted symbols with those previously received.Stop and Wait (SAW) – Each HARQ process, up to a maximum of 8, operates independently on one data block until that block is correctly decoded or transmission is aborted by the Node B.Synchronous ACK/NAK – The UE transmits an ACK or NAK for a given block at a fixed time following reception of the data.Asynchronous retransmission – The Node B sends a retransmission any time after an NAK is received. The earliest this can occur is 10 ms after the previous transmission. A more typical value is expected to be 12 ms, due to internal delays in the Node B scheduling algorithm. A retransmission could occur later than 12 ms depending on channel quality reported by the UE and other internal scheduling decisions.Priority Pre-emption – The Node B can pre-empt a retransmission of a lower priority data block by choosing a different HARQ process, or by flushing the previous block and transmitting new data. The Node B HARQ process toggles the New Data Indicator (NDI) whenever it sends a new data block.


5-19




Section 5-19UMTS UniversityUE HARQ Process Flowchart

UE HARQ Process Flowchart

The control flow for a HARQ process in the UE is as follows:

1. When a data block is received, compare the New Data Indicator (NDI) bit with the value received in the previous block.

If NDI is different, flush data in the buffer and store new dataIf NDI is the same and the buffer is empty, this data has already been decoded correctly, so discard it and send an ACK. This can happen if the Node B interprets an ACK as a NAK, and retransmits the data block.If NDI is the same and the buffer is not empty, soft combine the new data with data already in the buffer.

2. Attempt to decode the data in the buffer.

If correctly decoded, deliver the data to the re-ordering queue, flush the buffer, and send an ACK.If incorrectly decoded, keep the data in the buffer and send a NAK.


5-20




Section 5-20UMTS UniversityHARQ Protocol Signaling on HS-SCCH

HS-SCCH Information

HARQ Protocol HS-SCCH Information

The Node B sends control Information for the HARQ protocol on the HS-SCCH. The first slot of the HS-SCCH is scrambled with the UE’s H-RNTI, which identifies the UE to which this HSDPA assignment belongs. A 16-bit CRC masked with the UE’s H-RNTI is computed over both parts.

The information on HS-SCCH includes:

Channelization Code Set – Which HS-PDSCH codes to use, and how many channels.Modulation Scheme – QPSK or 16-QAMHARQ Process ID – Which HARQ process should decode the next HSDPA assignment.Transport Format Resource Indicator (TFRI) – A 6-bit value that maps to the Transport Block size of the data.Redundancy and Constellation Version – The redundancy version indicates to the Turbo decoder which combination of systematic and parity bits will be sent. For 16-QAM, the constellation version indicates how the symbols were mapped to the constellation.New Data Indicator (NDI) – A 1-bit value that is toggled whenever new data is sent to a given HARQ process, to allow it to distinguish a retransmission from a new transmission.


5-21




Section 5-21UMTS UniversityHARQ Protocol Errors

What if Node B interprets an ACK as a NAK?

HARQ Protocol Errors

Errors can occur in the HARQ protocol if the Node B misinterprets the UE’s ACK/NAK. If the Node B receives nothing in the HS-DPCCH slot in which it expects an ACK or NAK, it treats it as a NAK. If the Node B interprets an ACK as a NAK, a packet may be retransmitted when it was not necessary to do so, using HSDPA bandwidth that could have been used for a new packet or allocated to another UE.

The UE HARQ process detects this condition by the fact that the New Data Indicator bit is the same value as the previous transmission, so it discards the data and sends another ACK.


5-22




Section 5-22UMTS UniversityHARQ Protocol Errors (continued)

What if Node B interprets NAK as a ACK?

HARQ Protocol Errors (continued)

Another type of protocol error occurs if the Node B misinterprets a NAK as an ACK. In this case, the Node B assumes the UE correctly decoded the data, so it sends a new data block to the same HARQ process. The HARQ process must discard the previous transmission and attempt to decode the new block, sending the ACK or NAK accordingly.

This is a worse error than mistaking an ACK, because data is lost and must be recovered by higher layer protocols (e.g., RLC).


5-23




Section 5-23UMTS UniversityHARQ Protocol Error Impact

One extra retransmission• Erroneous NAK

– ACK read as NAK or DTX (misdetection)– Node B conformance spec. requires < 1% probability1

RLC retransmits (AM) or data is lost (UM)• Erroneous ACK

– NAK read as ACK (misdetection) or– UE misses HS-SCCH and DTX read as ACK (false alarm)– Node B conformance spec. requires < 1% probability1

• UE reads HS-SCCH intended for another UE– Soft buffer may be corrupted– Reordering protocol may get confused– Design assumed very low probability due to 16 bit CRC over HS-SCCH

HARQ Protocol Error Impact

In addition to the errors discussed previously, errors can occur due to UE mistakes when reading the HS-SCCH. If the UE misses an HS-SCCH assignment, it sends nothing on the HS-DPCCH.

HARQ protocol errors can be grouped into two categories, according to the impact they have on the overall throughput.

Extra transmission required – If the Node B mistakes an ACK for a NAK, one extra transmission will be required. In the case of a missed assignment, the retransmission must be self-decodable because the UE never received the first transmission.RLC retransmission required – If the Node B misinterprets an ACK, then it will not send a retransmission and the data will be lost, forcing RLC to recover if possible. An even more expensive error of this type occurs if the UE mistakes an HS-SCCH assignment intended for another UE. The UE will attempt to decode a data block, possibly corrupting its soft-combining buffer if there are already symbols present. If the UE correctly decodes the buffer, then upper layers must detect that this data does not belong in the data stream (e.g., ciphering will fail). The probability of this is low, as the UE identity encoded on the HS-SCCH is protected by a 16-bit CRC.

Of these, RLC retransmission required is the more expensive impact, and therefore the system was designed to minimize the probability of these types of errors.1Node B conformance requirements for ACK detection are specified in Release 6.


5-24




Section 5-24UMTS UniversityRe-ordering Protocol

Features of the MAC-hs Re-ordering Protocol• Each reordering queue operates independently• Control information in MAC-hs header

– Queue ID (QID), 3 bits– Transmission Sequence Number (TSN), 6 bits

• In-sequence delivery of MAC-d PDUs to RLC– HARQ protocol may deliver data out of sequence– RLC requires in-sequence delivery

• Uses transmit windows to eliminate sequence number ambiguity• Two mechanisms for flushing missing PDUs

– Timer mechanism– Window mechanism

Re-ordering Protocol

The re-ordering protocol operates independently on each re-ordering queue, which corresponds to each priority queue transmitted by the Node B. The re-ordering protocol supports the following features:

MAC-hs header information – The MAC-hs header contains a re-ordering queue identifier. Re-ordering queues are in a one-to-one mapping with the Node B priority queues. The MAC-hs header also contains a 6-bit sequence number which identifies the transmission order of MAC-hs PDUs.In-sequence delivery – RLC requires in-sequence delivery of MAC-d PDUs, but the HARQ protocol may deliver data out of sequence due to retransmissions. The re-ordering protocol puts the MAC-hs PDUs back into the original transmit order from each of the Node B priority queues.Transmit Window – The Node B uses a transmit window to ensure that it does not overflow the 6-bit sequence number and introduce sequence number ambiguity.Flushing Missing PDUs – The re-ordering protocol recovers from the case where a PDU is never successfully received. This can occur when the Node B misinterprets an NAK as an ACK, and therefore never retransmits the PDUts or when the Node B aborts transmission of a PDU due to reaching the maximum number of retransmissions.


5-25




Section 5-25UMTS UniversityRe-ordering Protocol – MAC-hs Header

Re-ordering Protocol – MAC-hs Header

The MAC-hs PDU header consists of the following fields:

Version Flag (VF) – Always set to 0 for this release.Queue Identifier (QID) – Identifies the priority queue in the Node B from which the data came, and the re-ordering queue in the UE to which the data is being sent.Transmission Sequence Number (TSN) – Used by the re-ordering protocol to ensure in-order delivery of MAC-d PDUs when retransmissions occur.Size Index Identifier (SID) – When HSDPA operations begin, the RNC sends a signaling message to the UE that maps valid MAC-d PDU sizes to a set of up to 7 SIDs.Number (N) – Indicates the number of consecutive MAC-d PDUs of the size given by the previous SID. The maximum number of MAC-d PDUs in a MAC-hs PDU is 70.Flag (F) – One-bit flag field to indicate the end of the MAC-hs header.Padding – Padding as needed to fill the scheduled MAC-hs size.


5-26





Re-ordering Protocol –In-sequence Delivery of MAC-hs PDUs

Re-ordering Protocol – In-sequence Delivery of MAC-hs PDUs

When a HARQ process sends a NAK, a hole is created in the re-ordering queue for which that PDU was intended. As subsequent PDUs are received, the re-ordering queue buffers those PDUsto prevent them from being delivered out of order to the RLC layer above MAC-hs.

When the missing PDU is received correctly, the re-ordering queue inserts it into the correct position in the buffer and delivers it and all subsequent consecutive PDUs that are awaiting delivery.


5-27




Section 5-27UMTS UniversityRe-ordering Protocol Quiz

Question:Is it possible for the HARQ process to indicate to the re-ordering queue that a PDU has been NAK’ed?

Re-ordering Protocol – Quiz

Hint: Where is the Queue ID stored?


5-28




Section 5-28UMTS UniversityRe-ordering Protocol Quiz – Answer

Question:Is it possible for the HARQ protocol to indicate to the re-ordering protocol that a PDU has been NAK’ed?

Answer: NoThe Queue ID is sent in the MAC-hs header. The HARQ process doesn’t know to which re-ordering queue the PDU belongs until it successfully decodes the PDU.

Notes


5-29




Section 5-29UMTS UniversityRe-ordering Protocol – Transmit Window

Re-ordering Protocol – Transmit Window

The Node B MAC-hs uses a transmit window size to prevent sequence number ambiguity. The sequence number space is 64. The allowed values of the transmit window size are 4, 8, 12, 16, and 32. The window size is implementation dependent, and may be set independently for each re-ordering queue.

After the Node B has transmitted a MAC-hs PDU with TSN=SN, any MAC-hs PDU with

TSN ≤ SN – TRANSMIT_WINDOW_SIZE

should not be retransmitted.

The upper edge of the transmit window is advanced whenever the Node B sends a new MAC-hsPDU. The lower edge of the transmit window is advanced when the Node B receives an ACK for the oldest PDU, or when the Node B aborts retransmission of that PDU and discards it.


5-30





Flushing the Re-ordering Queue –Window Method

Flushing the Re-ordering Queue – Window Method

The Node B transmit window and the UE receive window work together to ensure that the transmitter does not overrun the receiver and to flush the receiver’s queue if the transmitter discards a PDU.

The example shown illustrates the following steps, for a Transmit Window Size and Receive Window Size of 4:

1. Node B transmits 4 PDUs. The first PDU is NAKed, so the transmitter stops at sequence number 3. It cannot send any more new PDUs until the missing PDU is correctly received or until retransmission of that PDU is aborted. PDU with sequence number 2 is also NAKed.

2. Node B decides to discard PDU with sequence number 0, and transmits two new PDUswith sequence numbers 4 and 5, advancing the transmit window. The UE does not receive PDU 4 correctly, but does receive a PDU 5 which lies outside the receive window. The UE advances the receive window to include the PDU 5, and delivers all correctly received consecutive PDUs that are below the receive window to the upper layer.


5-31





Flushing the Re-ordering Queue –Window Method (continued)

Flushing the Re-ordering Queue – Window Method (continued)

Continuing the previous example:

3. Node B retransmits PDU 2, and the UE correctly decodes it and sends an ACK. The UE delivers all consecutive PDUs within the receive window.

4. Node B misinterprets the ACK for PDU 2 as a NAK, and retransmits it again. The PDU is within the UE’s receive window, but the UE has already decoded and delivered the PDU. The UE discards the PDU as a duplicate, but sends another ACK.


5-32





Flushing the Re-ordering Queue –Timer Method

Flushing the Re-ordering Queue – Timer Method

The window method is a fast method for flushing PDUs that the Node B has decided to discard, or if the Node B mistakes a NAK for an ACK and advances the transmit window without retransmitting a PDU that the UE has not correctly received. But a problem can also arise if the UE misses the last PDU in a series. If the UE doesn’t decode the HS-SCCH correctly, or if the Node B mistakes a NAK for an ACK, the UE’s re-ordering queue could stall. A timer mechanism is used to avoid this problem.

The example shown illustrates the following steps:

1. Node B transmits 4 PDUs. The UE does not receive PDU 0, and the Node B has no more data to send after PDU 3. The UE starts a timer for PDU 0 when it receives PDU 1.

2. When the timer expires, the UE delivers all consecutive PDUs within the receive window with sequence number larger than the one for which the timer was running.


5-33





Summary of Re-ordering Queue Flushing Methods

• Window Method– Used for flushing holes and eliminating sequence number

ambiguity– Window size is configured independently for each queue– UE advances receive window when a PDU is received outside the

window– UE delivers PDUs that are outside the window after advancing

• Timer Method– Single timer for each queue– UE starts a timer when a hole is detected– When timer expires, deliver all PDUs before the hole, and

consecutive PDUs after the hole– If there is still a hole, start a new timer

• Missing PDUs resolved by RLC layer– RLC AM will request retransmission– RLC UM detects missing PDUs but no retransmission is possible

Summary of Re-ordering Queue Flushing Methods

Two queue flushing methods are used to avoid re-ordering queue stalls. The window method is the faster method, as long as data is flowing steadily from Node B. The timer method is a slower method, but is needed to avoid stalling when there is a gap in data flow and there are still holes in the re-ordering queue.

It is RLC’s responsibility to recover from missing MAC-hs PDUs. MAC-hs PDUs are disassembled into MAC-d PDUs, each of which contains an RLC PDU. Each RLC PDU contains a sequence number. RLC detects when there are gaps in the sequence numbers. In Acknowledged Mode (AM), the receiving RLC can request retransmission of the missing RLC PDUs. In Unacknowledged Mode (UM), the receiving RLC can detect missing PDUs, but no retransmission is possible. Protocol layers above RLC UM must recover from the missing data.


5-34




Section 5-34UMTS UniversityRLC Considerations

• HSDPA Cell Repointing Procedure can cause loss of data– Acknowledged Mode (AM) can recover– Unacknowledged Mode (UM) cannot recover– Synchronized procedure can minimize data loss, but is

slow

• Sequence Number Space– Higher data rates of HSDPA can cause sequence

number wrap

RLC Considerations

The RLC protocol is impacted by the higher data rates of HSDPA and the HSDPA Cell Re-pointing Procedure. These are examined in the next two slides.


5-35





RLC Considerations –HSDPA Cell Re-pointing Procedure

Flush MAC-d PDUs already transferred to previous Node B

• RLC UM channels will lose data• RLC AM channels will require RLC retransmissions

RLC Considerations – HSDPA Cell Re-pointing Procedure

When the serving Node B changes due to UE mobility, MAC-d PDUs that have already been transferred from the RLC layer in the RNC to the MAC-hs layer in the Node B will be lost.

For RLC UM, there is no mechanism to recover these PDUs. UM uses a sequence number space of 7 bits, so up to 128 PDUs could be lost.

For RLC AM, the PDUs can be recovered, but require RLC retransmissions. Release 5 specifications require the UE RLC layer to send a status report following HSDPA Cell Repointing Procedure. The status report carries RLC positive and/or negative acknowledgments so the RNC RLC layer can determine which PDUs need to be retransmitted.

Section 5 discusses a synchronized procedure that minimizes data loss and RLC retransmissions, but adds significant delay.


5-36





RLC Considerations –Sequence Number Space and Data Rates

• RLC AM– Sequence Number Space 12 bits– Typical Release 99 RLC PDU size is 320 bits

Limits HSDPA data rate to 1.6 Mbps, due to RLC protocol limitations and round trip time limit

– Solution: use 640 bit RLC PDUs

• RLC UM– Sequence Number Space 7 bits – Up to 128 PDUs could be lost during HSDPA Cell Repointing

Procedure– Ciphering desynchronization could occur– Solution: use UM only services with average data

rate < 40 kbpsCan still achieve high instantaneous data rates

RLC Considerations – Sequence Number Space and Data Rates

The sequence number spaces of RLC AM and UM affect the maximum data rate that can be achieved with HSDPA.

RLC AM

RLC AM uses a sequence number space of 12 bits. In Release 99, a typical RLC PDU size is 320 bits. The observed round trip time is about 200 ms. RLC AM uses a status prohibit timer to prevent the receiver from sending status reports more frequently than the round trip time. These factors combine to produce a theoretical maximum data rate of 1.6 Mbps, which is well below the theoretical maximum of the HSDPA Physical Layer.

The recommended solution is to double the RLC PDU size to 640 bits.

RLC UM

The RLC UM sequence number space is only 7 bits. Up to 128 PDUs could be lost during HSDPA Cell Repointing Procedure, which could lead to ciphering desynchronization.

The recommended solution is to use RLC UM only for applications with a low average data rate (e.g., bursty applications that send a small amount of data at a time). Note that an application can still take advantage of HSDPA’s high instantaneous data rates, even though its average data rate might be quite low.


5-37





1. Where in the network architecture does the MAC-hs layer reside, and why?

2. How many MAC-hs Priority Queues may be allocated per UE?

3. What is the purpose of the SID field in the MAC-hs header, and what values does it assume?

4. What entity within the UE is responsible for generating ACK or NAK for HSDPA data blocks?

5. What entity within the UE is responsible ensuring in-order delivery of MAC-d PDUsto RLC?

6. List 2 mechanisms for flushing UE re-ordering queues, and explain why both are needed.

7. What happens to data at the previous Node B MAC-hs layer when a HSDPA Cell Repointing Procedure occurs?

Notes


5-38





1. Where in the network architecture does the MAC-hs layer reside, and why?

• Node B• To support high data rates of HSDPA, the MAC-hs layer must be close to

the Physical Layer.

2. How many MAC-hs Priority Queues may be allocated per UE?• 8, because the QID field in the MAC-hs header is 3 bits

3. What is the purpose of the SID field in the MAC-hs header, and what values does it assume?

• It indicates the size of a group of consecutive MAC-d PDUs that are packed into the MAC-hs PDU.

• Values are signaled to the UE by RNC when HSDPA channels are established.

Notes


5-39





4. What entity within the UE is responsible for generating ACK or NAK for HSDPA data blocks?

• HARQ Process

5. What entity within the UE is responsible ensuring in-order delivery of MAC-d PDUs to RLC?

• Re-ordering Protocol

6. List 2 mechanisms for flushing UE re-ordering queues, and explain why both are needed.

• The window mechanism allows for fast flushing during periods of continuous data transfer.

• The timer mechanism flushes data that might otherwise stall in the MAC-hs layer during periods when the Node B has no new data to send to the UE.

7. What happens to data at the previous Node B MAC-hs layer when a HSDPA Cell Repointing Procedure occurs?

• For RLC UM, it is lost.• For RLC AM, the RLC layer must retransmit missing PDUs.

Notes


5-40





Layer 2 Protocols –What We Learned

HSDPA Network and UE protocol architecture. Example data flow from RNC MAC-d to UE MAC-d.MAC-d multiplexing and how MAC-d flows are mapped to priority queues.HARQ protocol and how error scenarios affect the protocol.Re-ordering protocol.HSDPA impact on RLC protocol.

Notes


6-1

Section 6: Layer 3 Protocols and Procedures80-W0331-1 Rev B




Section 6:Layer 3 Protocols and Procedures

6SECTION

Layer 3 Protocols and Procedures

Notes


6-2





Define HSDPA terminology used by RRC and signaling layers.Illustrate call flow for setting up HSDPA operations.Examine UE measurements and associated messaging used in HSDPA.Define the messages and information elements used to configure HSDPA.Illustrate call flows for changing UE states while starting and stopping HSDPA.Illustrate call flows for performing the HSDPA Cell Repointing Procedure.Discuss some problem areas in Release 5 that will be addressed in Release 6.

Notes


6-3






25.308 High Speed Downlink Packet Access (HSDPA) Overall Description Stage 2

25.331 Radio Resource Control (RRC) Protocol Specification

References

Notes


6-4




Section 6-4UMTS UniversityHSDPA RRC Functions

• HSDPA Channel Configuration– Start and stop HSDPA operations

• Call State Management– RRC must be in Cell_DCH when HSDPA is active

• Active Set Update Procedure– Active Set are cells for which a DCH radio link exists– Maximum of 8 cells– Procedure adds and removes DCH radio links

• HSDPA Cell Re-pointing Procedure– Serving HS-DSCH Radio Link is the cell transmitting HS-DSCH– Maximum of 1 cell– Procedure changes Serving HS-DSCH Radio Link from one cell

to another cell

HSDPA RRC Functions

The RRC layer is responsible for signaling procedures related to HSDPA operations. This includes:

HSDPA Channel Configurations – The RNC RRC layer is responsible for starting and stopping HSDPA operations, and for configuring the radio bearers, logical channels, and physical channels to support HSDPA.Call State Management – The UE RRC layer must be in Cell_DCH state with an established DCH transport channel and one or more DPCH physical channels when using HSDPA channels. The RNC RRC layer manages UE state changes to and from Cell_DCHstate.Active Set Update Procedure – All cells with an established DCH radio link are members of the UE’s Active Set. The UE provides measurement information to the RNC,and the RNC signals changes to the Active Set using the Active Set Update message.HSDPA Cell Re-pointing Procedure – The RNC RRC layer is responsible for establishing an HSDPA radio link, which is called the Serving HS-DSCH Radio Link. In response to UE measurement information, the RNC RRC layer signals changes to the Serving HS-DSCH Radio Link using a reconfiguration message.


6-5




Section 6-5UMTS UniversityMobile Originated PS Data Call Setup

Mobile Originated PS Data Call Setup

A mobile originated PS data call begins by establishing an RRC connection and transitioning the UE to either Cell_DCH or Cell_FACH state. Signaling Radio Bearers are established first to exchange messages, then Radio Access Bearers are established to carry the user data. The message sequence is:

1. Call setup always begins with RRC connection establishment. The RRC Connection Setup message indicates whether the UE should operate in Cell_DCH or Cell_FACH.

2. GPRS Mobility Management (GMM) indicates the desire for service, but first the UE is authenticated. Note that the authentication step is optional.

3. After authentication, ciphering and integrity protection are enabled with the Security Mode Command. The security procedure serves as an implicit GMM Service Accept.

4. The Session Manager (SM) requests that the PDP context be activated. This step may include QoS negotiation which indicates the need for HSDPA operation.

5. The PS Core Network requests UTRAN to establish the Radio Bearers that will carry the user data. If high speed data transfer is required, UTRAN will establish DCH Radio Bearers and assign the UE to Cell_DCH state if it is not already there. This step may include establishing HSDPA operation, or that may occur later after the UE sends a Measurement Report message.

6. After the PDP Context is activated, higher layers of the data protocol stack may perform any negotiation or setup required to allow data to begin flowing.


6-6




Section 6-6UMTS UniversityEstablish HSDPA Operation from Cell_DCH

Establish HSDPA Operation from Cell_DCH

The sequence of messages to establish HSDPA operation are as follows:

1. With the UE already in Cell_DCH state, the RNC decides to start HSDPA operation, based on availability of Downlink data from the PS Core Network and QoS negotiations that occurred at the NAS layer when the data call was established.

2. RNC requests that the UE perform measurements to determine the best cell in its Active Set. Event 1d causes the UE to send a report whenever there is a change in the best cell. Release 5 modified this event to require the UE to immediately report the initial best cell when the measurement is setup.

3. RNC notifies the Node B that controls the best cell to set up an HS-DSCH radio link on that cell.

4. RNC sends a reconfiguration message to notify the UE that the HS-DSCH is available, and to signal all the parameters associated with HSDPA operations. The message may be any one of the following messages: Radio Bearer Setup, Radio Bearer Reconfiguration, Radio Bearer Release, Transport Channel Reconfiguration, Physical Channel Reconfiguration.


6-7





UE Measurements –Change of Best Cell (Event 1d)

Reporting event 1D

Measurement quantity

Time

P CPICH 2

P CPICH 1

P CPICH3

UE Measurements – Change of Best Cell (Event 1d)

UTRAN may configure the UE to measure and report when there is a change of best cell among those cells in the Active Set. UTRAN may use this to trigger a Serving HS-DSCH Cell Change procedure.

The measurement quantity may be pathloss, CPICH Ec/No, or CPICH RSCP. When the quantity is pathloss, 25.331 gives the following equation to determine change of best cell:

When the quantity is CPICH Ec/No or CPICH RSCP, 25.331 gives the following equation to determine change of best cell:

In both equations:

M represents the measured result of a cell.H represents the hysteresis value for event 1d sent in the Measurement Control Message.CIO represents the cell individual offset for a particular cell, given in the Measurement Control Message.

2/10101dBestBestNotBestNotBest

HCIOLogMCIOLogM ++⋅≥+⋅

2/10101dBestBestNotBestNotBest

HCIOLogMCIOLogM −+⋅≤+⋅


6-8





UE Measurements –Measurement Control Message

UE Measurements – Measurement Control Message

The Measurement Control Message may be used to configure many different types of measurements. Those pertaining to HSDPA operations are intra-frequency measurements.


6-9





UE Measurements–Event 1d Parameters

UE Measurements– Event 1d Parameters

Parameters pertaining to Event 1d include:

Cell Info List – Configures the cells to measure by listing cells to remove and cells to add. For cells to add, parameters such as the cell individual offset, reference time difference to the cell, primary scrambling code, and primary CPICH transmit power may be included.Measurement Quantity – Defines the filter coefficient and the quantity that the UE should measure (pathloss, CPICH Ec/No, or CPICH RSCP).Reporting Quantity – Defines reporting quantities in addition to those quantities that are mandatory for the event.Reporting Cell Status – Defines the types and number of cells to be reported.Measurement Validity – Defines the UE states in which measurements are to be performed: Cell_DCH, all states except Cell_DCH, or all states.Triggering Condition – Defines which sets of cells may trigger the event.Hysteresis – Hysteresis value used in the best cell calculations.Time to Trigger – Time between detecting the event and sending the report message.Use CIO – Boolean that indicates whether or not to use the cell individual offsets in the best cell calculations.


6-10




Section 6-10UMTS UniversityHSDPA Configuration – Messages

Messages to Set up HSDPA Configuration• Radio Bearer Setup• Radio Bearer Reconfiguration• Radio Bearer Release• Cell Update Confirm

Messages to Modify L1 and MAC-hs only• Transport Channel Reconfiguration

Messages to Modify L1 only• Physical Channel Reconfiguration

HSDPA Configuration – Messages

Several messages may be used to control HSDPA operation.

Messages to Set up HSDPA Configuration

Messages that contain all of the necessary IEs are Radio Bearer Setup, Radio Bearer Reconfiguration, Radio Bearer Release, and Cell Update Confirm. These message can assign a new H-RNTI, indicate the mapping from logical channel to HS-DSCH, and provide all the MAC-hs and Layer 1 parameters.

Cell Update Confirm message may be used to transition directly from Cell_FACH state into Cell_DCH and simultaneously activate HSDPA operations.

Radio Bearer Release message may be used to activate HSDPA operations at the same time that some radio bearers are released. This may occur in a concurrent call scenario, for example, when voice radio bearers are released but data radio bearers remain active.

Messages to Modify L1 and MAC-hs HSDPA Configuration

The Transport Channel Reconfiguration message can assign a new H-RNTI and alter MAC-hs or Layer 1 parameters, but cannot establish or modify logical channel mappings.

The Physical Channel Reconfiguration message can only assign a new H-RNTI and alter Layer 1 parameters (including Serving HS-DSCH Radio Link).


6-11





HSDPA Configuration –Radio Bearer Reconfiguration Message

HSDPA Configuration – Radio Bearer Reconfiguration Message

The Radio Bearer Reconfiguration message contains all the IEs needed to establish or modify the HSDPA configuration, including:

Radio Bearer Mapping Information – Existing RBs may be reconfigured to map the data flow onto HSDPA channels. Other existing RBs may be affected by these changes. New RBs cannot be setup with this message.MAC-hs Information – IEs to configure the MAC-hs, including HARQ processes, re-ordering queues, and MAC-hs header.Layer 1 Information – IEs to configure the Layer 1, including which HS-SCCH codes to monitor, which member of the Active Set is the Serving HS-DSCH Radio Link, and CQI parameters.


6-12





HSDPA Configuration –Layer 1 Information Elements

• Downlink Scrambling Code• HS-SCCH Channelization Codes• Serving HS-DSCH Radio Link Indicator• Measurement Feedback Information

– Power offsets– CQI feedback cycle and repetition factor

HSDPA Configuration – Layer 1 Information Elements

UTRAN signals the following Information Elements to the UE to configure Layer 1 HSDPA operations:

Downlink Scrambling Code – Defaults to the Primary Scrambling Code.HS-SCCH Channelization Codes – Set of OVSF codes which the UE must monitor to decode the HS-SCCH. Up to four may be assigned to a UE.Serving HS-DSCH Radio Link Indicator – Of all the DPCH radio links configured, UTRAN identifies one as the Serving HS-DSCH Radio Link.Measurements Feedback Information – Parameters that control CQI feedback sent on HS-DPCCH, including power offset, feedback cycle, and repetition factor.

These IEs may appear in any of the messages that configure HSDPA operation.


6-13





HSDPA Configuration –Layer 2 Information Elements

• Radio Bearer Mapping Information– DCCH/DTCH mapping to HS-DSCH or DCH + HS-DSCH– MAC-d flow ID (0..7)

• MAC-hs Information– Number of HARQ Processes (1..8)– Division of UE soft buffer bits between processes– For each re-ordering queue:

Which MAC-d flow maps to itReordering timerWindow sizeMAC-d PDU size to SID mapping

HSDPA Configuration – Layer 2 Information Elements

UTRAN signals the following information elements to the UE to configure Layer 2 HSDPA operations:

Radio Bearer Mapping Information – Indicates which logical channels (DCCH or DTCH) are mapped to HS-DSCH. Note that a logical channel may be mapped to both DCH and HS-DSCH. Also indicates mapping from logical channel to MAC-d flow.MAC-hs Information – Indicates the number of HARQ processes used and whether to divide the soft buffer bits equally among processes or to use an explicit assignment of buffer sizes. For each re-ordering queue, the mapping from MAC-d flow, timer and window size, and Size Index Identifier (SID) mapping are given. The SID mapping defines the mapping between the MAC-d PDU size and the SID field of the MAC-hs header.


6-14




Section 6-14UMTS UniversityUE State Transitions

• Cell_DCH -> Cell_DCH (start or stop HS-DSCH)• Cell_DCH -> Cell_FACH (stop HS-DSCH and DCH)• Cell_FACH -> Cell_DCH (start HS-DSCH and DCH)

UE State Transitions

Packet data operations are inherently bursty. During the course of HSDPA operations, the UE may have periods of time where the high-speed data transfer is not needed. UTRAN internal algorithms determine when the UE transitions in and out of HSDPA operations. When not using HSDPA, UTRAN decides whether to continue to operate in Cell_DCH or to transition to Cell_FACH where only lower data rates are possible.


6-15





UE State Transitions –Stopping HS-DSCH in Cell_DCH

UE State Transitions– Stopping HS-DSCH in Cell_DCH

UTRAN can decide at any time to stop HS-DSCH operations for a given UE. If the Downlink data activity falls below an internal threshold, or if HSDPA operations become impractical due to high mobility of the UE, UTRAN may decide to stop HS-DSCH but keep the UE operating in Cell_DCH.

1. With the UE in Cell_DCH state, the RNC decides to stop HSDPA operation.2. RNC notifies the Node B that controls the Serving HS-DSCH Radio Link stop

transmitting to the UE on the HS-DSCH.3. RNC sends a reconfiguration message to notify the UE that the HS-DSCH is not

available. This is accomplished by removing Radio Bearer mappings to HS-DSCH, removing MAC-d flows, and/or indicating that no radio links are the Serving HS-DSCH Radio Link. The message may alter the data rate of the DCH. The message may be any one of the following messages: Radio Bearer Setup, Radio Bearer Reconfiguration, Radio Bearer Release, Transport Channel Reconfiguration, or Physical Channel Reconfiguration.


6-16





UE State Transitions –Cell_DCH to Cell_FACH

UE State Transitions – Cell_DCH to Cell_FACH

UTRAN can decide at any time to stop HS-DSCH operations for a given UE. If the Downlink data activity becomes very low, UTRAN may decide to stop HS-DSCH and put the UE in Cell_FACH state. Uplink and Downlink data can still be transferred, but only at lower data rates on the common FACH and RACH transport channels.

In Cell_FACH state, the UE communicates with only one cell. The message that directs the UE to Cell_FACH indicates a single radio link chosen by UTRAN. If that cell is not the best cell, the UE may perform a cell reselection and Cell Update procedure after entering Cell_FACH.

1. With the UE in Cell_DCH state, the RNC decides to stop HSDPA operation and transition the UE to Cell_FACH.

2. RNC notifies the Node B that controls the Serving HS-DSCH Radio Link stop transmitting to the UE on the HS-DSCH.

3. RNC sends a reconfiguration message to notify the UE that the HS-DSCH is not available and to transition to Cell_FACH state. This is accomplished by remapping DCCH/DTCH logical channels to RACH and FACH. The message may be any one of the following messages: Radio Bearer Setup, Radio Bearer Reconfiguration, and Radio Bearer Release.


6-17





UE State Transitions –Cell_FACH to Cell_DCH with HS-DSCH

UE State Transitions – Cell_FACH to Cell_DCH with HS-DSCH

For a UE operating in Cell_FACH, UTRAN may decide to transition to Cell_DCH and start HSDPA operation in one step.

1. With the UE in Cell_FACH state, the RNC decides to start HSDPA operation, based on availability of Downlink data from the PS Core Network and QoS negotiations that occurred at the NAS layer when the data call was established.

2. RNC notifies the Node B that controls the best cell to set up an HS-DSCH radio link on that cell. The DCH radio links are set up at the same time.

3. RNC sends a reconfiguration message to map DCCH/DTCH channels to HS-DSCH and DCH. UTRAN must pick a Serving HS-DSCH Radio Link, and will most likely choose the cell which the UE used in Cell_FACH state.

4. RNC requests that the UE perform measurements on its Active Set to determine the best cell by setting up event 1d. UE reports immediately the current best cell, and continues to report whenever the best cell changes.

A two-step process is also possible, in which the DCH is set up in the transition to Cell_DCH, and then HS-DSCH is added after the UE reports the best cell. However, this takes much longer and likely will not be used.


6-18





HSDPA Cell Re-pointing Procedure –Overview

Node B Node B

Active set DPCH

Node B Node B

Active set DPCH

Before After

HSDPA Channels

HSDPA Channels

HSDPA Cell Re-pointing Procedure – Overview





6-19





HSDPA Cell Re-pointing Procedure –Messages and Information Elements

• Serving HS-DSCH Radio Link Indicator IE– Contained in one of 6 reconfiguration messages

Cell Update ConfirmRadio Bearer Setup, Release, ReconfigurationTransport Channel ReconfigurationPhysical Channel Reconfiguration

• NOT in the Active Set Update Message!– HSDPA operation stops if Serving HS-DSCH Radio

Link is removed from Active Set– UTRAN should perform HSDPA Cell Re-pointing

Procedure before removing the HSDPA radio link from the Active Set.

HSDPA Cell Re-pointing Procedure – Messages and Information Elements

To perform the HSDPA Cell Re-pointing Procedure, UTRAN sends any of the messages that contain the Serving HS-DSCH Radio Link Indicator IE.

This information element is not contained in the Active Set Update message. If a new cell that is not yet in the Active Set is also the best possible cell for HSDPA operations, that cell cannot become the Serving HS-DSCH Radio Link in one step. An Active Set Update procedure mustoccur first to establish the DCH and add that cell to the Active Set. After that, the HSDPA Cell Repointing Procedure may be performed.

If the cell that is the Serving HS-DSCH Radio Link is removed from the Active Set before a new cell is chosen to be the serving cell, HSDPA operations stop and must be restarted. In general, UTRAN should avoid doing this by performing the HSDPA Cell Re-pointing Procedure first, then removing the old radio link from Active Set. However, sometimes rapidly changing radio conditions do not allow UTRAN to perform these operations sequentially, and the UE must be able to handle the sudden loss of the Serving HS-DSCH Radio Link.


6-20





HSDPA Cell Re-pointing Procedure –Synchronized vs. Unsynchronized

Synchronized Cell Change• Uses an action time at which the change occurs.• Introduces significant delay in procedure (600 ms typical).• Very little data will need to be retransmitted.• Not suitable for very high data rate services.

Unsynchronized Cell Change• Change happens ASAP.• May lose data when MAC-hs buffers are flushed at old Node B.• May introduce jitter in data delivery due to AM retransmissions.• May be used if the old and new cells are controlled by the same

Node B.

HSDPA Cell Re-pointing Procedure– Synchronized vs. Unsynchronized

The HSDPA Cell Re-pointing Procedure may be either synchronized or unsynchronized. From the UE’s perspective, the difference is whether or not an action time is included in the message that triggers the cell change.

From the network’s perspective, a synchronized procedures gives the previous Node B a chance to transmit any data in its HS-DSCH buffers and thus minimize data loss. However, the procedure introduces significant delay, during which the UE continues to operate in a sub-optimal cell, and therefore is not suitable for very high data rate services.

The unsynchronized procedure is much faster, but introduces either data loss (for UM services), or delay due to retransmissions (for AM services). It is suitable for cell changes that do not change Node Bs, as the MAC-hs buffers can be transferred to the new cell without data loss.


6-21





HSDPA Cell Repointing Procedure –Synchronized Inter-Node B

HSDPA Cell Repointing Procedure – Synchronized Inter-Node B

A Inter-Node B synchronized Serving HS-DSCH Cell Change procedure consists of the following steps (assuming the new cell is not already in the Active Set):

1. A new Node B is added to the Active Set, in response to measurements reported by the UE. The target Node B is first instructed to set up a radio link, then the RNC sends the Active Set Update message to the UE. The UE now has a DPCH on both target and source Node Bs, and an HS-DSCH on the source Node B.

2. The UE reports that the best cell is the one just added to the target Node B. This triggers the beginning of the HSDPA Cell Repointing Procedure. RNC sends a message to the current Serving HS-DSCH Node B to prepare to release the HS-DSCH. This gives the Node B a chance to finish transmitting data already in its MAC-hs buffers. The commit message contains an action time that specifies the time after which no more HS-DSCH data is to be sent by this Node B.

3. RNC instructs the target Node B to establish HSDPA channels for the UE. The commit message contains the same action time, indicating the time at which HS-DSCH operations begin for this UE.

4. RNC sends a reconfiguration message to change the Serving HS-DSCH Radio Link. This message contains the same action time, at which time the UE begins receiving HS-DSCH data from the new cell.


6-22





HSDPA Cell Repointing Procedure –Unsynchronized Inter-Node B

HSDPA Cell Repointing Procedure – Unsynchronized Inter-Node B

A Inter-Node B unsynchronized HSDPA Cell Repointing Procedure consists of the following steps (assuming the new cell is not already in the Active Set):

1. A new Node B is added to the Active Set, in response to measurements reported by the UE. The target Node B is first instructed to set up a radio link, then the RNC sends the Active Set Update message to the UE. The UE now has a DPCH on both target and source Node Bs, and an HS-DSCH on the source Node B.

2. The UE reports that the best cell is the one just added to the target Node B. This triggers the beginning of the HSDPA Cell Repointing Procedure. RNC sends a message to the target Node B instructing it to establish HSDPA channels for the UE.

3. RNC sends a message to the Serving HS-DSCH Node B to release the HS-DSCH.4. RNC sends a reconfiguration message to notify the UE that the HS-DSCH is available on

the new Serving HS-DSCH Radio Link. No action time is specified, so the UE makes the change as soon as possible.

During an unsynchronized change, there may be a period of time in which the UE receives no data, or some data may be lost if the UE does not make the change fast enough, and the new Serving Node B has already begun transmitting. Data may also be lost due to flushing the buffers at the old Serving Node B.


6-23




Section 6-23UMTS UniversityProblem Areas in Release 5 Signaling

• Significant delays in HSDPA procedures because:– Active Set Update cannot change Serving HS-DSCH

Radio LinkFixed in Release 6

– Synchronized procedure requires activation time (~600 ms)– Race condition in Radio Link removal can lead to stoppage

of HSDPA• Delays could impact system capacity for high

mobility.– HSDPA best performance achieved for stationary users

Problem Areas in Release 5 Signaling

Several limitations of Release 5 signaling introduce significant delays for high mobility users. The standards organization is working on solutions, but these will only become available in Release 6.

Active Set Update cannot change Serving HS-DSCH Radio Link – This limitation introduces extra signaling when a new DCH radio link must be added before the HSDPA Cell Repointing Procedure can occur. It would be better if the HS-DSCH and DCH could be established in a single message.Synchronized Procedure Activation Time – The synchronized procedure is the preferred method for minimizing data loss, but the current activation time mechanism introduces a long delay.Race Condition in Radio Link Removal – Different processing elements in the RNC may be involved in making the decision for Active Set membership and Serving HS-DSCH Radio Link. A race condition may arise if the Serving HS-DSCH Radio Link is removed from the Active Set before performing the HSDPA Cell Repointing Procedure.

Due to these limitations, continuous (non-bursty) very high data rates possible with HSDPA are likely to be achieved only for stationary users.


6-24





QUIZ

Notes


6-25





1. In what state must the UE be operating for HSDPA operations?

2. What measurement event is typically configured to support HSDPA operations?

3. What messages may UTRAN send to establish HSDPA operations for the first time?

4. List four Physical Layer HSDPA information elements.

5. List four Layer 2 HSDPA information elements.

6. What messages may UTRAN send to perform HSDPA Cell Repointing Procedure?

7. What is the difference between synchronized and unsynchronized HSDPA Cell Re-pointing Procedures?

Notes


6-26





1. In what state must the UE be operating for HSDPA operations?• Cell_DCH

2. What measurement event is typically configured to support HSDPA operations?

• Event 1d Change of Best Cell

3. What messages may UTRAN send to establish HSDPA operations for the first time?

• Radio Bearer Setup• Radio Bearer Reconfiguration• Radio Bearer Release• Cell Update Confirmation

4. List 4 Physical Layer HSDPA information elements.• HS-SCCH Channelization Codes• Serving HS-DSCH Radio Link• CQI Feedback Repetition• CQI Feedback Cycle

Notes


6-27





5. List 4 Layer 2 HSDPA information elements.• Radio Bearer mapping• Reordering queue window size• Reordering queue timer• MAC-d PDU to SID mapping

6. What messages may UTRAN send to perform HSDPA Cell RepointingProcedure?• Radio Bearer Setup• Radio Bearer Reconfiguration• Radio Bearer Release• Cell Update Confirmation• Transport Channel Reconfiguration• Physical Channel Reconfiguration

7. What is the difference between synchronized and unsynchronized HSDPA Cell Re-pointing Procedures?• Synchronized procedure uses an action time to minimize data loss and ensure

that the cell change occurs at the same time for the UE and the source and target Node Bs.

• Unsynchronized procedure occurs ASAP, but may result in data loss or in gaps in the reception of data.

Notes


6-28





Layer 3 Protocols and Procedures –What We Learned

HSDPA terminology.HSDPA call flows.UE measurements and reporting.Messages and information elements to configure HSDPA.UE state transitions.HSDPA Cell Repointing Procedure.Problem areas in Release 5 signaling.

Notes


7-1

Section 7: Summary80-W0331-1 Rev B




Section 7:Summary

7SECTION

Summary

Notes


7-2





• Set of high speed channels• Channels are shared by multiple users• Each user may be assigned all or part of the total

bandwidth every 2 ms.

High Speed Downlink Packet Access (HSDPA)

In HSDPA, the Node B allocates a set of high speed channels. These channels are assigned to a user using a fast scheduling algorithm that allocates the channels every 2 ms. All or part of the channels may be assigned to a given user during any 2 ms period.

The rapid scheduling of HSDPA is well-suited to the bursty nature of packet data. During periods of high activity, a given user may get a larger percentage of the channel bandwidth, while it gets little or no bandwidth during periods of low activity.


7-3






In a Release 99 PS network, the NAS layer protocols are terminated at the SGSN. The RRC, RLC, and MAC protocols are terminated at the RNC. The Physical Layer protocol is terminated at the Node B.


The location of MAC-hs in Node B has an important implication for HSDPA operation. In Release 99, a UE may be in soft handover with multiple Node Bs. Transport channel frames are constructed by the MAC sublayer in the RNC and sent over the Iub interface to each Node B with which the UE is in soft handover. The UE receives identical Transport channel frames from each Node B.

HSDPA requires fast scheduling of the shared channels, and allocates the channels in 2 ms intervals called subframes. To meet this requirement, the Transport channel frames are constructed by the MAC-hs sublayer operating in the Node B. By design, the HSDPA channels cannot operate in soft handover because the MAC-hs sublayer of each Node B operates independently.


7-4





Theoretical Maximum Data Rate is 14.4 Mbps



The following assumptions are needed to achieve the theoretical maximum data rate of 14.4 Mbps:

Multi-code transmission – All 15 HS-PDSCH channels must be assigned to a single UE during one 2 ms TTI. This uses up a significant portion of the OVSF tree, leaving very few codes for non-HSDPA users and overhead channels.Consecutive assignments – The Node B must send back-to-back assignments to a single UE, and the UE must be able to correctly decode every block without requiring retransmission.Lower Coding Gain – Using an effective code rate of 1 increases the data rate, but the channel conditions must be very good for the UE to correctly decode every data block on the first transmission.16-QAM – This modulation scheme works well only in very good channel conditions.

In a practical scenario, the practical maximum data rate will be considerably less than 14.4 Mbps, due to less than ideal channel conditions, the need for retransmission, and the need to share the channel with other HSDPA users and Release 99 users.

Other factors that reduce the practical maximum data rate will be discussed in subsequent slides.


7-5






Release 99 Channel Release 5 HS-DSCHNew Physical Layer blocks in Release 5:

• Bit Scrambling

• Physical Layer HARQ functionality

• HS-DSCH Interleaving

• Constellation re-arrangement for 16-QAM


Channel coding is done to add robustness to the information bits. Usually, channel coding is performed by adding redundant bits determined by an FEC coding scheme such as CRC, convolutional coding, turbo coding, etc. The HS-DSCH channel coding involves a number of other functions performed by the Node B’s Physical Layer. The main reason for this additional processing is the dynamic size of the transport block transmitted in an HS-DSCH TTI. Other reasons include large HS-DSCH payload size and the possible use of 16-QAM modulation for HS-PDSCH. Comparing the coding chain for the Release 99 channel with the Release 5 HS-DSCH channel. Some blocks have been removed and some new blocks have been added.

HS-DSCH coding chain does not require:1. Concatenation, because there is always only one transport block per HS-DSCH TTI. The

transport block size, however, varies from 137 bits to 27952 bits. In case of retransmission, the transport block size remains the same as of the original transmission.

2. First DTX insertion, because HS-DSCH doesn’t support fixed position transport channel and thus Blind Transport Format Detection (BTFD).

3. Second DTX insertion, because there is just one transport channel mapped on to HS-PDSCH.

4. Radio frame segmentation, because HS-DSCH has a fixed TTI of 2 ms, which is equal to the HS-PDSCH subframe duration.

5. Transport channel multiplexing, because there is just one transport channel mapped on to HS-PDSCH.


7-6




Section 7-6UMTS UniversityUE MAC-hs Architecture

Dis-assembly

MAC-d Flows

HS-DSCH Signaling on HS-SCCHSignaling on HS-DPCCH

Re-ordering queue distribution

MAC-d Flow Mapping

HARQ processes

Re-ordering Queue

Dis-assembly

Re-ordering Queue

...

...

...

ACK/NAK HARQ Process ID, New Data Indicator

User Data

Signaling

MAC-hs

UE MAC-hs Architecture

When the UE Physical Layer decodes a data block addressed to it, the associated HARQ process determines whether to ACK or NAK the block. If an ACK is sent, the data block is passed to the assigned re-ordering queue.

Re-ordering of MAC-hs PDUs is necessary because up to 8 HARQ processes can be operating onsequentially transmitted data. MAC-hs PDUs can be received out of order when a HARQ process sends a NAK.

The re-ordering queue passes the block up to the disassembly entity when it receives consecutive data blocks. The disassembly entity takes apart the MAC-hs PDU into its constituent MAC-d PDUs and passes them up to the appropriate MAC-d flow for processing by the MAC-d layer.


7-7





Re-ordering Protocol –In-sequence Delivery of MAC-hs PDUs

Re-ordering Protocol – In-sequence Delivery of MAC-hs PDUs

When a HARQ process sends a NAK, a hole is created in the re-ordering queue for which that PDU was intended. As subsequent PDUs are received, the re-ordering queue buffers those PDUsto prevent them from being delivered out of order to the RLC layer above MAC-hs.

When the missing PDU is received correctly, the re-ordering queue inserts it into the correct position in the buffer and delivers it and all subsequent consecutive PDUs that are awaiting delivery.


7-8





HSDPA Cell Re-pointing Procedure –Overview

Node B Node B

Active set DPCH

Node B Node B

Active set DPCH

Before After

HSDPA Channels

HSDPA Channels

HSDPA Cell Re-pointing Procedure – Overview




qualcomm hsdpa

Technology