instructional workshop on wireless networks : physical layer aspects

IEEE Symp./ IISc -2001 IIT Madras 1

OFDM Physical Layer -- Fundamentals, Standards, &

Advances

K. Giridhar

Associate Professor of Electrical EngineeringTelecom and Computer Networks (TeNeT) Group

IIT Madras, Chennai 600036http://www.tenet.res.in

Instructional Workshop on Wireless Networks : Physical Layer AspectsDRDO-IISc Program on Mathematical Engineering, Feb. 14, 2003


Contents

Wireless Propagation -- Overview OFDM Fundamentals Comparing TDMA, CDMA, and OFDM OFDM Standards Case Study: IEEE 802.11a OFDM WLAN Key Advances in Wireless Technology Space-Time Processing for OFDM Summary


Basics of Radio Propagation

Distance

Powe

r

10-100 m(1-10 secs)

0.1 -1 m(10-100 msecs)

Exponential

Long-term Fading

Short-term Fading


Multi-path Propagation

r(t) = 0 s(t-0) + 1 s(t-1) + 2 s(t-2) + 3 s(t-3)


Multi-path Propagation -- contd.r(t) = 0 s(t-0) + 1 s(t-1) + 2 s(t-2) + 3 s(t-3)

channelInput (Tx signal)

Output(Rx signal)

ImpulseResponse h(t)

3 - 0

time3

0

freq.

Frequency Response H(f)


Frequency Selective Fading

Time2.0 secs 2.5 secs 3.0 secs

Delay Spreadrms =

5secs

Gain

(in vo

lts)

Fading

Frequency Selective Fading Channels can provide-- time diversity (can be exploited in DS-CDMA)-- frequency diversity (can be exploited in OFDM)


Contents



TDMA, CDMA, and OFDM Wireless Systems

Time Division Multiple Access (TDMA) is the most prevalent wireless access system to date GSM, ANSI-136, EDGE, DECT, PHS, Tetra

Direct Sequence Code Division Multiple Access (DS-CDMA) became commercial only in the mid 90’s IS-95 (A,B, HDR,1x,3x,...), cdma-2000 (3GPP2), W-CDMA (3GPP)

Orthogonal Frequency Division Multiplexing (OFDM) is perhaps the least well known can be viewed as a spectrally efficient FDMA technique IEEE 802.11A, .11G, HiperLAN, IEEE 802.16 OFDM/OFDMA

options


TDMA (with FDMA) Principle

Power

Time

Freq.

Time-slots

Carriers


Direct Sequence CDMA Principle(with FDMA)

Power

Time

Freq.

User CodeWaveforms


OFDM (with TDMA & FDMA) Principle

Power

Time

Freq.

Time-slots

Carriers

Tones


Other Multiple Access Techniques Multi-Carrier TDMA

DECT, PACS Frequency Hopped Spread Spectrum

Bluetooth CSMA/CA

IEEE 802.11 (1 or 2 Mbps standard) DS-CDMA with Time Slotting

3GPP W-CDMA TDD (Time Division Duplex)

Packet Switched Air Interface is vital for high bit-ratesand high capacity (for data users) -- GPRS, DPRS, etc.


What is an OFDM System ? Data is transmitted in parallel on multiple

carriers that overlap in frequency


FEC IFFT

DAC

LinearPA

add cyclic extension

bits

fc

OFDM symbol

Pulse shaper &

view this as a time tofrequency mapper

Generic OFDM Transmitter

Complexity (cost) is transferred back from the digital to the analog domain!

Serial toParallel


Add

Cyclic

Prefix

Serial/

Parallel

]0,[ns

]1,[ns

],[ Nns

Parallel/

SerialIFFT

]0,[nd

]1,[nd

],[ Nnd

OFDM Transmitter -- contd.

S/P acts as Time/Frequency mapper IFFT generates the required Time domain waveform

Cyclic Prefix acts like guard interval and makes equalization easy (FFT-cyclic convolution vs channel-linear convolution)

1

0

2],[1],[

N

k

Nkij

eknsN

ind


OFDM Receiver

Cyclic Prefix is discarded

1

0

2],['1],[

N

i

Nikj

eindN

knr

FFT

]0,[nr

]1,[nr

],[ Nnr

Parallel/

Serial

Serial/

Parallel

Remove

Cyclic

Prefix

]0,[' nd]1,[' nd

],[' Nnd

FFT generates the required Frequency Domain signal

P/S acts like a Frequency/Time Mapper


AGC

fc

VCO

Sampler FFTError

gross offset

Slot &

fine offsetFreq. OffsetEstimation

TimingSync.

(of all tones sent in one OFDM symbol)

Generic OFDM Receiver

RecoveryP/S and

Detection


OFDM Basics To maintain orthogonality

where = sub-carrier spacing = symbol duration

If N-point IDFT (or FFT) is used Total bandwidth (in Hz) =

= symbol duration after CP addition

fTs

1

fsT

fNW

CPS TT


Condition for Orthogonality

Time

T

Base frequency = 1/T

T= symbol period


OFDM Basics -- contd. If the Cyclic Prefix > Max. Delay Spread,

then the received signal after FFT, at the nth tone for the kth OFDM block can be expressed as

where is additive noise is channel frequency response

],[],[],[],[ knwknsknHknr

],[ knw],[ knH


Tx Waveform over a OFDM Symbol(magnitude values, for 802.11a)


Sync Basis Functions(of equal height for single-ray channel)

Shape gets upset by(a) Fine Frequency Offset(b) Fading


OFDM -- PHY layer tasks

Signals sent thro’ wireless channels encounter one or more of the following distortions:

additive white noise frequency and phase offset timing offset, slip delay spread fading (with or without LoS component) co-channel interference non-linear distortion, impulse noise, etc

OFDM is well suited for high-bit rate applications


Frequency Offset Carrier recovery and tracking critical for OFDM

Offsets can be comparable to sub-carrier spacing in OFDM Non-coherent detectors possible with differential coding

Residual freq. offset causes constellation rotation in TDMA loss of correlation strength over integration window in CDMA

(thereby admitting more CCI or noise) increased inter-channel interference (ICI) in OFDM

OFDM can easily compensate for gross freq. offsets (offsets which are an integral multiple of sub-carrier width)


Timing Synchronisation Timing recovery (at symbol level) is easily achieved in OFDM

systems Can easily overcome distortions from delay spread

Can employ non-coherent timing recovery techniques by introducing self-similarity => very robust to uncompensated frequency offsets

If cyclic prefix is larger than the rms delay spread, range of (equally good) timing phases become available

=> robust to estimation errors


Slot and Timing Synchronization in OFDMExample: 4 tones per slot (OFDM symbol) T

self-symmetry can be exploited for non-coherent timing recovery

zero tones

IFFT PA

T secst

IFFT PA

T secst

T/2 T

Traffic Slot

Preamble/Control Slot


Effect of Delay Spread

Typical rms delay spread in macro-cells Urban : 1-4 secs, Sub-urban : 3-6 secs Rural (plain, open country) : 3-10 secs Hilly terrain : 5-15 secs

TDMA requires equalization (even if rms delay spread is only 20-30% of symbol duration)

higher bit-rates would imply more Inter-Symbol Interference (ISI)

therefore, equalization complexity increases with bit rate


Effect of Delay Spread -- contd. 1

Effect of delay spread on DS-CDMA is multi-fold On the Uplink, the time diversity inherent in the delay

spread can be used to mitigate fading On the Downlink, multipath delay spread upsets

channelization (short) code orthogonality

Sectorisation vital in CDMA to reduce CCI on the Uplink

However, sectorisation reduces delay spread as well, thereby reducing the RAKE performance


Effect of Delay Spread in OFDM

Delay spread easily compensated in OFDM using : Cyclic Prefix (CP) which is longer than the delay spread Thereby, converting linear convolution (with multipath

channel) to effectively a circular convolution enables simple one-tap equalisation at the tone level

However, the frequency selectiveness could lead to certain tones having very poor SNR=> poor gross error rate performance

Data Payload CP

3.2secs 0.8secs

Example: IEEE 802.11 A (and also in HiperLAN)


Delay Spread Compensation in OFDM

Two basic ideas to combat freq. selectivity in OFDM

Feed-forward only techniques Temporal FEC and interleaving Transmit diversity and space-time coding

Feed-back based techniques (similar to approaches used in Multi-Carrier Modulation in the ADSL modems)

Water-pouring (bit-loading) Pre-equalisation or pre-distortion

Sectorisation in macro-cell OFDM can help reduce delay spread


AGC

Sampler DFTError

-- Gross Freq. Offset-- Channel Estimation and Equalization

OFDM Receiver Algorithms -- Recap

RecoveryP/S and

Detection

Freq.

-- Fine Freq. Offset-- Timing Estimation


Conventional OFDM

Frequency Domain Equalisation-- Conventional OFDM

DFTFrequency

DomainEqualiser

RemoveCP

RxAlgos.

Detection& P/S

IDFT AddCP

TxMod.

SymbolMapping

& S/P


Tx -- low-complexity, TDMARx -- implements SC-FDE; Linear Equaliser or DFE

Frequency Domain Equalisation-- Single Carrier FDE (SC-FDE)

DFTFrequency

DomainEqualiser

RemoveCP

RxAlgos. DetectorIDFT

AddCP

(of symbols)Tx

Mod.SymbolMapping

to permit FDE


TDE + FDE for OFDM

Time & Frequency Domain Equalisation-- for OFDM in large delay spread channels

DFTFrequency

DomainEqualiser

RemoveCP

RxAlgos.

Detection& P/S

IDFT AddCP

TxMod.

SymbolMapping

& S/P

Time-Domain

Equaliser


Fading and Antenna Diversity Short-term fading exhibits spatial correlation

Two antennas, spaced /4 meters or greater apart, fade independently

Spatial diversity combining can mitigate fading Switch diversity (least complex, modest improvement) Selection diversity Equal gain combining Maximal ratio combining (most complex, optimal)

TDMA, CDMA, and OFDM systems will invariably require antenna diversity to overcome fading


Fading and Channel Estimation Use of midamble in GSM and EDGE to avoid channel

tracking within the slot duration Unlike in TDMA and OFDM, fading affects not only

signal quality, but also system capacity in DS-CDMA Fast closed-loop power control required which can

track short-term fading For RAKE combining, multipath delays and gains are

required to be estimated and tracked By using orthogonal signaling, IS-95 uplink does not

need gain estimation, but requires delay estimation In OFDM systems, the long symbol duration makes

channel estimation and tracking very important


Channel Estimation in OFDM -- Example

Traffic slots may contain a few equally spaced tones for phase correction (due to residual freq. offset, phase noise, fading)

Control slot may also contain MAC messages

Frame (say, 4 slots)Control +

Training Slot Traffic Slot 1 Traffic Slot 3Traffic Slot 2

PhaseCorrectionTones

TrainingTones (for channelidentification)

MAC message(broadcast)

Control +Training Slot


Fading Compensation in OFDM OFDM using a FDE, observes only “flat” fading at the

sub-carrier level

Fading manifests as ICI terms in the Frequency Domain

In OFDM Phy Layer, two basic ways to reduce ICI Reduce OFDM symbol duration (increase sub-carrier width)

802.16 has FFT sizes ranging from 256 to 4096

Transmit pulse shaping can reduce ICI (by providing excess “time-width”)


Other PHY Issues in OFDM High peak-to-average ratio of the signal envelope

Linear Power Amp., with 5-8dB back-off required (costly)

To support mobility (fast fading) it will require More training tones per symbol and also in every slot Tx diversity and/or ST coding support Exploit time, frequency, and space diversity /

processing


Phy Layer Issues in Macro-cell OFDM Macrocells will require larger cyclic extensions / prefix

Microcells may not be economical during initial deployment GPS locked base stations required

To control ACI from neighbor BS sites (at cell edge) CCI can be estimated / controlled only if it is tone-aligned

Strict power control required may be required on uplink To minimize cross-talk between tones of different users

sharing the same OFDM symbol (time slot) To avoid uplink power control

allocate only one user per uplink slot or, make uplink a pure TDMA (not OFDM)


Phy Layer Issues in OFDMA Strict power control required required on uplink

(OFDMA) To minimize cross-talk between tones of different

users sharing the same OFDM symbol (time slot) To avoid uplink power control

allocate only one user per uplink slot (OFDM) or, make uplink a pure TDMA (single-carrier)


MAC Layer Issues in Macro-Cell OFDM Many proprietary broad-band FWA based on OFDM are

configured as primarily data networks providing Bridging functionality (Ethernet packets on air) Routing functionality (IP packets on air)

Some of the key issues then are How many modes (scheduling options) should MAC

support? How is voice and other streaming data to be handled?

Indeed, mixing of voice and data not good for statistical multiplexing CDMA example – the new cdma2000 / HDR standard, where

distinct voice-only and data-only base stations are proposed


Contents



DS-CDMA versus OFDM

channelInput (Tx signal)

Output(Rx signal)

ImpulseResponse h(t)

time3

0

freq.

Frequency Response H(f)

DS-CDMA can exploit

time-diversity

OFDM can exploitfreq. diversity


Comparing Complexity of TDMA, DS-CDMA, & OFDM Transceivers

Timing Sync.

Freq. Sync.

Timing Tracking

Freq. Tracking

ChannelEqualisation

Analog Front-end(AGC, PA, VCO, etc)

TDMA OFDMVery elegant, requiring

no extra overhead

CDMAEasy, but requires

overhead (sync.) bitsDifficult, and requiressync. channel (code)

Easy, but requiresoverhead (sync.) bits More difficult than TDMA Gross Sync. Easy

Fine Sync. is Difficult

Modest Complexity Usually not requiredwithin a burst/packet

Requires CPE Tones(additional overhead)

RAKE Combining in CDMA usually more complex than

equalisation in TDMA

Modest Complexity(using dedicated correlator)

Easy, decision-directedtechniques can be used

Frequency DomainEqualisation is very easy

Complexity or cost is very high (PA back-off

is necessary)Very simple

(especially for CPM signals)

Complexity is high inAsynchronous W-

CDMA

Modest to High Complexity(depending on bit-rate and

extent of delay-spread)

Fairly Complex(power control loop)


Comparing Performance of TDMA, DS-CDMA, & OFDM Transceivers

Fade Margin(for mobile apps.)

Range

Re-use & Capacity

FEC Requirements

Variable Bit-rateSupport

Spectral Efficiency

TDMA OFDM

Required for mobileapplications

CDMA

Required for mobileapplications

Modest requirement(RAKE gain vs power-

control problems)

Range increase by reducing allowed noise rise (capacity)

Difficult to support large cells (PA , AGC limitations)

Modest (in TDMA) andHigh in MC-TDMA

Re-use planning iscrucial here

FEC is vital even forfixed wireless access

FEC is usually inherent (to increase code decorrelation)FEC optional for voice

Powerful methodsto support VBR

(for fixed access)

Very High(& Higher Peak Bit-rates)Modest

Modest

Low to modest support

Poor to Low

Very elegant methodsto support VBR & VAD

Very easy to increasecell sizes


Contents



Proprietary OFDM Flavours

Wideband-OFDM(W-OFDM) of Wi-LAN

www.wi-lan.com

Flash OFDMfrom Flarion

www.flarion.com

Vector OFDM(V-OFDM) of Cisco, Iospan,etc.

www.iospan.com

Wireless Access (Macro-cellular)

-- 2.4 GHz band-- 30-45Mbps in 40MHz-- large tone-width (for mobility, overlay)

-- Freq. Hopping for CCI reduction, reuse-- 1.25 to 5.0MHz BW -- mobility support

-- MIMO Technology-- non-LoS coverage, mainly for fixed access-- upto 20 Mbps in MMDS

Wi-LAN leads the OFDM Forum -- many proposals submitted to IEEE 802.16 Wireless MANCisco leads the Broadand Wireless Internet Forum (BWIF)


OFDM based Standards

Wireless LAN standards using OFDM are HiperLAN-2 in Europe IEEE 802.11a, .11g

OFDM based Broadband Access Standards are getting defined for MAN and WAN applications

802.16 Working Group of IEEE 802.16 -- single carrier, 10-66GHz band 802.16a, b -- 2-11GHz, MAN standard


Key Parameters of 802.16a Wireless MAN

• Operates in 2-11 GHz• SC-mode, OFDM, OFDMA, and Mesh support• Bandwidth can be either 1.25/ 2.5/ 5/ 10/ 20 MHz• FFT size is 256 = (192 data carriers+ 8 pilots +56 Nulls) • RS+Convolutional coding

• Block Turbo coding (optional)• Convolutional Turbo coding(optional)

• QPSK, 16QAM, 64QAM• Two different preambles for UL and DL


Preamble structure for 802.16a Wireless MAN

TbTg

CP 128 128

Preamble structure of 802.16a Uplink

Two different preamble structures for DL and UL

TgTg Tb Tb

CP 64 64 64 64 CP 128 128

Preamble structure of 802.16a Downlink


Calculations for 802.16a -- Example: 5MHz

Carrier frequency 2-11 GHz Channel Bandwidth 5 MHz Number of inputs to IFFT/FFT 256 Number of data subcarriers 192 Number of pilots 8 Subcarrier frequency spacing f 19.53125 KHz (5 MHz/256) Period of IFFT/FFT Tb 51.2 s (1 / f)Length of guard interval 12.8 s (Tb / 4)Length of the preamble for Downlink 128 s (640 sub-carriers)Length of the preamble for Uplink 76.8s (384/5 MHz)Guard interval for Uplink preamble 25.6 s (128/5 MHz)OFDM symbol duration 64 s (320/5 MHZ)


Hiperaccess(PMP, 25Mbps, 40GHz)

orETSI’s FWA (2-11 GHz)

Broadband Wireless Standards ETSI BRAN activity

HiperLan > HiperLink > HiperAccess

HiperLan (1,2)(19 or 54Mbps, 5GHz)

Hiperlink(155Mbps, 17GHz

upto 150m)

2-5 miles, LoS(> 11GHz) or non-LoS (<11GHz)


IEEE 802.16(10 to 66 GHz)

Broadband Access Standards -- contd. IEEE LAN and MAN standards

IEEE 802.11a or.11b, or .11g

IEEE 802.16a,b(2 to 11 GHz)

2-5 miles, LoS(> 11GHz)

1-3 miles, non-LoS


Contents



IEEE 802.11a Overview Carrier frequency= 5 GHz Total allotted bandwidth= 20 MHz x 10 =

200MHz Size of the FFT= 64 Number of data subcarriers= 48 Number of Pilot subcarriers= 4 FFT period= 3.2 µs Channel bandwidth used= 64/3.2 µs => 20

MHz


Rate Dependent ParametersCoded bits

persubcarrier

(NBPSC)

Coded bitsper OFDM

symbol(NCBPS)

Data bitsper OFDM

symbol(NDBPS)

Data rate(Mbits/

s)

ModulationCoding rate

(R)

6

9

12

18

24

36

48

54

BPSK

BPSK

QPSK

QPSK

16 QAM

16 QAM

64 QAM

64 QAM

1/2

1/2

3/4

3/4

1/2

3/4

2/3

1

1

2

2

4

4

6

6

288

48

96

96

192

192

48

288

24

36

48

72

96

144

192

2163/4


802.11A -- Frame and Slot Structure

Details of the preamble field

10 short symbols (0.8*10 = 8s) 2 long symbols (1.6 + 2*3.2 = 8s)

0 1 2 3 4 5 6 7 8 9 GI 2 T1 T2

Freq. Offset estimation and channel estimationSignal detect, AGC, Timing

Recovery, Freq. acquisition

Number of Sub-carriers = 64 (only 48+4=52 are non-zero)

P1 P2 MACHeader

Data Data ……. Data Preamble2

Data …

8 s 8 s 4 s 4 s


PPDU Frame format


Preamble Structure -- Implications

0 1 2 3 4 5 6 7 8 9

Only every 4th tone is non-zero. Thisimplies 10 replicas (in time) within 4+4 = 8secs

Even if delay spread in 0.2 secs (for a 100m cell), we can use 9 of 10 replicas to recover timing; use less than 9 for higher fade rates


Auto-correlation and Piece-wise Cross-correlation for Slot Boundary Detection

79k

0k

* 159to0nfor16) k(nk)yy(n z(n) ||

Auto-correlation for timing and freq. estimation

Piece-wise Cross-correlation can also be used


Timing Recovery in 802.11A --Simulation Results

N=0 represents start of 1st preamble; length of channel impulse response set to 8 samples (0.4secs)

Probability of the corresponding n being detected as the startof the frame at different SNRs

Value of index n inthe transmitteddata s(n) 5 db 10 db 15 db 20 db No noise

n<7 (outside theacceptable range)

0.062 0.008 0 0 0

N=7 0.032 0.009 0.002 0 0 N=8 0.057 0.048 0.022 0.013 0.013 N=9 0.096 0.091 0.081 0.080 0.083 N=10 0.144 0.195 0.226 0.236 0.231 N=11 0.204 0.276 0.322 0.327 0.313 N=12 0.148 0.216 0.205 0.208 0.228 N=13 0.118 0.109 0.113 0.106 0.103 N=14 0.070 0.036 0.027 0.027 0.026 N=15 0.033 0.036 0.002 0.003 0.003 N=16 0.019 0.008 0 0 0 n>16 (outside theacceptable range

0.017 0.003 0 0 0

Performance of timing recovery algorithm using 1st preamble

AcceptableRange


Auto-correlation Result

autocorrelation result

0

0.2

0.4

0.6

0.8

1

1.2

1 11 21 31 41 51 61 71 81 91 101 111 121 131 141 151 161


Piece-wise cross-correlation Result

Cross correlation Result

0

0.5

1

1.5

2

2.5

3

3.5

4

1

10 19 28 37 46 55 64 73 82 91

100

109

118

127

136

145

154


• Quantity of interest is the Standard Deviation, f of the frequency estimate.

• It is given by: f = [E (( fest - fo )2 )] 1/2

Fine Frequency Offset Estimation

Approximate by using ensembleaveraging of many Monte-Carlo runs


MMSE Technique5 10 15 20 25

10-3

10-2

10-1

snr(db)

S.D

300 Hz

30 Hz

5 10 15 20 2510-3

10-2

10-1

snr(db)

S.D

30 Hz

300 Hz

Self-Correlation

Comparison of the Two Fine Frequency Estimation Algorithms


64-QAM Without Pilot De-rotation64 QAM before pilot correction

-2

-1.5

-1

-0.5

0

0.5

1

1.5

2

-2 -1 0 1 2


64-QAM After Pilot De-rotation

64 QAM after pilot rotation

-1.5

-1

-0.5

0

0.5

1

1.5

-1.5 -1 -0.5 0 0.5 1 1.5


BER Curves for Different Channel Models For AWGN Channel

AWGN case

-5

-4

-3

-2

-1

00 5 10 15

Eb/n0 in db

BE

Rin

db QPSK1/2

12Mbps16QAM 1/224Mbps64QAM2/348MBPSBPSK1/26Mbps


Contents



Motivation for Advances Increase Erlang Capacity (Re-use

Efficiency) – more users per square area Increase Range and/or Reliability Increase Channel Capacity (Spectral

Efficiency) -- higher average bit rate or lower Tx power

Increase Coverage -- must for fixed wireless Support for asymmetric and bursty traffic

-- high peak to average bit rate traffic like Internet

Support for mobility, inter-operability etc.


Wireless Advances -- contd.

Transmit Diversity

Smart Antennas

Sectorisation

CCI Suppression

Freq. Hopping

Multi-user Detection

Power Control

VAD, AMR, VBRReceive Diversity

Fixed Beamforming

Transmit Diversity

Spatial Multiplexing

Space-Time CodingLink

Adaptation

Re-use Re-use EfficiencyEfficiency

RangeRange

Spectral Spectral EfficiencyEfficiency

DCS

Turbo Coding OFDM


ST Block Code Example

-d*(k+1), d(k)

d*(k), d(k+1)

TxRxr(k+1), r(k)

Recall Example – Permutation Tx Diversity Scheme

Alamouti and other Tx diversity / coding schemes are suitable only for frequency-flat channels

OFDM converts frequency selective channel to parallel flat channels (one for every sub-carrier)


Contents



MIMO OFDM In addition to time and space, OFDM

systems can exploit frequency diversity

If feedback channels are available, Space-Time-Frequency “water pouring” possible!

OFDM can convert delay-spread diversity into space diversity (diversity conversion!)


Permutation Tx Diversity for OFDM

Courtesy:http://www.research.att.com/~justin/


ST Coded Tx Diversity for OFDM

Courtesy:http://www.research.att.com/~justin/


Contents



Why OFDM for Broadband Access? Why not CDMA ?

DS-CDMA cannot support high bit rates efficiently Advantages of OFDM

Fundamentally, well suited for high bit rate applications Simple frequency domain equalisation

lower complexity than RAKE or TDMA equalization Timing recovery is very straight forward Timing jitter easier to handle (due to long symbol duration) Good support for highly variable bit rate applications

Coarse granularity from time-slots(1 time-slot=1 OFDM symbol)

Fine granularity from tones (blocks) inside a time-slot


Summary -- contd. 1 OFDM is emerging as popular solution for wireless

LAN, and also for fixed broad-band access

The questions that remain to be answered are Will OFDM be good when there is vehicular mobility?

Pulse-shaping or large tone-widths reduce throughput

What about macro-cellular, non-LoS coverage issues?

What about OFDM deployment in unlicensed bands?

Will OFDM be cost-effective? If not right now, when? Analog (linear PA) with dynamic PAR control


Summary -- contd. 2 Space-Time processing for OFDM is a very hot area of

current research

The cost-effectiveness of many of these space-time techniques is not clear at present

Multiple RF/IF chains versus faster base-band (MIPS) costs

Will 4G see a combination of OFDM, DS-CDMA & TDMA ?

Key Question is: Where are those high-bit rate, high usage applications ? -- at low cost ?

Thank You!

instructional workshop on wireless networks : physical layer aspects

Documents