Holger Claussen

Department Head, Autonomous Networks & Systems ResearchBell Labs

1. April 2012

Future Cellular Networks

1. Traffic Growth in Wireless Networks

2. The Future of Small Cell Networks

3. Future Macrocellular Networks

AGENDA

4. Industry Trends

5. Conclusions

Traffic growth for communication networks

10-1

100

101

102

103

Traffic (Tb/s)

Wireless Voice

P2P

Wireless data

grows fastest

2010 2015 202010

-2

10

YearData from: RHK, McKinsey-JPMorgan, AT&T, MINTS, Arbor, ALU, and

Bell Labs Analysis: Linear regression on log(traffic growth rate) versus log(time) with Bayesian learning to compute uncertainity.

Transition to LTE will increase capacity trough more spectrum, better scheduling, and MIMO, but these gains are not sufficient.

Energy consumption of networks will become an increasing problem. “More of the same” will not be good enough.

How can we address these address these problems?

Historic Capacity Gains in Wireless Networks

Wireless Network Capacity Gains 1950-2000

15x by using more spectrum (3 GHz vs 150 Mhz)

5x from better voice coding5x from better voice coding

5x from better MAC and modulation methods

2700x from smaller cells

Total gain 1 million fold

Source: William Webb, Ofcom.

Small Cells - A Necessary Topology Evolution for Future Data Growth

Moving to hierarchical cell structures with small cells can:

• Significantly increase the capacity in the same bandwidth

• Significantly reduce the energy consumption of networks

Small Cell types and their target segments

Home Femto

Enterprise Femto

Metro Indoor

Metro Outdoor

How will small cell deployments deployments evolve in the future?

Separate carrier for femtocells

• private access

• public access

Co-channel operation on a single shared

carrier

• private access NOT FEASIBLE due to high

Frequency deployment options for femtocells today and in the future

Today

Future

Increasing spectral efficiency per area

• private access NOT FEASIBLE due to high interference – coverage holes exist around femtocells with restricted access if no alternative carrier is available

• public access, potential HO issues for fast moving users

Co-channel operation with one shared & one

clean macrocell carrier

• private access – requires one clean macrocell carrier to serve UEs that are in range of femtocells with restricted access

• public access

Technical feasibility of co-channel operation

1

Adding co-channel

femtocells has only very

little impact on the

macro cell throughput

Adding more femtocells

does not affect their

throughput significantly

Results

• Co-channel deployment of femtocells in a macrocellular network is possible without significant impact on the macrocell performance.

• This allows efficient spatial frequency re-use.

• Femtocell throughputs indoors are very high since the wall separation to interference sources results in a high SINR. 64-QAM support recommended.

• Power self-optimization for both DL and UL of the

0 2 4 6 8 10 12 14 16 18 200

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

← 64-QAM

rate 1/2

← 16-QAM

rate 1/2

← 4-QAM

rate 1/2

Downlink throughput 3dB from the capacity limit [Mbit/s]

CDF

macro cell, N = 0

macro cell, N = 10

macro cell, N = 100

femto cell, N = 10

femto cell, N = 100

The femtocells are capable of high data rates when

64-QAM or higher modulation is used

Minimum throughput

is around 1Mbps

Example: Downlink results with

shadow fading, bandwidth = 3.84 MHz

• Power self-optimization for both DL and UL of the femtocell is necessary to ensure a low impact on the macrocellular network and to achieve a consistent cell range independent from the distance to the macrocell.

• For co-channel operation with only one available carrier, public access for femtocells is required.

References

[1] H. Claussen, “Performance of macro- and co-channel femtocells in a hierarchical cell structure," in Proc. 18th IEEE International Symposium on Personal, Indoor and Mobile Radio Communications (PIMRC), Athens, Greece, Sept. 2007.

[2] H. Claussen, “Co-channel operation of macro- and femtocells in a hierarchical cell structure," International Journal of Wireless Information Networks, vol. 15, no. 3, pp. 137-147, Dec. 2008.

[3] H. Claussen, L. T. W. Ho, and L. G. Samuel, “An overview of the femtocell concept," Bell Labs Technical Journal, vol. 13, no. 1, pp. 221-245, May 2008.

Results with Ocelot Coverage EngineScenario Overview for coverage analysis

Simulated test scenario Predicted transmission for various standard wall

types versus cosine of angle of incidence

Realistic house floor plans used for the simulation

Results with Ocelot Coverage EngineCoverage Analysis Results

Covera

ge p

robability

Covera

ge p

robability

Private Access model Public Access model

Co-channel operation of femtocells and macrocells is possible without causing coverage problems when public access is allowed or one clean carrier is available

Distance from femto [m] Distance from femto [m] Results

• Co-channel operation with a private access results in coverage holes for un-registered users.

• With a public access model or when a clean carrier is available the coverage probability around co-channel femtocells is very high.

References

[1] J. D. Hobby and H. Claussen, “Deployment options for femtocells and their impact on existing macrocellular networks,” Bell Labs Technical Journal, vol. 13, no. 4, pp. 145-160, Feb. 2009.

Impact of femtocells on the network energy consumption

• Telecommunications is a large consumer of energy (e.g. Telecom Italia uses 1% of Italy’s total energy consumption, NTT uses 0.7% of Japan’s total energy consumption)

• Increasing costs of energy and international focus on climate change issues have resulted in high interest in improving the efficiency in the telecommunications industry

Opportunity:Small cells have the potential to reduce the transmit power required for serving a user by a factor in the order of 103

compared to macrocells.

Problem: Most femtocells today are not serving users but are still consuming power:

50 Millon femtos x 12W = 600 MW 5.2 TWh/a

Comparison: - Nuclear Reactor Sizewell B, Suffolk, UK: 1195MW- Annual UK energy production: ~400 TWh/a

Source: BBC News - How the world is changing

Reducing energy consumptionIdle mode procedures for femtocellsWhen femtocells become more widely deployed, their energy consumption becomes a concern.

Idle mode procedures can:

• Significantly reduce energy consumption

• Reduce power density in the home

• Reduce mobility procedures and associated signalling

• Reduce interference caused by pilot transmissions

Further work is required for street level deployments

References:

Femtocell activation based on noise rise from active UE

allows to activate the femto only for serving a call

Femtocell energy consumption - Today Femtocell energy consumption – Optimized design

References:

[1] I. Ashraf, L. T. W. Ho, and H. Claussen, “Improving energy efficiency of femtocell base stations via user activity detection," in Proc. IEEE Wireless Communications and Networking Conference (WCNC), Sydney, Australia, Apr. 2010.

[2] H. Claussen, I. Ashraf, and L. T. W. Ho, “Dynamic idle mode procedures for femtocells," Bell Labs Technical Journal, to be published in 2010.

Measurements for noise rise controlled idle modes - Residential house

Measureme

nt p

oint

(1) UMTS Vodafone

PN = −−−−90dBm

(2) UMTS Three

PN = −−−−90dBm

(3) GSMO2

PN = −−−−81dBm

(4) GSMVodafone

PN = −−−−81dBm

Signal Analyzer

Ref 0 dBm Att 0 dB*

A

Center 1.923 GHz Span 5 MHz500 kHz/

3DB

RBW 100 kHz

VBW 300 kHz

SWT 2.5 ms

1 PK

MAXH

-100

-90

-80

-70

-60

-50

-40

-30

-20

-10

0

Noise rise during call to macrocell

Measurement equipment and example result

Antenna

Measureme

nt p

oint S1

Signal P1

[dBm]S2

Signal P2

[dBm]S3

Signal P3

[dBm]S4

Signal P4

[dBm]

1 6/7 −−−−79 6/6 −−−−61 5/5 −−−−22 4/5 −−−−25

2 6/7 −−−−61 6/6 −−−−52 3/5 −−−−18 3/5 −−−−15

3 6/7 −−−−67 5/6 −−−−57 3/5 −−−−28 3/5 −−−−22

4 6/7 −−−−74 6/6 −−−−70 3/5 −−−−30 3/5 −−−−31

5 6/7 −−−−67 6/6 −−−−65 3/5 −−−−33 3/5 −−−−34

6 6/7 −−−−82 6/6 −−−−75 4/5 −−−−28 4/5 −−−−27

7 6/7 −−−−79 6/6 −−−−68 3/5 −−−−32 3/5 −−−−40

8 6/7 −−−−69 4/6 −−−−57 4/5 −−−−29 4/5 −−−−26

Result: All calls for both GSM and UMTS are easily detectable

P1...P4 are the measured noise powers during a call of the test mobile to the macrocell.

S1...S4 is the signal strength indicator displayed by the mobile.

Future Small Cell DeploymentsEnterprise Femtocells / Street level deployments

Click on this video to start

The femtocell concept will be extended to support enterprise applications & street level deployments

This requires several changes:

• Support for more active users

• Higher power

• Different self-optimization algorithms

• Connects via Ethernet

Example: Distributed Coverage Optimization

Algorithm uses local measurements as inputs and adjusts coverage to balance the needs of following objectives:

• minimise coverage holes

• balance load

• minimise overlap and leakage

References

[1] I. Ashraf, H. Claussen, L.. T. W. Ho, “Distributed Radio Coverage Optimization in Enterprise Femtocell Networks”, in Proc. International Conference on Communications (ICC), 2010.

[2] L. T. W. Ho, I. Ashraf, and H. Claussen, “Evolving femtocell coverage optimization algorithms using genetic programming," in Proc. 20th IEEE International Symposium on Personal, Indoor and Mobile Radio Communications (PIMRC),Tokyo, Japan, Sept. 2009.

Coverage

Future Small Cell DeploymentsCost-effective deployment of large number of cells

Concept: Residential Picocells

• Slightly increase coverage of Femtocells to a cell radius of around 60m (Residential Picocells).

• Increase support to 8 users.

• Use small cell deployed by the user to supplement macrocell coverage

• Allow public access for users of the same operator.

IP Internet

PSTN

GGSN

SGSN

MSCOperator IP Network

operator.

• Use the user’s internet connectionas backhaul.

• This results in no costs for the cell deployment, the site, electricity, and backhaul for the operator.

Questions

What is the financial impact?

What is the impact on the total energy

consumption of the network?

RNC

Macro-cellNode B Residential

Picocell

Traditional UMTS Architecture Small Cell Architecture

Residential Picocell Controller/Gateway

Residential Picocells can significantly reduce the total annual network costs Results

• Macro-cellular networks become less economically viable with an increasing demand of high data rate services due to high operational expenses

• This problem can be addressed by user-deployed publicly accessible residential picocells

• A large fraction of the user demand can be covered by installing home base stations in only a small fraction of the customer’s homes

• Residential picocell deployment in combination with a macro- 0 20 40 60 80 1000

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

fraction of installed HBS

fraction of user coverage

40% market share30% market share

20% market share

10% market share

• Residential picocell deployment in combination with a macro-cellular network for area coverage can potentially reduce the annual network costs by 30% to 70% (2007 technology) [1].

fraction of installed HBS

0 20 40 60 80 1000

5

10

15x 10

6

percentage of customers with Home-BSAnnual network cost [$]

16 users/macrocell

32 users/macrocell

64 users/macrocell

128 users/macrocell

256 users/macrocell

Significant potential for

network cost reduction

References

[1] H. Claussen, L. T. W. Ho, and L. G. Samuel, “Financial analysis of a pico-cellular home network deployment," in Proc. IEEE International Conference on Communications (ICC), Glasgow, UK, June 2007, pp. 5604-5609.

Residential Picocells enable significant improvements in energy efficiency Results

• A mixed macro- and residential picocell architecture can significantly reduce the energy consumption of cellular networks for high data rate user demand in urban areas where macrocells are capacity limited

• The power consumption can be reduced by up to 60% for high data rate demand in urban areas (2007 technology) [1],[2].

• With more dynamic idle mode control and efficient power scaling with load a 46x efficiency improvement is possible

4

5

6

7

8

9

10x 10

4

Total Power [W

]

Macrocells Only

PIFm=15%, PIF

s=15%, γ

m=0.4, γ

s=0.4

PIFm=30%, PIF

s=30%, γ

m=0.2, γ

s=0.3

PIFm=50%, PIF

s=50%, γ

m=0.2, γ

s=0.05

46x scaling with load a 46x efficiency improvement is possible in 2016.

• Operators with high market share benefit more from the advantages since high small cell coverage is achieved with a lower fraction of customers with small cells.

References:

[1] H. Claussen, L. T. W. Ho, and F. Pivit, “Leveraging advances in mobile broadband technology to improve environmental sustainability," Telecommunications Journal of Australia, vol. 59, no. 1, pp. 4.1-4.18, Feb. 2009.

[2] H. Claussen, L. T. W. Ho, and F. Pivit, “Effects of joint macrocell and residential picocell deployment on the network energy efficiency," in Proc. 19th IEEE International Symposium on Personal, Indoor and Mobile Radio Communications (PIMRC), Cannes, France, Sept. 2008.

2010 2011 2012 2013 2014 2015 20160

1

2

3

Year

Total Power [W

]

46x

improvement

for 2016

What is the future forfuture forMacrocells?

lightRadio Cube and Active Antenna Array

Picocell –Macrocell solutions

Enables intelligent antenna

techniques

lightRadio Cube

Active Antenna Array

Digitallink

or Cloud

techniques

Avoids Cable losses (3 dB)

RF

Ethernet or CPRI

Circular Antenna Array Compact higher-order sectorization strategies

Beam 2

Beam3

Beam 1

.

.

.

24 column Array. Diameter: ~21” @2.45 GHz

Bell Labs prototype

Each beam/sector is generated

using 7 adjacent antenna elements.

Performance comparable with MU-MIMO

Large Scale Antenna SystemsReducing Energy ConsumptionOnly one antenna panel is powered to simulate a call to an end-user. All powered but only at a fraction

End-user

Power used = 16W

© 2011 GreenTouch Consortium

End-user

Power used = 1W

Proof of Concept Demonstration, London 2011.Collaborators: Bell Labs, Freescale, Huawei, imec, Samsung

What are other trendsfor Network for Network Evolution?

• Many applications are moving to computing clouds to reduce costs and allow dynamic scalability of apps.

• Networks have to support this trend and this inturn will generate new business for our customers.

• Traditional telecom functions will move to computing clouds. This requires developing combined networking and computing solutions.

Industry TrendsNetwork Virtualization and Cloud Computing

Current open problems (not exhaustive list):

• Missing dependability/real-time performance prevents us from running telco functions in the cloud.

• Missing optimization of cloud infrastructure to meet multiple SLAs of more demanding networked cloud apps (e.g. virtual telco)

• Security: missing confidentiality of data and code is a major showstopper for many commercial cloud applications.

Industry TrendsEvolution towards fully Autonomous Networks

Planning and optimisation is

performed manually through

network planning engineers

and drive testing.

Algorithms for auto-

configuration and self-

optimisation are developed

manually, and used to

automate the network

configuration and optimisation.

Algorithms are generated

using automated means,

(genetic programming).

Design process is done offline,

and resulting algorithm is

implemented in the network.

The algorithm generation process is

now distributed and performed locally

and continuously by the network nodes

themselves. Nodes now have ability to

autonomously specialise their individual

behaviour according to environment.

Summary & Conclusions

• Small cells are a necessary topology evolution for future data growth.

• Reserving carriers for femtocells will not be acceptable in the future since this restricts the macrocellular capacity too much.

• Energy efficiency becomes critical when small cells become widely deployed

• The femtocell concept will be extended to enterprise and outdoor applications

• A promising direction for the future evolution of small cells is to change their

objective from providing coverage in the home to supplementing macrocellular objective from providing coverage in the home to supplementing macrocellular

coverage.

• Light Radio enables many interesting radio concepts for Pico- to Macrocell

applications, and significantly reduces the infrastructure footprint.

• Many telco functions can move to computing clouds. Security and confidentiality are key for commercial success.

• Future networks will become more intelligent and highly autonomous