Dynamic Resource Allocation for Efficient Wireless Packet

Dynamic ResourceAllocation for

Efficient WirelessPacket Data

Communcations

B. Srikrishna

Introduction

Problem

OFDM ResourceAllocation

Our Algorithm

Results

CDMA ResourceAllocation

Results

Summary

References

Dynamic Resource Allocation for EfficientWireless Packet Data Communcations

B. Srikrishna

Assistant ProfessorDepartment of Electrical EngineeringIndian Institute of Technology Madras

Joint work with:M. Chandrashekar

V. SandeepParimal Parag

March 17, 2006



Communcations

B. Srikrishna

Introduction

Problem


Our Algorithm

Results


Results

Summary

References

Broadband Wireless

Internet

Wireless Local Area Network (WLAN)

Wireless Cellular

I Demand for mobile wireless internet access

I Need support for multimedia data transfer



Communcations

B. Srikrishna

Introduction

Problem


Our Algorithm

Results


Results

Summary

References

Broadband Wireless: Challenges

t

t

I High data rates and limited/expensive spectrum⇒ need high spectral efficiency

I Shared resources and multiple access

I Multipath fading channel

I Bursty traffic characteristics



Communcations

B. Srikrishna

Introduction

Problem


Our Algorithm

Results


Results

Summary

References

Broadband Wireless: Some Proposals

I Code-Division Multiple Access (CDMA)-basedI 1xEV-DO (HDR)I HSDPAI 1xEV-DV

I Orthogonal Frequency Division Multiplexing(OFDM)-based:

I FlashOFDMI IEEE 802.16eI IEEE 802.20



Communcations

B. Srikrishna

Introduction

Problem


Our Algorithm

Results


Results

Summary

References

Key Techniques

User 1 User 2 User 3

Periodic

reallocation of

resources

I Adaptation to channel and traffic conditionsI Dynamic resource allocation

I Reallocation period of the order of a few milliseconds



Communcations

B. Srikrishna

Introduction

Problem


Our Algorithm

Results


Results

Summary

References

Downlink Resource Allocation Problem

Channel information

User 1

User 2

User K

TrafficBasestation

I Physical resources: power and bandwidth

I Maximize system throughput

I Total transmit power constraint

I Fairness or Quality of Service (QoS) constraints



Communcations

B. Srikrishna

Introduction

Problem


Our Algorithm

Results


Results

Summary

References

Maximizing Capacity

User 1

User 2

User K

Basestation

Select theuser with

best channel

Channel K

Channel 2

Channel 1

I All power and bandwidth resources to one user

I User with best achievable rate chosen:

i = arg maxk

Rk ,

where Rk is the rate that can be supported by user k.

I No fairness or QoS constraint



Communcations

B. Srikrishna

Introduction

Problem


Our Algorithm

Results


Results

Summary

References

Maximizing Capacity: Parallel Channels

User 1

User 2

User K

Basestation

For each

best channeluser with

Select the

parallelchannel

Parallel Channels to each user

I Bandwidth resources split to achieve parallel channelsI For each channel n, user with best channel conditions

chosen:in = arg max

kRk,n.

I Water-filling power allocationI No fairness or QoS constraint



Communcations

B. Srikrishna

Introduction

Problem


Our Algorithm

Results


Results

Summary

References

Fairness and Quality of Service (QoS)

I Various notions of fairness or QoS

I Round-Robin

I Proportional Fairness [Tse02]

i = arg maxk

Rk

Rk,av,

where Rk,av is the average rate that can be supportedby user k.

I Modified-Largest Weighted Delay First (M-LWDF)[Andrews00]

i = arg maxk

γkWkRk ,

where Wk is the Head-Of-Line (HOL) packet delay foruser k, and γk = Ck

Rk,av.



Communcations

B. Srikrishna

Introduction

Problem


Our Algorithm

Results


Results

Summary

References

Resource Allocation in OFDM

Power

User 1

User 2

User 3Subc

arri

ers

I Available resources:I SubcarriersI Transmit power

I Channel is frequency-selective ⇒ subcarriers notidentical.



Communcations

B. Srikrishna

Introduction

Problem


Our Algorithm

Results


Results

Summary

References

OFDM Resource Allocation Algorithms

I Channel Aware Only (CAO) SchedulingI Proportionally Fair (PF) subcarrier allocation [Rhee00]I PF subcarrier allocation + power optimization [Shen05]I Max utility subcarrier allocation + power optimization

[Song05]

I Channel Aware Queue Aware (CAQA) SchedulingI MLWDF for OFDM-TDMA [Andrews00]I MLWDF at subcarrier level [Parag05]

I Our WorkI Joint Subcarrier and Power Allocation (JSPA) approachI Optimize power allocation after each subcarrier is

allocated



Communcations

B. Srikrishna

Introduction

Problem


Our Algorithm

Results


Results

Summary

References

MLWDF for OFDM-TDMA

I All subcarriers allocated to a single user in each slotI Select user i as:

I i = arg maxk

γkWkRk

I Wk : Head-Of-Line (HOL) packet delay for user kI Rk : Rate achievable for user k (water-filling)

I γk =Ck

Rk,av

I Ck = − log δk

Dkto achieve P[delay > Dk ] < δk

I Throughput optimal single-user scheduling ruleI Maximum stability: achieves stable queues if any

algorithm can achieve it

I Single-user scheduling in each time slot not optimal



Communcations

B. Srikrishna

Introduction

Problem


Our Algorithm

Results


Results

Summary

References

Subcarrier-wise Allocation

I Approach 1: MLWDF at the subcarrier level [Parag05]

I For each subcarrier n:I in = arg max

kγkWkRk,n

I Wk : Head-Of-Line (HOL) packet delay for user kI Rk,n: Rate achievable for user k on subcarrier nI Power allocation needed to allocate subcarriersI Uniform/fixed power allocation assumption

I Approach 2: [Song04]I Mean packet waiting time instead of Head-Of-Line

(HOL) packet delayI Other utility functions based on mean packet waiting

time



Communcations

B. Srikrishna

Introduction

Problem


Our Algorithm

Results


Results

Summary

References

Joint Subcarrier and Power Allocation

power

Split power equally amongst subcarriers

Start

Checkif all subcarriers

are allocated

Allocate a subcarrier to a user

Start

Split power equally amongst subcarriers

Allocate all subcarriers to users

Update each user’squeue

No

Yes

End

Optimize powerallocation with

Update each user’squeue

End

Optimize power

allocation with

power



Communcations

B. Srikrishna

Introduction

Problem


Our Algorithm

Results


Results

Summary

References

Joint Subcarrier and Power Allocation (JSPA)

I Optimal JSPA too complexI Sub-optimal JSPA

I Power optimization after each subcarrier is allocatedleads to better allocation of the remaining suncarriers

I Power allocation to each user proportional to thenumber of subcarriers allocated

I HOL delay is estimated after each subcarrier is allocated

I Some practical constraints includedI Discrete-rate constraint: Integer bit M-QAM

constellationsI Extension to band-wise allocation: reduced

signaling/feedback



Communcations

B. Srikrishna

Introduction

Problem


Our Algorithm

Results


Results

Summary

References

Simulation Setup

I 128 subcarrier OFDM system

I 12 users, Bernoulli packet arrival, 100 slot buffer

I 6-tap multipath channel, average channel conditions aredifferent for each user

I QPSK to 64-QAM



Communcations

B. Srikrishna

Introduction

Problem


Our Algorithm

Results


Results

Summary

References

Results: Throughput vs. Arrival Rate

1 1.5 2 2.5 3 3.5 4 4.5 5 5.51

2

3

4

5

6

7

Arrival rate (Mbps)

Thro

ughp

ut (M

bps)

CAO+FPACAO+FPA+PAOCAO+JSPAMLWDFCAQA+FPACAQA+JSPAPtotal= 5 dBW

Homogenous rate users



Communcations

B. Srikrishna

Introduction

Problem


Our Algorithm

Results


Results

Summary

References

Results: Max. Arrival Rate vs. Transmit Power

4 4.5 5 5.5 6 6.5 7 7.5 82.5

3

3.5

4

4.5

5

5.5

6

6.5

7

7.5

Ptotal in dBW

Arriv

al ra

te (M

bps)

CAO+FPACAO+FPA+PAOCAO+JSPAMLWDFCAQA+FPACAQA+JSPAHomogenous rate users

I Max. arrival rate for less than 0.5% packets dropped



Communcations

B. Srikrishna

Introduction

Problem


Our Algorithm

Results


Results

Summary

References

Results: Delay Performance

5 10 15 20 25 30 35 40

10−1

100

101

delay (time slots)

P(de

lay>

x)CAO+FPACAO+FPA+PAOCAO+JSPAMLWDFCAQA+FPACAQA+JSPA

Ptotal = 8dBW Arrival rate = 3 MbpsHomogenous rate users

I Best and worst delay performance among users plotted



Communcations

B. Srikrishna

Introduction

Problem


Our Algorithm

Results


Results

Summary

References

Results: Band-wise Allocation

1 1.5 2 2.5 3 3.53

3.5

4

4.5

5

5.5

6

Delay spread (rms) µsec

Arriv

al ra

te (M

bps)

L = 1L = 2L = 4L = 8L = 16Ptotal = 5dBW

I L = Number of subcarriers in a sub-band



Communcations

B. Srikrishna

Introduction

Problem


Our Algorithm

Results


Results

Summary

References

Resource Allocation in CDMA

User 1

Power

code

sSp

read

ing

User 2

I Available resources:I Spreading codesI Transmit power

I For any given user, all spreading codes are similar (interms of channel conditions).



Communcations

B. Srikrishna

Introduction

Problem


Our Algorithm

Results


Results

Summary

References

Multiuser Scheduling

I Most algorithms are single-user scheduling algorithmsI Proportionally Fair [Tse02]I MLWDF [Andrews00]

I Recent results on multi-user scheduling algorithmsI Greedy and pairwise greedy allocation [Kumaran05]I Gradient-based scheduling [Agrawal04]



Communcations

B. Srikrishna

Introduction

Problem


Our Algorithm

Results


Results

Summary

References

Results: Maximum Supportable Traffic

0 1 2 3 4 5 6 7 8 90

0.1

0.2

0.3

0.4

Average Arrival Rate (Mbps)

Frac

tion

of p

acke

ts d

ropp

ed

MLWDFTWO USER



Communcations

B. Srikrishna

Introduction

Problem


Our Algorithm

Results


Results

Summary

References

Results: Delay Performance

0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4

10−2

10−1

100

Delay d (in sec)

Prob

[del

ay >

d]

MLWDFTWO−USER



Communcations

B. Srikrishna

Introduction

Problem


Our Algorithm

Results


Results

Summary

References

Summary

I Dynamic resource allocation is essential to achieve highspectral efficiency

I Adaptation based on both channel and trafficinformation

I Some new results for OFDM and CDMA systemsI Joint subcarrier and power allocation in OFDMI Multiuser scheduling in CDMA

I Several open problems:I OptimalityI Quantifying the signaling/feedback overhead



Communcations

B. Srikrishna

Introduction

Problem


Our Algorithm

Results


Results

Summary

References

References

E. F. Chaponniere, P. Black, J. M. Holtzman, andD. Tse, “Transmitter directed multiple receiver systemusing path diversity to equitably maximize throughput,”U. S. Patent No. 6449490, September 2002.

M. Andrews, K. Kumaran, K. Ramanan, A. Stolyar,R. Vijayakumar, and P. Whiting, “CDMA data QoSscheduling on the forward link with variable channelconditions,” Bell Labs Technical Memorandum, 2000.

W. Rhee and J. M. Cioffi, “Increase in capacity ofmultiuser OFDM system using dynamic subchannelallocation,” in Proceedings of the 51st IEEE VehicularTechnology Conference, vol. 2, Spring 2000, pp.1085–1089.

Z. Shen, J. G. Andrews, and B. L. Evans, “Adaptiveresource allocation in multiuser OFDM systems withproportional rate constraints,” IEEE Transactions onWireless Communications, vol. 4, no. 6, pp. 2726–2737,November 2005.



Communcations

B. Srikrishna

Introduction

Problem


Our Algorithm

Results


Results

Summary

References

References

G. Song and Y. Li, “Cross-layer optimization for OFDMwireless networks-Part II: Algorithm development,” IEEETransactions on Wireless Communications, vol. 4, no. 2,pp. 625–634, March 2005.

P. Parag, S. Bhashyam, and R. Aravind, “A subcarrierallocation algorithm for OFDMA using buffer andchannel state information,” in Proceedings of the 62nd

IEEE Vehicular Technology Conference, vol. 1,September 2005, pp. 622–625.

G. Song, Y. G. Li, J. L. J. Cimini, and H. Zheng, “Jointchannel-aware and queue-aware data scheduling inmultiple shared wireless channels,” in Proceedings of theIEEE Wireless Communications and NetworkingConference, vol. 3, March 2004, pp. 1939–1944.



Communcations

B. Srikrishna

Introduction

Problem


Our Algorithm

Results


Results

Summary

References

References

R. Agrawal, V. Subramanian, and R. Berry, “Jointscheduling and resource allocation in CDMA systems,”in 2nd Workshop on Modeling and Optimization inMobile, Ad Hoc, and Wireless Networks (WiOpt ’04),Cambridge, UK, March 2004.

K. Kumaran and H. Viswanathan, “Joint power andbandwidth allocation in downlink transmission,” IEEETransactions on Wireless Communications, vol. 4, no. 3,pp. 1008–1016, May 2005.

Documents

Dynamic Resource Allocation for Efficient Wireless Packet