Proportional Fair Frequency-Domain Packet Scheduling for

3GPP LTE Uplink

Suk-Bok Lee, Ioannis Pefkianakis, Adam Meyerson, Shugong Xu, Songwu Lu

IEEE INFOCOM 2009 proceedings.

Speaker： Tsung-Yin Lee

2

Outline

Introduction The Model Problem Formulation

Heuristic Algorithm Simulation Conclusion

3

OFDMA for LTE

Orthogonal Frequency Division Multiple Access (OFDMA) has been considered as a strong candidate for the broadband air interface robustness to multipath fading higher spectral efficiency bandwidth scalability

4

Disadvantage of OFDMA

one major disadvantage of OFDMA is that the instantaneous transmitted RF power can vary dramatically within a single OFDM symbol high peak-to-average power ratio (PAPR)

(decrease battery life)

5

Single-Carrier FDMA

selected for LTE uplink multiple access scheme

keeping most of the advantages of OFDMA SC-FDMA has significantly lower PAPR

benefits the mobile terminal in terms of transmit power efficiency

6

LTE Uplink Scheduler

a scheduler needs to know the instantaneous radio channel conditions across all users and all resource blocks (RBs) LTE UL each user transmits a Sounding Refere

nce Signal (SRS) to the BS channel quality indicator (CQI)

7

Proportional Fair (PF) algorithm

PF algorithm as a basic scheduling principle and apply the PF algorithm directly over each RB one-by-one independently

SC-FDMA requires that all the RBs allocated to a single user must be contiguous in frequency within each time slot [5][6]

[5] Moray Rumney. 3GPP LTE: Introducing SIngle-Carrier FDMA Agilent MeasurementJournal, 2008.[6] 3GPP TSG-RAN WG2 Meeting #57, R2-070585, “Resource fragmentation in LTEuplink”, St. Louis, USA, Feb, 2007.

8

The Model

The base station can allocate m RBs to a set of n users At each time slot multiple RBs (with the contig

uity constraint) can be assigned to a single user indicate whether or not RB c is assigned to

user i at time slot t denote the instantaneous channel rate for u

ser i on RB c at time t

9

Problem Formulation (1/3)

the well known PF algorithm aims to maximize the logarithmic utility function

In order to maximize , one should maximize where di(t) is total data transmitted to user i at time t ( this paper change di(t) to ) [7][10][14]

[7] M. Andrews. A survey of scheduling theory in wireless data networks. IMA, 2005.[10] H. Kushner and P. Whiting. Asymptotic properties of proportional-fair sharingalgorithms. Allerton, 2002.[14] D. Tse. Multiuser diversity in wireless networks. http://www.eecs.berkeley.edu/ dtse/stanford416.ps , 2002.

10

Problem Formulation (2/3)

Let be the PF metric value that user i has on RB c at time slot t

We can establish PF objective function when scheduling time slot t as follows:

(1)

11

Problem Formulation (3/3)

for LTE UL we need to incorporate the contiguous RB constraint into this objective (1) due to the physical layer requirement of SC-FDMA serve users with suboptimal PF metric value fo

r some RBs so as to optimize the PF objective (1)

12

Hardness Result

Theorem 1 LTE UL PF-FDPS problem (i.e. maximizing ob

jective (1) with the contiguous RB constraint) is NP-hard [11]

[11] S.-B. Lee, I. Pefkianakis, A. Meyerson, S. Xu, and S. Lu. Proportional FairFrequency-Domain Packet Scheduling for 3GPP LTE Uplink. UCLA TR-090001, 2009.

13

Heuristic Algorithm

Paper’s heuristics do not give guaranteed error bound, and moreover we believe that no practical greedy algorithms can give an approximation to this particular problem [11]

[11] S.-B. Lee, I. Pefkianakis, A. Meyerson, S. Xu, and S. Lu. Proportional FairFrequency-Domain Packet Scheduling for 3GPP LTE Uplink. UCLA TR-090001, 2009.

14

carrier-by-carrier in turn (1/2)

As a starter, our first greedy heuristic Alg1 schedules data from RB1 to RBm in sequence, and for each RB c it assigns the best user i who 1) has the maximum PF metric value on c 2) satisfies the contiguity constraint.

15

carrier-by-carrier in turn (2/2)

16

largest-metric-value-RB-first (1/2)

Assign all the “in-between” RBs to a candidate user it assigns RB5 to i, which as a result comes wit

h assignment of RB4 to i, since i is already assigned RB3

RB3 RB4 RB5

RB3 Already Assigned RB5 Assigned NowAssigned RB

Between RB3 and RB5

17

largest-metric-value-RB-first (2/2)

18

riding peaks (1/3)

Seeing the drawback of Alg2, we would like to utilize each user’s high valued RBs as much as possible Fundamental physical layer characteristic is tha

t in multi-carrier systems the channel SNR values (i.e. CQI) are correlated in both time and frequency

19

riding peaks (2/3)

if for each user i RB c has good channel rate, then the neighboring RBs (c−1, c+1) have high channel rate as well with high probability

20

riding peaks (3/3)

21

RB grouping (1/2)

Alg3 relies on the strong frequency-domain correlation, it is easily cheated by the small-scale variation

22

RB grouping (2/2)

This RB grouping might be helpful to catch a bit large-scale fluctuation divide m RBs into n groups apply the “peak riding” over those RB groups

23

Simulation Parameter (1/2)

use traces generated as specified in 3GPP deployment evaluation [2]

[2] Technical specification group radio access networks - Deployment aspects. 3GPP TR 25.943

24

Simulation Parameter (2/2)

paper use an algorithm that optimizes objective (1) without the constraint as our reference, and we refer to this algorithm as OPT∗ (upper bound of the optimum)

Jain’s fairness index [9], measured by the data-rate fairness criterion

[9] R. Jain, D. M. Chiu, and W. Hawe. A Quantitative Measure of Fairness and Discrimination for Resource Allocation in Shared Systems. DEC Research Report TR-301.

25

System throughput and fairness with varying number of users

26

Average number of users scheduled per 1 TTI

27

Conclusion

Due to its single carrier property of SC-FDMA, LTE UL requires the RBs allocated to a single user to be contiguous in frequency NP-hard nature of this problem Four Heuristic Algorithm to solve this problem

28

Comment

In LTE scheduling problem, we will handle uplink and downlink from different scheme We should combine uplink and downlink to

increase network performance