1. An Efficient Subcarrier and Power Allocation Algorithm for Dual-Service Provisioning in OFDMA Based WiBro Systems February 1, 2005 Mohmmad Anas, Kanghee Kim, Jee Hwan Ahn, and Kiseon Kim anas@gist.ac.kr Multimedia Communication Systems Lab. Dept. of Information and Communications, GIST ICOIN 2005 Jan 31 st – Feb 2 nd, 2005

2. Multimedia Communication Systems Lab. [2/34] Resource Allocation Issues  Resource Allocation in OFDMA Systems  Subcarrier Allocation Different subcarriers are allocated to different users according to subcarrier allocation strategy  Power Allocation Depending upon the user’s capacity requirement power is allocated to the user’s subcarriers [OFDMA: Orthogonal Frequency Division Multiple Access] Subcarrier Allocation Power Allocation Two set of flows Flow 1 Flow 2 Mix of flows Complete Channel State Information is known 1) How to allocate subcarriers for dual-service provisioning 2) How to allocate power for dual-service provisioning

3. Multimedia Communication Systems Lab. [3/34] Contributions  An efficient resource (sequential subcarrier and power) allocation algorithm is proposed for dual-service (GP and BE) provisioning in OFDMA based systems. GP: Guaranteed Performance; BE: Best Effort; EQ: EQual power allocation; PA: Power Allocation; RA: Resource Allocation Proposed-EQ Proposed-PA Two set of flows having different required QoS GP Users BE Users Mix of flows Complete Channel State Information is known Proposed-RA

4. Multimedia Communication Systems Lab. [4/34] Contents  I NTRODUCTION  O VERVIEW OF OFDMA S YSTEM  R ESOURCE A LLOCATION I N OFDMA S YSTEM  Subcarrier Allocation for Dual-Class Users  Power Allocation for Dual-Class Users  S IMULATION R ESULTS  C ONCLUDING R EMARKS

5. Multimedia Communication Systems Lab. [5/34]  Research Background  Motivations and Objective  I NTRODUCTION

6. Multimedia Communication Systems Lab. [6/34]  Broadband Wireless Applications  NextG Mobile Communication Systems Integration of Voice, Video, Image, Data etc. User-level QoS Provisioning  OFDMA is chosen as PHY-MAC of several NextG standards IEEE 802.16-2004, ETSI HiperLAN/2, DVB etc.  Resource Allocation in OFDMA Systems  Adaptive/Dynamic Resource Allocation is advantageous over Fixed Resource Allocation Wireless Channel is time-varying and frequency-selective  Previous works mainly focus on joint subcarrier and power allocation problem for system-level QoS provisioning Research Background

7. Multimedia Communication Systems Lab. [7/34] Motivations and Objective Objective  OFDMA is PHY-MAC standard for several NextG communication systems  Resource Allocation in multiuser-OFDM/OFDMA is mostly studied for system-level QoS provisioning  Joint subcarrier and power allocation for user-level QoS provisioning is NP-hard problem [ NP-hard: non-deterministic polynomial time hard ]  [Shen’03] Joint resource allocation problem can be decoupled into subcarrier allocation and power allocation problems with the reduction in complexity almost by half and can be solve using canonical methods Motivations  To present an efficient, decoupled subcarrier and power allocation algorithm, for user-level QoS provisioning [ QoS: data rate and BER ]

8. Multimedia Communication Systems Lab. [8/34]  OFDMA Block Diagram  System Model  Traffic Model  O verview of OFDMA S ystems

9. Multimedia Communication Systems Lab. [9/34] OFDMA Block Diagram Assumptions: - Base Station has perfect CSI - Base station informs the mobile for Resource Allocation - Subcarrier is not shared among users @ a certain time instant OFDMA Transceiver Subcarrier Selector Subcarrier and Power/Bit Allocation Algorithm OFDMA Transceiver b1 b1 data Subcarrier and bit information Channel State Information (CSI), H k,n Subcarrier and bit allocation Base StationUser k Subcarrier and power information for mobile k Channel Estimator b2 b2 bK bK

10. Multimedia Communication Systems Lab. [10/34] System Model  Assumption  Each user experience independent fading, h k,n, with AWGN with mean zero and variance, σ 2 = N 0 (B/N)  Received SNR for the k th user’s n th subcarrier, where, H k,n : carrier-to-noise ratio p k,n : power allocated to the k th user’s n th subcarrier N 0 : noise power spectral density B: total bandwidth N: total number of subcarriers  [Goldsmith’97] for

11. Multimedia Communication Systems Lab. [11/34] Traffic Model  NextG broadband wireless communication expects mix of user applications i.e., voice, video, image, and data  Assumptions: Mixture of two kind of users differentiated on the basis of QoS (data rate and BER): Guaranteed Performance (GP) and Best Effort (BE) GP Users: require bounded BER, and throughput guarantee BE Users: require bounded BER, and NO throughput guarantee Total number of users in the system is K First K 1 (known) users are GP users Last ( K-K 1 ) users are BE users All GP users have same BER requirements ( ) All BE users have same rate and BER requirements ( )

12. Multimedia Communication Systems Lab. [12/34]  Joint Subcarrier and Power Allocation  Joint Problem Formulation  Decoupled Problem Formulation  Proposed-EQ  Proposed-RA Algorithm  R esource A llocation in OFDMA S ystem for Dual-Service Provisioning for Dual-Service Provisioning

13. Multimedia Communication Systems Lab. [13/34] Joint Subcarrier and Power Allocation (I)  Data rate of user k,  Problem Statement: Maximize the sum capacity for BE (Best Effort) users under the constraints of GP (Guaranteed Performance) users data rate requirements and total transmitted power constraints. where, Ω k : set of subcarrier allocated to k th user T: OFDMA symbol duration i.e.,

14. Multimedia Communication Systems Lab. [14/34] Joint Subcarrier and Power Allocation (II)  Joint Subcarrier and Power Allocation Problem Formulation: Maximization of sum capacity of BE users Rate of ANY one GP user Total Power Constraint Other GP users rate are limited by the proportionality equation  [Hoo’98] NP-hard problem  Decouple subcarrier and power allocation Subject to:

15. Multimedia Communication Systems Lab. [15/34] Decoupled Subcarrier and Power Allocation  Power Allocation Problem (for known subcarrier allocation) Formulation:  Subcarrier allocation to service combined GP and BE users Power is assumed to be equally distributed for subcarrier allocation Subject to:

16. Multimedia Communication Systems Lab. [16/34] Proposed-EQ Algorithm (I)  Based on suboptimal subchannel allocation in [Rhee’00]  Proposed-EQ: Subcarrier allocation to combined GP and BE users… We assume power is equally distributed among subcarriers as in [Rhee’00] 1.Initialization (enforce zero initial conditions) (a) set R k =0,  k =  for k = 1, 2,…, K and A = {1, 2,…, N} (b) p=P total /N  EQual power is allocated among subcarriers 2.For k = 1 to K 1 (allocate best subcarrier for each GP user) (a) find n satisfying | H k,n |  | H k,j | for all j  A (b) let  k =  k  {n}, A = A – {n} (c) (d) while A ≠ , repeat Step 2 until the Rate requirements for GP users are fulfilled

17. Multimedia Communication Systems Lab. [17/34] Proposed-EQ Algorithm (II) 3.For k = K 1 +1 to K (allocate left subcarriers to BE users) (a) find n satisfying | H k,n |  | H k,j | for all j  A (b) let  k =  k  {n}, A = A – {n} (c) 4. While A ≠  (iteratively give lowest rate BE user first choice) (a) find k satisfying R k ≤ R i for all i, K 1 +1 ≤ i ≤ K (b) for the found k, find n satisfying | H k,n |  | H k,j | for all j  A (c) for the found k and n, let  k =  k  {n}, A = A – {n} (d)  Omitting Step 2.(d) and Step 3 in proposed-EQ reduces it to method in [Rhee’00]

18. Multimedia Communication Systems Lab. [18/34] Proposed-RA Algorithm (I)  Proposed-RA: Resource allocation to provide service to combined GP and BE users  Proposed-EQ is used as subcarrier allocation algorithm  An efficient power allocation algorithm for dual-service (combined GP and BE users) provisioning is proposed: Proposed-PA

19. Multimedia Communication Systems Lab. [19/34] Proposed-RA Algorithm (II)

20. Multimedia Communication Systems Lab. [20/34]  Resource Allocation in OFDMA System  Simulation Parameters  Optimal Power Allocation  Proposed-RA vs. Proposed-EQ  S imulation R esults

21. Multimedia Communication Systems Lab. [21/34] Simulation Parameters System OFDMA System, Number of Subcarriers, N = 64 Number of Users, K = [4, 8, 12, 16] Channel 6-tap exponentially decaying power profile with Rayleigh fading Max. delay spread 5  s Doppler frequency30 Hz GP UsersBE Users Number of UsersFirst 50%Last 50% Required BER10 -5 10 -3 Required data rate1 bps/HzNot applicable ExampleVoice, VideoInternet Data

22. Multimedia Communication Systems Lab. [22/34] Proposed-EQ vs. Proposed-RA (I) (a) (b) BE UsersGP Users BE UsersGP Users BE UsersGP Users BE UsersGP Users Figure 3. Capacity vs. user index. (a) K = 4; (b) K = 8 Proposed-EQ Proposed-RA Observations: We notice that GP users adapt to their data rates (1 bps/Hz) after proposed-RA algorithm. While the BE user’s capacity performance is same for proposed-EQ and proposed-RA.

23. Multimedia Communication Systems Lab. [23/34] Proposed-EQ vs. Proposed-RA (II) (c)(d) BE UsersGP Users BE UsersGP Users BE UsersGP Users BE UsersGP Users Figure 4. Capacity vs. user index. (a) K = 12; (b) K = 16 Proposed-EQ Proposed-RA Observations: We notice that GP user’s QoS requirements (1 bps/Hz) are satisfied after proposed-RA algorithm.

24. Multimedia Communication Systems Lab. [24/34] Proposed-EQ vs. Proposed-RA (III) Figure 5. min-user’s capacity of GP and BE users vs. number of users Observations: min-user’s capacity of GP users (A, B) remain constant while those of BE users (C, D) decreases with the increase in number of users. Adaptive resource allocation (C, D) performs better than static resource allocation i.e., TDMA (E). A: B: C: D: E: C, D A, B E

25. Multimedia Communication Systems Lab. [25/34]  Conclusions  C ONCLUDING R EMARKS

26. Multimedia Communication Systems Lab. [26/34] Conclusions  QoS aware subcarrier and power allocation for OFDMA based broadband wireless communication system is studied Proposed a subcarrier allocation algorithm to service dual-class users differentiated on the basis of data rate and BER: Proposed-EQ  The suboptimal subchannel allocation method in [Rhee’00] is the motivation for proposed-EQ algorithm Proposed an efficient power allocation algorithm for dual-class users (combined GP and BE) provisioning assuming subcarrier allocation determined by the proposed-EQ: Proposed-RA  Result shows that proposed-RA works well to provide guaranteed performance to GP users and to maximize the sum- capacity for BE users for given BER.

27. Multimedia Communication Systems Lab. [27/34] References [Wong‘99] C.Y. Wong, R.S. Cheng, K.B. Letaief, and R.D. Murch, “Multiuser OFDM with Adaptive Subcarrier, Bit, and Power Allocation”, IEEE JSAC, Vol. 17, No.10 (1999) [Rhee‘00] W. Rhee, and J.M. Cioffi, “Increase in Capacity of Multiuser OFDM system Using Dynamic Subchannel Allocation,” IEEE VTC-Spring (2000) [Yin‘00] H. Yin, and H. Liu, “An Efficient Multiuser Loading Algorithm for OFDM-based Broadband Wireless Systems,” IEEE GLOBECOM (2000) [Jang‘03] J. Jang, and K.B. Lee, “Transmit Power Adaptation for Multiuser OFDM Systems,” IEEE JSAC, Vol.21, No. 2 (2003) [Shen‘03] Z. Shen, J.G. Andrews, and B.L. Evans, “Optimal Power Allocation in Multiuser OFDM Systems,” IEEE GLOBECOM (2003) [Hoo‘98] L. Hoo, J. Tellado and J.M. Cioffi, “Dual QoS Loading Algorithm for DMT Systems Offering CBR and VBR Services,” IEEE GLOBECOM, Vol. 1 (1998) [Goldsmith‘97] A.J. Goldsmith and S.G. Chua, “Variable-Rate Variable-Power MQAM for Fading Channels,” IEEE Trans. On Communications, Vol. 45 (1997) [Shakkottai‘03] S. Shakkottai, T.S. Rappaport, and P.C. Karlsson, “Cross-layer Design for Wireless Networks,” IEEE Communications Magazine, vol. 41 (2003)

28. Multimedia Communication Systems Lab. [28/34] Thank you!