1. Performance evaluation of adaptive sub-carrier allocation scheme for OFDMA Thesis presentation16th Jan 2007 Author:Li Xiao Supervisor: Professor Riku Jäntti Instructor: Lic.Sc Boris Makarevitch Place: Communications Laboratory HELSINKI UNIVERSITY OF TECHNOLOGY Communications Laboratory

2. Agenda Introduction Overview of OFDM OFDM based multiple access schemes Adaptive sub-carrier allocation algorithm Simulation Conclusions HELSINKI UNIVERSITY OF TECHNOLOGY Communications Laboratory

3. Introduction Background Multi-carrier transmission methods attract much focus to support high speed and reliable wireless communications A good OFDMA sub-carrier allocation scheme should use spectral as efficiently as possible and achieve minimum cost of service based upon user’s QoS requirement Objectives Transmission power minimization as cost of service in Downlink and Uplink Performance evaluation of adaptive sub-carrier allocation for OFDMA Methodology Adaptive OFDMA sub-carrier allocation algorithm implementation in Matlab Performance comparison among adaptive OFDMA sub-carrier allocation scheme and other static schemes HELSINKI UNIVERSITY OF TECHNOLOGY Communications Laboratory

4. OFDM Dividing the total bandwidth into a number of sub-carriers OFDM realization Intersymbol interference Intercarrier interference Cyclic prefix HELSINKI UNIVERSITY OF TECHNOLOGY Communications Laboratory

5. OFDM Based Multiple Access Schemes AdvantageDisadvantage OFDM-TDMA Easiest implementation Simple resource allocation No intra-cell MAI Low processing requirement Power saving Low signaling overhead High latency Lowest flexibility High peak to average power ratio OFDM-CDMA Spectral efficiency Frequency diversity MAI and inter-cell interference resistance Highest flexibility Simple resource allocation Low signaling overhead Implementation complexity Requirement of power control Only coherent modulation possible Intra-cell interference High peak to average power ratio OFDMA Simple implementation Resource allocation flexibility Adaptation to channel characteristics (adaptive scheme) Better BER performance (adaptive scheme) Inter-cell interference Low spectral efficiency High peak to average power ratio Signaling overhead (adaptive scheme) HELSINKI UNIVERSITY OF TECHNOLOGY Communications Laboratory

6. OFDMA Each user transmits on a certain number of OFDM sub-carriers during all time slots Static sub-carriers assignment and dynamic sub-carriers assignment Multirate system Multiuser diversity Adaptive modulation (bit rate, transmission power, channel coding rate or scheme) HELSINKI UNIVERSITY OF TECHNOLOGY Communications Laboratory

7. Mobile WiMAX Extension of WiMAX for fixed access Scalable OFDMA High data rate Quality of Service Scalability Security Mobility ParametersValues System Channel Bandwidth (MHz) 1.2551020 Sampling Frequency (MHz)1.45.611.222.4 FFT Size12851210242048 Number of Sub-Channels281632 Sub-Carrier Frequency Spacing 10.94kHz Useful Symbol Time (Tb = 1/f) 91.4 us Guard Time (Tg = Tb/8)11.4 us OFDMA symbol Duration (Ts = Tb+Tg) 102.9 us Number of OFDMA Symbols in 5ms Frame 48 HELSINKI UNIVERSITY OF TECHNOLOGY Communications Laboratory

8. Adaptive sub-carrier allocation algorithm Adaptive means Number of sub-carriers each user needs is adaptive Sub-carriers allocation among users is adaptive Bit loading to sub-carriers is adaptive Adaptive modulation scheme for each sub-carrier Users’ QoS requirement Minimum Reserved Rate Bit Error Rate HELSINKI UNIVERSITY OF TECHNOLOGY Communications Laboratory

9. Downlink system structure of OFDMA BS has the perfect knowledge of instantaneous channel information for all users Bandwidth of each sub-carrier is smaller than channel coherence bandwidth Each sub-carrier can only be occupied by one user No free sub-carrier left HELSINKI UNIVERSITY OF TECHNOLOGY Communications Laboratory

10. Adaptive sub-carrier allocation algorithm Objective function Transmission power minimization Downlink: Minimize the interference from BS in question to the MSs in other cells Uplink: MS battery saving Constraints Bit rate (bit/symbol) BER requirement Three sub-algorithms Number of sub-carriers determination Sub-carriers allocation Bit loading HELSINKI UNIVERSITY OF TECHNOLOGY Communications Laboratory

11. Number of sub-carriers determination Inputs: Each user’s bit rate constraint and average channel gain for each user Output: Number of sub-carriers each user gets assigned Two types of sub-carriers: Minimum required sub-carrier and Extra sub-carrier Minimum required sub-carriers are to fulfill the user’s bit rate constraint in the case that maximum amount of bits will be transmitted in each sub-carrier Extra sub-carriers will share bits with minimum required sub-carriers so that the loaded bits in each sub-carrier can be reduced and with an adaptive modulation scheme transmission power to all user can decrease No free sub-carrier left HELSINKI UNIVERSITY OF TECHNOLOGY Communications Laboratory

12. Sub-carrier allocation Inputs: Channel State Information for each user and number of sub-carriers each user gets assigned Output: sub-carriers allocation Phase 1: Constructive initial allocation 1. List the sub-carriers for each user in descend order according to channel gain 2. Check sub-carriers user by user if the number of sub-carrier each user gets is achieved or the sub-carrier has already been assigned to some users 3. If both are NO, assign the sub-carrier to this user, otherwise skip this user to next user HELSINKI UNIVERSITY OF TECHNOLOGY Communications Laboratory

13. Sub-carrier allocation Phase 1 may achieve only a local minimum but not total minimum transmission power Phase 2: Iterative improvement For every iteration, swap a pair of sub- carriers allocated to two users such that the result power can be reduced further Power reduction factor is the cost function in order to select the pair of users and pair of sub-carriers which can reduce power most Iteration is over when the maximum possible power reduction is less than zero HELSINKI UNIVERSITY OF TECHNOLOGY Communications Laboratory

14. Bit loading Inputs: Sub-carriers allocation, channel gain and bit rate constraint Output: Bits loaded to achieve each user’s bit rate constraint Levin-Campello algorithm 1. Each time selecting the sub-carrier that requires the least additional power to add one more bit 2. Check if the maximum amount of bits loaded in this sub-carrier has already been achieved and if this user’s bit rate constraint has been fulfilled 3. If both are NO, loading one more bit to this sub- carrier, otherwise selecting the sub-carrier which requires second least additional power and repeat 2 HELSINKI UNIVERSITY OF TECHNOLOGY Communications Laboratory

15. Simulation Bandwidth (MHz)5 Sampling Frequency (MHz) 5.6 FFT size (NFFT)128 Number of users K2-10 Symbol time (us)25.81 Channel Sets200 HELSINKI UNIVERSITY OF TECHNOLOGY Communications Laboratory

16. Simulation results: Number of sub-carriers determination UserAllocated sub- carriers Minimum required sub- carriers Extra sub- carriers 119712 21376 31789 418810 5945 6954 724717 819811 HELSINKI UNIVERSITY OF TECHNOLOGY Communications Laboratory

17. Simulation results: Sub-carriers allocation HELSINKI UNIVERSITY OF TECHNOLOGY Communications Laboratory

18. Simulation results: Bit loading HELSINKI UNIVERSITY OF TECHNOLOGY Communications Laboratory

19. Simulation results: BER performance Minimum 11.33dB gain in SNR using Adaptive allocation OFDM without extra sub-carriers over OFDM Interleave-FDMA 11.84dB gain over OFDM-TDMA 14.35dB gain over OFDM-FDMA 6.91dB gain from extra sub-carrier presence compared with no extra sub-carrier case HELSINKI UNIVERSITY OF TECHNOLOGY Communications Laboratory

20. Simulation results: Convergence of algorithm HELSINKI UNIVERSITY OF TECHNOLOGY Communications Laboratory

21. Conclusions Adaptive sub-carriers allocation algorithm can enhance the BER performance compared with static schemes The use of extra sub-carriers can improve the BER performance and decrease the total transmission power further Speed of the algorithm (convergence speed) is fast to meet the real time application requirements The speed of algorithm is not affected by the number of users much which guaranttes it perform well in high load system BS could use algorithm to increase the total number of users that can be accommodated for a given power budget HELSINKI UNIVERSITY OF TECHNOLOGY Communications Laboratory

22. Future study Minimization of transmission power in Uplink Scalable OFDMA HELSINKI UNIVERSITY OF TECHNOLOGY Communications Laboratory Thank you!