Fast Power Allocation Algorithms for Adaptive MIMO Systems.
Thesis DisciplineElectrical Engineering
Degree GrantorUniversity of Canterbury
Degree NameMaster of Engineering
Recent research results have shown that the MIMO wireless communication architecture is a promising approach to achieve high bandwidth efficiencies. MIMO wireless channels can be simply defined as a link for which both the transmitting and receiving ends are equipped with multiple antenna elements. Adaptive modulation and power allocation could be used to further improve the performance of MIMO systems.
This thesis focuses on developing a fast and high performance power allocation algorithm. Three power allocation algorithms are proposed in this thesis and their performances are compared in various system sizes and transceiver architectures. Among the three algorithms proposed in this thesis, the fast algorithm may be considered as the best power allocation algorithm since the performance of the fast algorithm is almost as good as the fullsearch (optimal)algorithm and the mean processing time is considerably less than the fullsearch algorithm. The fast algorithm achieves about 97.6% agreement with the optimal throughput on average. In addition, the time taken to find the power scaling factors using the fullsearch algorithm is about 2300 times longer than the processing time of the fast algorithm in a 6 x 6 system when the SNR is 20dB.
As an extension to the power allocation process, excess power allocation methods are introduced. Excess power is the unused power during the power allocation process. The power allocation algorithm allocates power to each received SNR to maximize the throughput of the system whereas the excesspower allocation distributes the excess power to each SNR to improve both the instantaneous and temporal behavior of the system. Five different excess power allocation methods are proposed in this thesis. These methods were simulated in the Rayleigh fading channel with different Doppler frequencies, fD = 10Hz,50Hz and 100Hz, where the ACF of the channel coefficients are given by the Jakes' model. The equal BER improvement method showed a slightly better performance than the other methods. The equal BER improvement method enables the system to maintain the power scaling factors without sacrificing QoS for 19.6 ms on average when the maximum Doppler shift is 10Hz.