Power Allocation Techniques for Non Orthogonal Multiple Access Cellular Systems and Device to Device Groups

Abstract

The first part of the thesis proposes algorithms with realizable computational complexity newlinefor the optimal power distribution in the downlink NOMA cellular networks with three newlinedifferent objectives. Sum Rate (SR) and user fairness belong to the most important performance newlinemetrics of NOMA. Particularly, the metricWeighted Sum Rate (WSR) corresponds newlineto a balance between the total SR and user fairness, and it can reflect the priorities of different newlineusers. The thesis proposes low complexity power allocation algorithms for the newlinemaximization of the WSR within a sub-band and across multiple sub-bands. Since the newlineii newlinemetric WSR does not explicitly specify whether the Base Station (BS) benefits from the newlinepower allocation, alternate models based on the game theory are used in order to facilitate a newlinewin-win situation for both the BS and the users. However, the optimal selection of users is newlinenot addressed in the literature, and the existing power allocation solutions are sub-optimal. newlineThe thesis proposes a novel algorithm for the combined user selection and optimal power newlineallocation for the game, with the maximization of the BS revenue as the objective. Further, newlinesince NOMA operates by the principle of SIC at the receivers, the errors that occur during newlinethe SIC process affect the overall performance. However, many studies in the literature newlineassume that the SIC is error-free. Power optimization problems are generally solved assuming newlinethe theoretical Shannon rate. This assumption is not valid for a practical wireless newlinecommunication system employing an M-QAM, M-ary or similar modulation schemes. In newlinethis regard, the thesis analyzes the Bit Error Rate (BER) performance of the users for newlinea downlink NOMA network employing Quadrature Phase Shift Keying modulation and newlineproposes an algorithm for the power allocation with the minimization of the maximum newlineBER of the Near-User and Far-User as the objective. Extensive simulations reveal that the newlineproposed algorithms attain the optimal power allocation values that are obtained by the newlinecomputationally complex