Resource allocation problems in MIMO-OFDM systems

This thesis investigates resource allocation problems in MIMO-OFDM wireless communication systems. First, a stochastic adaptive multilevel waterfilling algorithm for optimal power allocation in a MIMO-OFDM system using imperfect channel estimates is presented. The algorithm has a self-learning capability that allows it to adapt to non-stationary fading channels. Analysis and simulations reveal the algorithm's excellent tracking ability in a slowly time-varying channel. Next, optimization of the ARITH-SINR linear cost function that takes into account power allocation issues, cross-layer issues, and the peak-to-average power ratio (PAPR) in a MIMO-OFDM system is presented. We show that the fairness constraints on the powers result in a polymatroid. As a result, optimal power allocation is performed via a Greedy Algorithm that attains the solution in O(nlogn) computations, where n is the number of fairness power constraints. Next, the problem is extended to consider antenna constraints that take into account the PAPR at each antenna. A computationally efficient Greedy-Revised Dual Simplex Algorithm for optimal power allocation is devised. The worst case complexity of this algorithm is O(n²) per iteration. Finally in numerical studies we show the user power allocations and spectral efficiency of the algorithms.Applied Science, Faculty ofElectrical and Computer Engineering, Department ofGraduat