OFDMA Constrained Resource Allocation with imperfect Channel Knowledge

Abstract

Adaptive resource allocation has been shown to provide substantial performance gain in OFDMA based wireless systems when perfect channel state information (CSI) is available at the transmitter. However, perfect CSI at the transmitter (CSIT) is rarely possible due to channel estimation errors and channel feedback delay. In this paper, we consider practical resource allocation strategies for downlink multiuser OFDMA systems, where the base station has imperfect CSI. We address the problem of subchannel assignment and power allocation, to maximize the ergodic sum utility under QoS, fairness and total power constraints. This constrained optimization problem is formulated as a non-convex finite dimensional stochastic problem. We use Lagrange dual decomposition framework to derive an algorithm that efficiently finds the optimal allocation. Simulation results show the impact of the channel estimation error on the performance of the system and illustrate the trade-off between maximized utility and the constraints.