Adaptive downlink OFDMA resource allocation

Optimizing OFDMA resource allocation with respect to communication performance requires solving a nonlinear mixed-integer programming problem. As a result, many researchers have fallen back on suboptimal heuristic algorithms. In a recent paper, we demonstrate that ergodic rate maximization is possible using a dual optimization framework that results in a practically optimal solution with complexity that is on the order of the number of subcarriers times the number of users. One of the primary disadvantages of considering ergodic rates is the assumption that the channel distribution information (CDI) is perfectly known at the transmitter. Therefore, this paper proposes an adaptive algorithm based on stochastic approximation methods that do not require knowledge of the CDI. This algorithm converges to the optimal solution with probability one, while for each OFDMA symbol, the complexity is on the order of the number of subcarriers times the number of users. There are no iterations in a given OFDMA symbol time; instead, the ldquoiterationsrdquo are actually performed across time (symbols). Simulation results based roughly on a third-generation partnership project, long-term evolution (3GPP-LTE) OFDMA system corroborate our claims.