Performance of low-feedback-rate, gradient-based OFDMA subcarrier allocation with partial channel information

We consider a downlink OFDMA subcarrier allocation problem where the allocation algorithms are gradient-based but only CSI for some subet of subcarriers from each user is available at the transmitter. We propose an iterative version of gradient-based OFDMA subcarrier allocation schedulers which performs better than the standard non-iterative gradient-based schedulers. In addition, we propose an adaptive, weighted M-Best feedback mechanism (called wM-Best) suitable for gradient-based schedulers than the standard M-Best feedback. Simulation results show improvement of system performance metrics, in term of average throughput and queue length (or equivalently average delay), of the iterative scheduler and also the wM-Best feedback.