OFDMA downlink resource allocation via ARQ feedback

In OFDMA downlink resource allocation, the base station exploits knowledge of the users' channel realizations in order to opportunistically assign users to appropriate subchannels, as well as to optimize the rates and powers across those subchannels. Because reverse-link bandwidth is scarce, the base station's channel knowledge must be obtained via some form of limited feedback. While the typical assumption for this feedback is that it comes in the form of heavily quantized SNR estimates computed at the user terminals, we propose to use ACK/NAK feedback that is already provided by higher-layer ARQ. Towards this aim, we propose a greedy resource allocation scheme, based on distributional (rather than point) estimates of SNR. We also show how these SNR distributions can be updated recursively for Markov time-varying channels.