The University of British Columbia

We present a cross-layer optimization problem where the coding rate of hybrid ARQ systems is adapted with channel, buffer, and input traffic state to minimize packet errors as well as buffer delay. Representing both the incoming traffic and the time-varying wireless channel as a finite state Markov chain, it is shown that the problem forms a partially observable Markov decision process (POMDP) problem. Since finding optimal policy is PSPACE complete, we investigate two policy-heuristic approaches for the purpose of efficient and real time solving of our formulated POMDP problem. Numerical results reveal that the performances of these two heuristic are almost same as the case when channel states are fully observable.