Resource allocation under channel uncertainty

Scheduling policies in multi-queue multi-server systems need to allocate servers based on the channel state information and the queue state information. In the downlink, channel state information available to the scheduler may be imperfect due to feedback delay and estimation errors. Motivated by this, we consider the downlink scheduling problem of allocating servers to multi-queue multi-server systems under channel uncertainty. We propose new policies which allocate the servers based on the predicted channel information. Simulations indicate that our policies have better delay and backlog properties than a policy proposed by Kar, Luo & Sarkar [1].