Buffer and channel adaptive transmission over fading channels with imperfect channel state information

We consider the problem of buffer and channel adaptive transmission over fading channels. In the model, data packets arrive stochastically into a finite-length buffer which transmits them over a time-varying correlated fading channel. We assume that knowledge of the buffer occupancy and the fading state is available at the transmitter. The objective is to vary the transmission power and rate according to the buffer and channel conditions so that the long-term system throughput is maximized under some average transmission power constraint. Here, maximizing the system throughput is equivalent to minimizing packet loss due to both buffer overflow and transmission error. We formulate this optimization problem as a Markov decision process (MDP) and use dynamic programming techniques to obtain the solution. Next, we look at the effects of error and delay in the channel state information and propose an optimal adaptive policy that is obtained by formulating a partially observable Markov decision process (POMDP). Simulation results are given to show the performance of the policies over fading channels with perfect and imperfect channel state information.