Capacity of Markov channels with receiver CSI and delayed feedback

While feedback does not increase the capacity of memoryless channels, the capacity of time-varying channels with feedback can be increased by exploiting the structure in the channel variations. Feedback information from the receiver is usually available at the transmitter only after some time delay. The capacity increase due to feedback depends on the feedback delay relative to the channel decorrelation time. We model time-varying channels as finite-state Markov channels and determine their capacity as a function of the feedback delay assuming perfect channel state information at the receiver. We apply the result to derive power control strategies to maximize the capacity for finite-state additive Gaussian noise channels and log-normal shadow fading channels