Optimal power and rate control for fading channels

Analysis of fading channels is usually made under the framework of either information theory or network theory. Information theory is usually concerned with the total throughput of the channel (Shannon capacity) while network theory considers also other performance measures such as average delay and buffer overflow probability. In a contemporary communication system operating in a fading environment, the performance measure is usually a combination of delay and throughput measures under various power constraints. Optimizing both delay and throughput is somewhat contradictory since achieving maximal throughput depends only on the fading channel while minimal average delay depends also on the arrival process of data to the transmitter. We define a performance criterion of both transmission power and average delay. Dynamic programming is employed to find the optimal power/rate control for this criterion as well as the system performance. Numerical results are presented as well as asymptotic analytic results for large buffers.