Multiaccess fading channels - Part II: Delay-limited capacities

In multiaccess wireless systems, dynamic allocation of resources such as tr ansmit power, bandwidths, and rates is an important means to deal with the time-varying nature of the environment. In this two-part paper, we consider the problem of optimal resource allocation from an information-theoretic p oint of view. We focus on the multiaccess fading channel with Gaussian nois e, and define two notions of capacity depending on whether the traffic is d elay-sensitive or not, In Part I, we have analyzed the throughput capacity region which characterizes the long-term achievable rates through the time- varying channel. However, the delay experienced depends on how fast the cha nnel varies. In the present paper, Part II, we introduce a notion of delay- limited capacity which is the maximum rate achievable with delay independen t of how slow the fading is. We characterize the delay-limited capacity reg ion of the multiaccess fading channel and the associated optimal resource a llocation schemes. We show that successive decoding is optimal, and the opt imal decoding order and power allocation can be found explicitly as a funct ion of the fading states; this is a consequence of an underlying polymatroi d structure that we exploit.