Optimal operating point for MIMO multiple access channel with bursty traffic

Multiple antennas at the transmitters and receivers in a multiple access channel (MAC) can provide simultaneous diversity, spatial multiplexing, and space-division multiple access gains. The fundamental tradeoff in the asymptotically large SNR regime is shown by Tse et al. (2004). On the other hand, MAC scheduling can provide a statistical-multiplexing gain to improve the delay performance as shown by Bertsimas et al. (1998) and Stolyar and Ramanan (2001). In this paper, we formulate and analytically derive bounds on the optimal operating point for MIMO-MAC channel for bursty sources with delay constraints. Our system model brings together the four types of gains: diversity, spatial multiplexing, space-division multiple-access, and statistical-multiplexing gains. Our objective is to minimize the end-to-end performance as defined by the delay bound violation probability as well as the channel decoding error probability. We find the optimal diversity gain and rate region in which the system should operate. As an example, we illustrate our technique and the optimal operating point for the case of a compound Poisson source. In addition, we note an interesting interplay between the intensity of the traffic and resource pooling with regard to both multiple-access and statistical-multiplexing gains.