Diffusion approximation for a heavily loaded multi-user wireless communication system with cooperation (Q1034821)

The paper considers a multiple-input multiple-output downlink system where data is buffered at the transmit end and the channel is assumed to be fixed for all transmissions over the period of interest. This communication system has a corresponding queueing system formulation where, even in the simple case of Poisson arrivals, independently for each user, it is not known how to minimize the average delay for a given load. Closed-form expressions for average delay are unavailable for many simple policies. However, when the ratio of the average arrival rates is specified in advance, the maximum possible throughput can be computed, and a simple policy can be shown to be throughput-optimal under Markovian assumptions. An exact expression for the performance of this policy is not available. In this paper, as a measure of performance, the authors prove a limit theorem justifying a diffusion approximation for the queueing system when heavily loaded and operating under this policy. Their limit theorem has general distributional assumptions on the arrivals and packet lengths. It does not require Markovian assumptions.