Random Access with Multi-Packet Reception

The queuing performance of a finite-user slotted Aloha type random access protocol with multi-packet reception capability at the base station is considered. The system evolution under this protocol can be described by a multi-dimensional Markov chain, which is not amenable to analysis due to complex interactions between queues of different users. A novel "user-centric" modeling and analysis approach is proposed, which focuses on the evolution of the one-dimensional Markov chain associated with the queue of a typical user in the system, assuming that all other users are in "steady state". While the proposed approach is applicable quite generally, this paper addresses the special case of identical users with Bernoulli packet arrivals in order to obtain insightful analytical expressions for various performance metrics of interest (throughput, average delay etc.). Both static and fading wireless channels are considered, under appropriate assumptions on the underlying power control mechanisms. Sufficient stability conditions as well as asymptotic performance measures in the infinite power and infinite user regime are derived for both cases. The fundamental throughput versus power tradeoff inherent in the random access protocol is established. The theoretical results derived in the paper are validated via simulations.