Throughput Regions of Multiple-Access Fading Channels with Markov Arrivals and QoS Constraints

In this paper, throughput regions of multiple-access fading channels are characterized when multiple users, experiencing random data arrivals, transmit to a common receiver under statistical quality of service (QoS) constraints. Random arrivals are modeled as discrete Markov or Markov fluid processes. Different multiple-access transmission strategies, namely time-division multiple access and superposition coding with fixed or variable decoding orders, are considered. For these arrival and transmission models, throughput in terms of maximum average arrival rates is formulated in the presence of buffer constraints, employing the notions of effective bandwidth of time-varying sources and effective capacity of time-varying transmissions. Throughput regions are computed for the two-user case and the impact of source burstiness, buffer constraints, signal-to-noise ratio, and different communication strategies is investigated.