RANDOM MULTIPLE-ACCESS OF BROADCAST CHANNELS WITH PARETO DISTRIBUTED PACKET INTERARRIVAL TIMES

We consider the random multiple access of a slotted broadcast communic ation channel. Packet arrivals for such channels are often modeled as Poisson processes because the latter have attractive theoretical prope rties and are well understood, even though a number of traffic studies have shown that packet interarrival times are not always exponentiall y distributed. For example, recent studies argue convincingly that tra ffic from a variety of working packet networks (LANs, WANs, etc.) is m uch better modeled using statistically self-similar processes to chara cterize the packet interarrival times of the aggregate traffic, which have much different properties than Poisson processes. Because of the great influence multiple access algorithms (used in the channel access subsystem) will have on the performance of third-generation digital w ireless mobile communication systems, we study in this article the pro blem of random multiple access under packet traffic generated by inter active data applications (e.g., reading/composing short e-mail message s, responding to paging type messages, transferring files between the mobile and a fixed computer, and querying of a database). We examine a nd demonstrate the performance robustness of the above RAAs when they are driven by a strongly bursty packet arrival process, commonly found in interactive data applications.