Analysis of a TDMA Network With Voice and Data Traffic

Abstract

An analysis of an integrated voice-data network with Demand Assignment Time Division Multiple Access (TDMA) is presented using the following model: (1) voice calls that cannot be serviced are blocked, whereas requests to transmit data messages are queued; (2) no traffic boundaries are assumed, i.e., any new traffic arrival may be assigned to any unassigned time slot; (3) message lengths are exponentially distributed with the mean voice message length assumed to be much larger than the mean data message length; (4) traffic requests are generated according to two independent Poisson processes; and (5) time slot assignments are made instantaneously and no priorities are assumed. Such a model applies to a single-channel TDMA network in which voice and data traffic arrivals are serviced on a first-come first-served basis. An approximate analysis, based upon physical insight, is presented that yields the blocking probability for voice messages, the mean number of queued data requests, and the mean value of the peaks of the data queue process. Comparisons with simulation results indicate that the analytical results are very accurate. Performance curves are presented and compared with analogous results for TDMA networks that handle only one traffic type.