Markov models of statistical multiplexing of telephone dialogue with packet switching

Existing methods of analysis of voice transmission by packet switching were designed mainly with respect to a Poisson stream of input packets, for which the probability of an active packet on each input port of the router is a constant value in time. This assumption is not always valid, since the formation of speech packets during a dialogue is a non-stationary process, in which case mathematical modeling becomes an effective method of analysis, through which necessary estimates of a network node being designed for packet transmission of speech may be obtained. This paper presents the result of analysis of mathematical models of Markov chain based speech packet sources vis-à-vis the peculiarities of telephone dialogue models. The derived models can be employed in the design and development of methods of statistical multiplexing of packet switching network nodes.