Performance analysis of an intrgrated voice/data system based on markov renewal process modeling

Abstract

In this dissertation work, a new modeling technique based on the Markov renewal process is proposed for the analysis of a class of integrated voice/data multiplexers. This is a generalization of the conventional two-dimensional Markov process model with the continuous time parameter. The conventional two-dimensional Markov process model has rather narrow application areas because of the exponential assumption on the data message length. On the other hand, the proposed model does not impose any restriction on the data message length distribution. Thus, with this model, it is possible to analyze a class of voice/data multiplexers which previously could not be dealt with. The service capacity available to data is approximated to be fixed during the service time of a data message. This approximation is possible since the service time of a data message is typically much smaller than the talkspurt/silence durations. As a direct application of the proposed model, a voice/data multiplexer with the non-gated service discipline is analyzed under the condition that the data message length is arbitrarily distributed. Numerical examples show that our model yields almost the same results as the conventional two-dimensional Markov process model, when the data message length is exponentially distributed. For the case of fixed length data messages, the effect of different service disciplines is shown by comparing the results of this work with those obtained by the discrete time parameter analysis. Simulation is done to confirm the correctness of the numerical results. As another application of the proposed model, an integrated voice/data multiplexer with an ARQ scheme is analyzed. In this multiplexer, a kind of ARQ scheme is applied to the data messages to guarantee error-free transmission of data messages. Previously, performances of the ARQ schemes were analyzed with data traffic only. One can find the significance of the proposed analysis technique more clearly with this app...