APPROXIMATING NONRENEWAL PROCESSES BY MARKOV-CHAINS - USE OF SUPER-ERLANG (SE) CHAINS

We study a class of point processes generated by transitions in Markov chains. We are primarily concerned with approximating superposed phase renewal processes by these point processes. We identify a subclass of Markov chains that we call Super-Erlang chains. These chains have special properties that facilitate the development of approximations. We outline an approximation procedure and provide computational results that demonstrate the potential of the approach. The primary motivation for this study is the analysis of open queueing networks.