Analysis of batch Bernoulli process subject to discrete-time renewal generated binomial catastrophes (Q2178311)

Motivated by packet based transmission systems the authors develop a discrete time counterpart of models which are developed already in continuous time: Counting processes with batch arrivals which are subject to catastrophes, where the occurrence of a catastrophe deletes packets which are in system individually with a constant probability (binomial catastrophes). The arrival counting process is a batch Bernoulli process and the inter-catastrophe process is a general discrete time renewal process. As usual, the introduction of discrete time poses technical difficulties which are resolved by selecting a late-arrival scheme with delayed access of the catastrophe's action on the packets. Using supplementary variable (for the residual time to next catastrophe) technique the system is described by a Markov chain and investigated under steady state conditions. The paper provides stationary distributions at various embedded time points in explicit form. Because these expressions incorporates infinite products the authors provide numerical procedures by truncation and discuss the influence of the underlying system parameters.