SPA Seminar: Non-homogeneous Markov chains in a buffered cellular network

Abstract: We consider a discrete-time model of a cellular network with retransmissions. We focus on the analysis of a transmitter's buffer built by failed signals and define its dynamic using Lindley-type recursion. The buffer process appears to be a non-homogeneous discrete-time Markov chain, where non-homogeneity comes from the transition probabilities which depend on a time step. We then derive the steady-state distribution of the buffer using classical ergodicity conditions for non-homogeneous chains.

We also discuss briefly some performance metrics of the network (such as latency and loss rate) and the trade-offs between high reliability and low delay systems. 

This is a seminar in our seminar series on Stochastic Pocesses and Applications (SPA).