Analytical Model of On-Demand Streaming Services Based on Renewal Reward Theory

Abstract

We propose an analytical model based on renewal reward theory to investigate the dynamics of on-demand streaming service. At the same time, we also propose a simple method combining multicasts and unicasts that can reduce the download rate from the streaming server without causing delay. By modeling the requests as a Poisson arrival and using renewal reward theory, we study the dynamics of this streaming service and derive the optimal combination of unicast and multicast. We even show how to estimate the fluctuation of download rate of the streaming service.