A Markov Chain Model of Streaming Proxy for Disconnecting Vehicular Networks

Abstract

Frequent loss of network connectivity makes media streaming very challenging in vehicular mobile communication scenarios. It has been shown earlier that a streaming proxy onboard a vehicle can be effective in mitigating the adverse effects of temporary network disconnections. During the connected period, a vehicular proxy pre-fetches as much media as possible ahead of the client's playback time. This pre-fetching is done by utilizing any excess bandwidth in the network connection. The pre-fetched contents are stored in a local storage and played out when the connectivity is temporarily lost. A number of research studies have been conducted to understand the performance dynamics of such streaming proxy systems in vehicular networks, however, no analytical model has been proposed yet. In this paper, we propose a 3-D Markov Chain model to analytically study the performance of such streaming proxies as a function of the system load. The model is validated by means of discrete- event simulation of a realistic networking scenario. The proposed model can be effectively used to dimension the streaming proxy systems in next-generation vehicular networks.