WGM: Wavelet-Based Gamma Model for Video Traffic in Wireless Multi-hop Networks

Abstract

While most of the existing literature concentrates on wireless traffic characterization and its effects on network performance, relatively little research work has focused on mathematical modeling and queuing analysis of video traffic in wireless multi-hop networks. The purpose of this paper is to characterize and model wireless multi-hop network video traffic. We begin with thorough analysis and investigations of characteristics of video traffic in typical wireless network scenarios, building upon which we present a novel Wavelet-based Gamma Model (WGM) for wireless multi-hop video streaming. Our analytic and simulation results demonstrate that the WGM provides a flexible and robust means to characterize wireless video traffic, in terms of statistical properties and self-similarity. On this basis, a WGM queuing system is constructed to deduce the theoretical value of buffer overflow probability, delay and delay jitter for wireless media streaming. A series of simulation experiments are conducted to demonstrate the generality and effectiveness of our modeling approach. Our experimental results show that the probability of buffer overflow indicates a hyperbolic decay as the buffer size increases; however, the buffer overflow probability decays exponentially fast, as the sending rate increases. The study also finds that, as the buffer size increases, the delay and delay jitter of video streaming first increase then become stable, but with an increase in the sending rate, both values decrease sharply. We believe that this conclusion could contribute to the design of appropriate network architectures and to elaborate efficient wireless multimedia communication protocols.