Fractal Analysis of traffic flow over Wireless Mesh Networks

In this paper, we used a ubiquitous measurement technique to assess traffic characteristics experienced by diverse sets of traffic flows routed over wireless mesh network environments. The question arises whether the traffic in wireless data networks exhibits long-range dependence as well. If so, it is important to determine if this traffic characteristic affects the provisioning and design of wireless mesh networks. We prove the existence of long-range dependence in packet delay when viewed as time-series data. Under the well-known time-domain and frequency domain estimators, packet delay on TCP/UDP flows shows asymptotic decay while Hurst index estimates suggest from moderate to strong intensity.