Traffic Prediction for Reliable and Resilient Video Communications Over Multi-Location WMNs

Abstract

This paper studies reliable and resilient deployment of video over IP across a multi-location organization, where a number of wireless mesh network (WMN) clouds are connected by a virtual private network (VPN). Particularly, we propose a scheme of enhanced SIP proxy server which can support an accountable network. In our proposed solution, the enhanced SIP proxy server consists of three modules, namely traffic load prediction, VPN bandwidth negotiation, and call admission control (CAC) in order to provide trustable service. We identify traffic load prediction as the key component among the three modules, and we further develop a linear predictor of variable sampling rate-normalized least mean square (VSR-NLMS) to estimate the traffic patterns. VSR-NLMS predictor employs adjustable sampling rate to achieve improved efficiency and performance. Numerical results show that our proposed scheme can automatically choose suitable sampling rate to track and predict the traffic load curve with acceptable accuracy and reasonable computational complexity.