Effective Bandwidth Prediction through Statistical Technique over Heterogeneous Networks

Abstract

Real time Streaming Media (Video Streaming) applications are mostly popular on the mobiles and computers using Internet. Due to higher demand of video streaming through wireless network and mobile devices, video are being transmitted through various heterogeneous networks so as to efficiently deliver to the clients devices. This has resulted in the lower quality of real time video, since real time streaming media has quality of parameter requirements like high bandwidth, low packet loss ratio, higher delay and jitter. Streaming media such as video through heterogeneous networks has more challenges due to unreliable wireless networks and device mobility; moreover bandwidth, delay and loss are unknown in advance and are unbounded. In this paper, effective bandwidth prediction through statistical technique over heterogeneous wireless communication networks is proposed. Statistical technique offers computationally efficient bandwidth prediction with reasonably better accuracy. Especially with mobile devices with limited computational power and battery life, necessitates better bandwidth prediction with efficient but computationally simpler algorithms. Bandwidth predictions assist in selecting effective network for video streaming when various heterogeneous networks are available. Detailed bandwidth prediction algorithm is presented with use of quality of service (QoS) parameters data sets available online.