Enhanced video streaming using dynamic quality control with bandwidth prediction

Abstract

Video streaming is one of the most widely used applications of internet users. However, bandwidth fluctuations especially during peak periods can seriously impact the Quality of Experience (QoE) of users by causing intermittent freezing of the video. This paper proposes a client-oriented video streaming application which dynamically changes the video quality by using a bandwidth prediction mechanism. Bandwidth prediction is achieved by incorporating the moving average prediction algorithm in the application's logic. The application has been developed in Java and protocols such as the Real Time Messaging Protocol (RTMP) and HTTP, as well as libraries of the System Information Gatherer and Reporter (SIGAR) and VLC media player have been integrated. Moreover, a new objective quality assessment metric, the freeze time, has been proposed to evaluate the performance of the scheme. The application was tested by streaming videos from YouTube in real time. Results show a major reduction in freeze time of over 60% and a gain of at least 2 dB is Peak Signal to Noise Ratio (PSNR) as compared to static video streaming.