Dynamic time-weighted popularity index: A video-on-demand case

Abstract

Information is doubling at a rate of once every few months. Video and media transport plays an important part of that growth. This growth has enthused the development of broadband access and slowly gaining increasing prominence especially for multimedia-rich Internet contents. Service operator are now figuring how to efficiently use available bandwidth for delivering targeted contents in a reliable and consistent manner. In this paper, we propose a dynamic time-weighted popularity index (TP) to exploit the unused capacity in the set-top-box (STB) by pre-storing selected contents dynamically from the content library with relative to content popularity in the STBs in order to reduce playout latency due to network capacity and delay from streaming video across the server, especially during congestion. Our aim is to quantify video popularity based on the individual subscribers' recent access patterns and interest as accurate estimation of popularity index is viable for the success of any VoD service. Experimental results show that the proposed TP index can serves as an important guide to select videos from a large content library and stored them in the STB's cache in order to decrease user perceived initial delay and increase user experience.