Popularity-Aware Caching Algorithm for Video-on-Demand Delivery over Broadband Access Networks

Video on Demand (VoD) service is regarded as one of the most promising services over increasingly deployed next generation broadband access networks. The distributed server architecture, in which the popular content is cached at a location closer to the viewer, is a widely used methodology to optimize the transport capacity of VoD delivery. However, due to the dynamic (change with time) nature of movie popularity distribution, identifying the popular content and updating the cache servers accordingly is not straightforward. In this paper, using a novel caching architecture for Passive Optical Networks (PON), we discuss the need for a fast caching algorithm that can respond to time changing movie popularity distribution, and we propose a novel Last-k caching algorithm, which identifies the popular content using the most recent statistics. The proposed algorithm estimates movie popularity using most recent inter-arrival times of movie requests and updates the cache accordingly such that the most popular content at any given time would reside in the cache. Simulations indicate that the proposed algorithm out-performs existing algorithms by effectively responding to the dynamic nature of movie popularity distribution.