Optimised batch patching with classes of service

Abstract

In this paper we present a new technique called Optimised Batch Patching with Classes of Service(OBP with CoS) which can be used to leverage the benefits of multicast within the context of 'near' video-on-demand systems in IP networks. OBP with CoS builds on an earlier scheme, known as Optimised Patching, but is different in two respects. Firstly, in OBP with CoS, the server artificially delays requests in order to increase the probability of accumulating duplicates, thereby allowing greater exploitation of multicast compared to Optimised Patching albeit at the expense of higher latency until commencement of service. Secondly, the client is able to request a latency class of service which reflects how long the client is prepared to wait until playout of the video. The server scheduling algorithm attempts to minimise network bandwidth consumption within the constraints imposed by the class of each request. We present analysis and simulations of our scheme in order to validate its effectiveness.