Adaptive channel allocation algorithm for video delivery

Abstract

In recent years video-on-demand has emerged as one of the potential applications of multimedia. Among all multimedia applications, videos require the largest amount of bandwidth. Being a scarce resource, bandwidth utilization in efficient and feasible way always have been the main concern in video broadcasting schemes. In request based scheme [8], a unicast channel is initiated on the arrival of a request in an inter-arrival time of two adjacent multicast channels. If several requests arrive in a very short duration then each of them initiates a separate unicast channel, requiring large number of unicast channels. However, all these requests are fetching almost same video data. In proposed algorithm, we cluster all incoming requests in a short time span and serve them by a single unicast channel. The spanning duration is determined by the threshold of user's tolerable waiting time. In our adaptive channel allocation algorithm, we took a waiting time of at most one minute for users and found that for popular videos with arrival rate more than 5.5 per minute, a saving of around 90% in bandwidth can be obtained. For less popular videos bandwidth saving is> 40%. The proposed scheme can precisely predict the waiting time for any user in case of unavailability of unicast channels, while there is no such mechanism in the request based scheme. Further, all the channels in the work are only unicast channels.