Optimal Stream Clustering Problems in Video-on-Demand 1
Prithwish Basu, R. Krishnan · 1998
A variety of resource sharing techniques have been proposed for supporting true video-on-demand by aggregating users into groups. Clustering of users by bridging the temporal skews between them is one such service aggregation technique. We present some recurrent problems in stream clustering and investigate optimal solutions with the main goal of minimizing average bandwidth. We evaluate the performance of heuristic and approximation algorithms for clustering and show that the static case optimal solution performs as an excellent heuristic in the dynamic case. Keywords: Dynamic service aggregation, clustering, video-on-demand. 1 This work is supported in part by the National Science Foundation under Grant No. NCR-9523958. 1 INTRODUCTION Supporting VCR-like interactions in true-video-on-demand requires the allocation of a channel to each interactive user. It has been observed that typical content access distributions are skewed. During peak demand, there is a high density of access...