Abstract:
Multicast Video-on-Demand (VoD) services have become some of the most popular real-time
multimedia applications available via the Internet for last few years. Multicast communication
with patching enables clients to join an existing multicast session with out any service latency. In
this research, we propose a new distributed patching technique DCAP (Distributed Client
Assisted Patching) where the initial portion of a movie is distributed to multiple clients to store
and provide as patch stream to other clients interested to join an ongoing session within a short
time. This scheme significantly reduces the server load without requiring larger client cache
space than the similar existing systems such as Client Assisted Patching. We present detailed
algorithms for the admission control of patching clients in this research. The policy of
distributing the initial part of the movie among different clients is also formulated. The analysis
of time requirement for admission controlling, buffer requirement for the patching clients and the
bandwidth requirement of the server and link connecting the servers are presented in this thesis.
To validate the theoretical results we have done simulation of the proposed system using Parsec,
a parallel simulator suitable for simulating different entities in the VoD systems. The detailed
analysis on the simulation results reveals that the new system outperforms the previous systems
in terms of number of requests served and average waiting although it requires more time in
admission controlling for finding suitable patch client during patching. Moreover the system is
more scalable and cost effective than many other existing systems.
