PhD Thesis Presentation
Gossip-based Real Time Communication over Internet

by Mr Wing Hei LUK

 :  23 Aug 2013 (Fri)
 :  8:45am
Venue  :  Room 2463, 2/F (Lifts 25-26), HKUST

Examination Committee
Prof Riki TAKEUCHI, MGMT/HKUST (Chairman)
Prof Chin-Tau LEA, ECE/HKUST (Thesis Supervisor)
Prof Albert K S WONG, ECE/HKUST (Thesis Co-supervisor)
Prof Rocky CHANG, Department of Computing, The Hong Kong Polytechnic University (External Examiner)
Real-time group communication forms a part of many interactive multimedia applications over the internet. In scenarios that involve large group sizes, sporadic sources, high user churns, and random network failures, gossip-based protocols can potentially provide advantages over structure-based group communication algorithms in ease of deployment, scalability, and resiliency against churns and failures.
We first propose a novel protocol called Redundancy Reduced Gossip (RRG) for real-time N-to-N group communication. We show that our proposed protocol can achieve a considerably lower traffic load than conventional push gossip protocols and conventional push-pull gossip protocols for the same probability of successful delivery and that its performance gain is higher in networks with smaller delays. We derive a mathematical model for estimating the frame non-delivery probability and the amount of overhead, and demonstrate the general correctness of the estimates by simulation. Lastly, we implement a functioning conferencing system using the protocol, completed with NTP synchronization, dynamic group size estimation, redundancy suppression, and other features needed for proper operation. We perform experiments over the campus network and PlanetLab. The prototype system demonstrates the ability of our protocol to maintain robust performance in real-world network environments.
To further reduce the traffic load, we extend RRG to incorporate random network coding. The extended scheme is called Network Coded Redundancy Reduced Gossip (NC-RRG). We show that a significantly lower traffic load than conventional push gossip protocols can be achieved for the same probability of successful delivery. Because the connectivity between peers is established in each cycle in our gossip scheme, redundancy can be effectively reduced via a progressive decoding and innovative transmission scheme. We derive a mathematical model for estimating the frame non-recovery probability and the overhead, and demonstrate the accuracy of the estimates by simulation. The delay and overhead tradeoff between the inter-frame and the intra-frame network coding schemes is also presented.
Recently, the idea of Content-Centric Networking (CCN) was proposed.  CCN replaces named hosts with named data on the internet and requires caching of content in network routers. We exploit the caching and content tracking capabilities of CCN to design a group communication protocol that can not only meet the stringent delay and loss requirements, but also reduce the total traffic volume significantly. Analytical and simulation results are provided to verify the performance and the advantages of the proposed protocol.
*** ALL ARE WELCOME !! ***
*** ALL ARE WELCOME !! ***