On Minimizing Packet Loss Rate and Delay for Mesh-based P2P Streaming Services


Visiting researcher: Mr Zhiyong Liu (CATR)
Host researcher: Dr. Dan He(University of Surrey)
Duration: 3 months
Status: Ongoing

Biography of the visiting researcher

Mr. Zhiyong Liu got M.Sc. degree of Communication and Information System in China Academy of Telecommunication Research, MIIT, in 2005. Since September 2005, he has been an engineer in China Academy of Telecommunication Research. He has published over 20 technical papers, two of which are indexed by EI. He has taken part in many research projects since 2005, of which the Europe-China Grid InterNetworking project (EC-GIN, from 20/11/2006 to 10/3/2010, European Sixth Framework STREP FP6-2006-IST-045256) is the most noticeable one. Within EC-GIN projects, he developed 3 software modules, contributed 12 deliverables, and got two software copyrights. He also holds 2 patents for inventions. He also leads several major projects from Ministry of Industry and Information Technology of China. His research interests include Next Generation Network Architecture, Mobile IPTV, Multimedia Communication Systems and Applications and Network Security Technologies, etc.

Background and motivation

4G technologies will eliminate the wireless access network bottleneck, making it possible for the development of higher quality video communications. Mobile IPTV services will be the killer application for 4G networks, getting more operators’ attention in a big way by becoming integrated in connected TV, video surveillance equipments, gaming consoles, etc. Basically there are two kinds of architecture for delivering mobile IPTV services on the Internet: the traditional client-server based paradigm and the novel peer-to-peer (P2P) based paradigm.
P2P based streaming technology enables one peer cache and relay content to other peers in a swarm in the meanwhile downloading the content. This characteristic allows P2P streaming to form a transport overlay and scale across plenty of peers involved in the transmission, without investing on building costly and complex CDN network. But many currently deployed P2P streaming systems suffer from unacceptable QoE, for example low video quality when network condition degrades or P2P overlay churns, and relatively long playback latency and startup delay. Another problem posed by P2P streaming systems is that they generated large amount of traffic on the underlying network. The main reason of these aforementioned problems lies in the network-oblivious policy in constructing P2P overlay. For example in mesh based P2P streaming system, peers form an overlay at the application layer, while using the underlying IP network to exchange chunks. But traditionally these two layers are completely independent, resulting in sub-optimal peer selection and P2P topology, always leading to poor Quality of Experience (QoE) and a waste of network resource, thus preventing P2P streaming services, which have rigorous QoS constraints, from the possibility to be deployed for commercial services.
In this project we study the design and evaluation of a novel peer selection and mesh construction algorithm, with the aim to build a mesh overlay which can minimize the maximum possible packet loss rate while meeting upper bound on mesh delay, thus optimizing the user perceived quality for P2P based streaming services.

Research activities

First, to jointly develop a full parametric opinion model formulate video quality as a function of various other parameters affecting video quality, so as to predict video QoS of each potential link between video receiver and active server. This parametric opinion model should be optimized to mobile video scenario.basing on the research results of video QoS, and to
Second, to develop novel Enhanced Optimum Server Selection algorithms for mobile IPTV video services based on the parametric opinion model. With this model, the receiver of video services can initiate QoS measurement procedure and predict video QoS of each potential link between video receiver and active servers, then select and change to the optimum video server to deliver the video to mobile users.


To be updated.
Powered by Drupal, an open source content management system