P2P Networking and Communication


Introduction:

 

P2P stands for "Peer-to-Peer", which is a new technology developed in the past few years. The intention of  P2P is to address the common problems in the traditional Client-Server architecture such as single point of failure and scalability issue. Peer can send/receive data from not only the server but other peers. At the beginning, P2P network is used for file sharing, which does not require that data be delivered by specific time. Recently, with the development of the video compression technologies and more user bandwidth, streaming video is becoming attractive, especially using P2P technology.

 

IP multicast is supported by some private networks, and will be fully supported in the next generation Internet. For example, Internet-2 , a private network that inter-connects most major universities in the US and is used for education and research, fully supports IP multicast, and application platforms such as Microsoft Research’s ConferenceXP platform  have been developed for such networks. Reliability in multicast scenarios is, however, typically more challenging than unicast scenarios as it suffers from several problems, including diverse client bandwidth and capabilities, bandwidth inefficiency in multicast retransmission, and potential sender overloading.

 

An overlay network is a computer network which is built on top of another network. Nodes in the overlay can be thought of as being connected by virtual or logical links, each of which corresponds to a path, perhaps through many physical links, in the underlying network. For example, many peer-to-peer networks are overlay networks because they run on top of the Internet.

 

Project Summary:

 

 

This project is to address the scalability and reliability problems in the IP-multicast network using P2P overlay technology.  The main observation is that  in the streaming  session, if congestion occurs in the IP-multicast network, a group of peers will have the same congestion problem. Once the problem occurs, the peer should try to find peers with better data reception for retransmission. These helper peers are denoted as retransmission node, or "rtx node" . Most of good "rtx node" will have a disjoint path to the server from the node with a receiving problem. The retransmission nodes for the receiver should be those end hosts that have been recently having little problem receiving packets from the original sender (source), have a good network connection to the receiver, and are typically closer to the receiver than the original sender.

 

To find the good disjoint "rtx node", we need to build a simple overlay network among the participants. Since the overlay network is used for retransmission purpose, the construction and maintenance of the overlay network should be light weight. Each receiver only needs to identify one or two "rtx node". With the identification of a couple of good retransmission nodes, a receiver can send the re-transmission request (with the original sequence number of the lost packets and some other necessary information) to one of its retransmission nodes who typically would have received or will receive that packet. The retransmission node then, upon receiving the re-transmission request, forwards the requested packet using a separate unicast RTP session to the requesting receiver.

 

The architecture we proposed is a novel, highly distributed overlay peer-to-peer retransmission architecture that exploits path-diversity by taking advantages of both IP-multicast and an overlay network. The architecture leverages both disjoint-path-finding and periodic selective probing to take into account peer’s recent packet loss probability, retransmission delay and recent retransmission success rate. The advantage of this architecture is that it helps to improve the reliability and scalability of  IP-multicast network for different applications such as VOD and live video conferencing. It can help peers to dynamically find good rtx node to address the packet loss issue with reduced delay and little overhead.

Our approach is an innovative exploitation of path diversity and peer-to-peer collaboration to address the QoS of IP multicast based multimedia applications. Simulation results show that it can significantly improve the reliability, delay, and scalability of IP-multicast based multimedia applications. Future work will include more quantitative evaluation of the impact of the proposed framework on the perceived quality of the multimedia applications in both simulation and real networks. The proposed architecture can also be further extended to bridge IP multicast networks and non- IP multicast networks and provide a unified framework to address both multicast connectivity and QoS performance.

 

Papers:

 

1.    Y. Zhu, W. Zeng and H. Lu, “Exploiting Overlay Path-diversity for Scalable Reliable Multicast,” IEEE International Conference on Multimedia and Expo, July 2007.

2.   W. Zeng, Y. Zhu, H. Lu, and H. Jiang, “A novel path-diversity overlay retransmission architecture for reliable multicast,”  IEEE Inter. Confer.  Multimedia and Expo, July 2006.

3. ��Technical Report on "Path-diversity P2P Overlay Retransmission for Reliable IP-multicast", Oct, 2007.

 


 

 

Last Updated :

11/21/2007 22:07:06