• RSS
  • LinkedIn
  • Facebook
  • Twitter

Wireless Peer-to-Peer Networks

Various peer-to-peer protocols exist for file sharing over the Internet. However, when all nodes are wireless with limited energies, none of the existing protocols will be efficient. We propose to develop a cross-layer design for energy-efficient wireless peer-to-peer protocols for various types of media. The protocol will most likely benefit from parallel downloading. An information theoretic study, building on the theories of relay networks and cooperative networks, will reveal the optimal strategy for this, and will certainly have implications in the source coding algorithms used in describing the shared media.

A P4P Approach for the Distribution of Multimedia Content over the Internet

Project Funded by Turk Telekom
The project will host 2 M.Sc./Ph.D. students. Students will have the chance to work closely with Turk Telekom and the research outcome will potentially impact TT’s tivibu service. Scholarships will be provided. The scholarship levels are highly competitive and will be based on the students’ qualifications.

Summary
In recent years, the peer-to-peer (P2P) technology has developed rapidly with various applications. P2P traffic has become very prominent in broadband networks and on the Internet. It has been measured that over 80% of the average Internet traffic is now from P2P applications, and upwards of 90% during peak hours.

P2P is a decentralized, self-organizing system of sharing resources between individual users. Being clients and servers simultaneously, P2P nodes obtain services from and provide services to other nodes. A user requesting a specific resource, with the use of a P2P protocol, may download it from many dynamically changing service providing peers on the Internet in parallel. Use of a multitude of peers from different parts of the Internet greatly reduces the download time. The P2P protocols are all defined at the application layer, fully independent of the routing protocols of the network layers, and therefore traditional routers cannot optimize P2P routing. Therefore, while current P2P protocols are beneficial to end-users, they may cause significant problems to Internet service provides (ISPs).

The P2P traffic has properties of potential uplink and downlink symmetry, and random interconnections between peers. It is likely for P2P protocols to choose cross-domain peers, increasing the cross-domain traffic significantly. Measurements conducted by Verizon in 2009 point out that the average hop count for P2P traffic is 5.5, with 90% cross- domain traffic penetrating the metropolitan networks (MANs). The typical MAN bandwidth planned for traditional Internet services cannot meet the growing P2P requirements. Further worsening the situation is the success of data download and video services that have increased the P2P user base with even faster traffic growth. For example, BBC in UK has launched the iPlayer service where users, after installing a P2P client server software, can stream or download a weekʼs worth of BBC TV and radio content. The service has been a great success, with over 1 million subscribed users having watched 3.5 million programs only 2 weeks after the launch. The service grew by 25% every month. In 2008, BBC has expanded the iPlayer service to iPhone and iPod users. However, just one week after the launch of the service, the cost of leasing of the traffic more than doubled for BBC, and its streaming server expenses tripled!

The increase in P2P use has been growing so fast that it has become a “traffic killer” for many of the ISPs. P2P applications bypass client-server software by enabling to terminals to directly communicate with each other and exchange data resources. As such, bandwidth requirements can easily reach the upper limit of the total network bandwidth resources, thereby, greatly impairing the performance of the critical network applications. The disordered data exchange between two nodes fails to optimize routes, and irrationally uses backbone network bandwidth. Therefore P2P applications occupy a large number of network devices and bandwidth resources, generating unstable network transmissions.

It has recently been suggested that ISPs can revert all of the aforementioned problems associated with P2P traffic to their benefit by a cross-layer design which can besummarized as a network-aware P2P. A working group within IETF, alto, has been working on a cross-layer P2P protocol design towards this end. This solution has been commonly referred to as the “Proactive Network Provider Participation for P2P (P4P).” The main idea behind P2P is to intelligently combine the benefits of Content Distribution Networks (CDNs) and P2Ps. The alto group meets regularly within IETF and will finalize the first release of the P4P protocol in their August 2010 meeting.

The research project summarized herein focuses on the development and application of a P4P platform for multimedia services that Türk Telekom may provide. The outcome of this work will be readily applicable to Türk Telekom services such as “tivibu.” The P4P platform that will be developed in this project would not only enhance the user experience, but also provide a controlled and distributed traffic across the network reducing the cost of infrastructure to support such services. The iPlayer experience of BBC show the importance of such a platform for future media services over the Internet.