The 6-hop peer-to-peer system

The proposed 6-Hop P2P (6-Hop) is an improved overlay network topology that intends to enhance the current P2P that houses a file sharing and media broadcasting (FSMB) application. A P2P system provides a serverless platform for the current FSMB application. It allows media files in each peer to be located by other peers in the network using any word in the file name. The current FSMB application is a closed system where only registered peers can join the overlay network. However, the current broadcasting system has shown some aspects of inefficiency. Two principle weaknesses identified are as follow: query messages flood the overlay network which taking up the bandwidth of the network; and, the search service is not reliable. When a new peer joins the P2P-based FSMB system, registered peers will randomly join clusters in the network. When a peer queries a specific or part of a file name, the current P2P examine all the peers and all the files in each peer. This has resulted in low recall rate and heavy query messages in the overlay network. The proposed 6-Hop P2P is an overlay network that adopts small-world characteristics attempting to improve the recall rate and to reduce the amount of query messages. The proposed P2P is designed using the "similarity interest" value to reduce unnecessary routing and to cut down the number of hops required by the search. It also rewires peers base on recall-rate learning algorithm to form clusters relying on the ‘interest’ bit value of QRP table. This paper presents the proposed 6-Hop model and the work-in-progress results with a short discussion on its prototype.

I. Introduction

P2P [1], [2], [3], [4], [5], [6], [7] has been used extensively for file sharing and media broadcasting. However, for P2P to deliver reliable services without high search time, low recall rate and minimum bandwidth consumption, challenges still persist. In this research, it has defined a set of objectives: 1.

To evaluate P2P in particular P2P with small world characteristics, and to justify the 6- Hop P2P design

2.

To design and prototype the proposed 6-hop P2P to deliver high clustering co-efficient and low average hop attributes

3.

To design new and enhanced the current algorithms to achieve high query recall rate and low message flooding by developing QRP ‘interest’ oriented clustering, rewiring using a recall-rate pattern learning algorithm

4.

To design and prototype algorithms and theories for local and aggregated ‘interest’ value using QRP data structure.