Methods for Efficient and Secure Service Availability in Peer to Peer Overlay Networks

Abstract

Peer-to-peer overlay traffic forms a major part of the Internet traffic. Peer-topeer newlineoverlays have prominent applications in file-sharing, communication, and newlinecontent distribution. Peer-to-peer paradigm is largely applied in file-sharing newlineover internet. More than 54 peta bytes of data is being shared on these networks. newlineDue to their decentralised nature, and large size, searching for a file newlineis an important service. Efficiency of search service affects the overall performance of the overlay. User directly perceives the effectiveness of search. newlineIn this work, approaches for improving efficiency and quality of search are newlinedeveloped and evaluated. Another important consideration for decentralised newlinenetworks is security. In a peer-to-peer overlay, the end systems act as routers. newlineTraffic passes through end-systems. Also end systems contribute to the storage, newlineand computational resources of the network. In this work, we address newlinean important problem known as Sybil attack which can exploit the routing newlineservice and resources shared by other honest participants for selfish purposes. newlineTwo basic approaches to search in peer-to-peer overlays are flooding and randomwalk. newlineFlooding has the advantage of quick response and enormous traffic newlineand randomwalk has the advantage of negligible traffic and very delayed newlineresponse. The in-between approach consists of intelligent selection of neighbours newlineto forward the queries. A content-oriented metric is proposed which newlineproved to have edge over the other metrics. Other metrics are built upon newlinethe observations over a period of time. This alone is insufficient to guide the newlinequeries. Therefore the proposed approach considers what type of content the newlineneighbours are sharing and how popular it is and compares this with what newlinetype of content query is looking for and how popular is the content. This information is subjective and is modelled using Fuzzy Sets.