In order to evaluate the trustworthiness of participating peers in unstructured peer-to-peer networks, Reputation aggregation methods are used in this method. Each and every peer of the network will collect the local scores of each transaction and will compute global scores by aggregating all the local scores with the help of global scores, each individual peer can interact with its suitable peers. But the existing method will not consider the score of the new peer. In this condition, requests are handled by existing peers who leads to failure in downloading process. To rectify this, NP-TRUST model is used to distribute the request to all peers including the newly joined peers. The proposed method is compared with gossip and DFR-TRUST model in Transaction Success rate and variation in file request.
Keyword:
Trust management Security and reputation aggregation
Copyright © 2017 Institute of Advanced Engineering and Science.All rights reserved.
Corresponding Author:Vimal S, Sathyabama University, Tamilnadu, Chennai. Email: vimalshan@gmail.com
INTRODUCTIONP2P networks are attaining more attention now by the way of file sharing. In file sharing, every peer can join the network and download files. Due to open nature and anonymous network an ideal environment for malicious peer to spread virus is offered by the system. Reputation aggregation methods are used to avoid downloading inauthentic files. The trust of the system will show the reliability and safety of peers and quantified trusts [1], [2] are known as reputation score by these trusts, individual peer will avoid transaction with malicious peers. The peer with high reputation score will provide safe files in high probability, thus every peer will select peer with High reputation score. Two types of reputation scores are; 1) Global score, 2) Local score.In all the transaction, peer receives a file to evaluate its reliability with help of local score. By aggregating local scores, every peer will share a unique global score. Then each peer refers to global score and selects the transacting peer whose global score is high. If peer foes not send a file to others then local score is not calculated. Also if global score of peer is not enough, then peer is not selected as uploader. The main purpose of DFR-TRUST model is to distribute the file requests among peers to provide safe file with high probability. In DFR-TRUST model, local score calculation will not depend on number of file providing and also will select middle evaluated peers. In P2Pnetwork, departure and participation of peers [3], [4], [5] occur frequently, thus the number of high evaluated safe peer's decreases with peer's departure.The purpose of NP-TRUST model is to distribute file requests among safe peers; this will include newly joined peers and initial peers. By giving chances new peers to be evaluated intentionally, file requests are distributed among all safe peers and new peers are utilized as up loader. By this way, NP-TRUST model avoids concentration of file requests on portion of safe peers.