A Reputation-based Trust Management System for P2P Networks

1. A Reputation-based Trust Management System for P2P Networks Authors: Ali Aydn Selcuk, Ersin Uzun, and Mark Resat Pariente Source: International Journal of Network Security, Vol. 6, No. 2, PP.227-237, Mar. 2008 Speaker: Shu-Fen Chiou (邱淑芬)

2. Outline • Introduction • Requirements • Trust and distrust rating • Resource query • Simulation results • Conclusion • Comment

3. Introduction C 1. Request a.mp3 a.mp3 responder A B 2. Calculate responders’ reputation requester 3. Dowload E a.mp3 responder (malicious) G D F a.mp3 responder

4. Requirements • Prevent the malicious node to provide the malicious files using the node’s reputation. • Security • Key management • Denial of service protection • Avoiding fake file downloads • Free riders problem

5. Trust vector • Every peer stores the trust vector for every other peer it has dealt with in the past. l: vector length 0: dishonest 1: honest

6. Trust and distrust rating • Minimum distrust is given priority in trust comparisons over maximum trust. complement 11101 00010

7. Resource query x1 responder Response{QID, RID, qID, hash(f1)} x2 responder x3 Request{qID} responder x9 … x4 responder x5 responder x6 responder x7 responder

8. E.g.

9. Simulation • Attackers • Naïve • Hypocritical • Collaborative • Pseudospoofing • Pseudospoofing with collaborators

10. Simulation results • Reduce an attacks’ effectiveness. • A large degree of protection can be obtained by just evaluating one most trusted response. The gain from > 2 appears to be negligible. • The protocol is similarly effective for both 1% and 10% malicious peer density.

11. Conclusion • Presented a reputation system for establishing trust in P2P network that helps preventing the spread of malicious content.

12. Comment • 初始信用值的設立 B A honest B’s initial reputation:0.5 0.5+0.1=0.6 B A dishonest B’s initial reputation:0.5 0.5-0.1=0.4