Font Size: a A A

Research On The Multifactor Fuzzy Trust Model On P2P Networks

Posted on:2010-06-06Degree:DoctorType:Dissertation
Country:ChinaCandidate:H Q LinFull Text:PDF
GTID:1118360302971178Subject:Computer system architecture
Abstract/Summary:PDF Full Text Request
Due to the excellent scalability and flexibility, the P2P (Peer-to-Peer) networks are applied to the fields of content sharing, cooperative work and distributed computing applications etc. But the open and anonymous natures of P2P networks induce trust absence among P2P users. It blocks the progress of the P2P networks badly. To resolve afore mentioned issues, one effective way is developing strategies for establishing trust through reputation. Based on the survry of study and progress of trust model, a hierarchical fuzzy system-based multifactor trust model is proposed.Identify the trust factors. In the most existing trust model, the feedback is used to evaluate the trust index of peers. Since these schemes only take into account a monofactorial feedback of the peers on past transactions, which cannot reflect the peers' total characteristics, the trustworthiness of peers is hard to be evaluated accurately. Since trust between peers is an essential attribute to a successful transaction and the factors to be considered also vary in different situations, we need to find out what factors will affect the completion of transaction. For example, in the P2P e-commerce network, the scale of the company, product quality, after-sales service, delivery time etc. may be considered to evaluate the trust. While in the distributed computing applications, the processor performance, online time and idle computing cycles etc. may be utilized to evaluate the trust on the capability of computation. The trust model should take into account of comprehensive trust factor. A P2P file sharing application is used as an example in this paper to illustrate our trust model. The trust factors are divided into five categories: capability factors, responsibility factors, server willing factors security factors and additional decision factors related with download task.Propose a hierarchical fuzzy trust model. The fuzzy theory is quite useful in manipulating imprecise or uncertain information. In P2P network, the trust relationships among peers are hard to assess due to uncertainties involved. Fuzzy logic offers the ability to handle uncertainty and imprecision effectively, and is therefore ideally suited to reasoning about trust. But the number of fuzzy rules grows exponentially with the number of input variables. The hierarchical fuzzy system provides a way to deal with this problem. Comparing to the conventional fuzzy system, the fuzzy rules number of our Multifactor Fuzzy System is reduced by 96.8%. In order to improve the evaluation of trust, the relationship among fuzzy variables should be more relaxed. It is shown in simulation that our trust model can not only resists the malicious behaviors, but also greatly improves the efficiency of P2P system and implement the workload banlance.Propose a novel trust data storage scheme: Trust Relationship Network (TRN). How to implement storage of trust data in the P2P networks which absent the trusty center servers is another key issue. Most existing trust model used DHT and random selection to store data. However, these methods can not work well in many occasions. The Trust Relationship Network (TRN) proposed by this thesis has two stable states: Random Trust Relationship Network (RTRN) and Supernode-Based Trust Relationship Network (SNTRN), which can combine the decentralized storage and centralized storage. The simulations show that the TRN has well performent in terms of hit numbers, network overhead and response time. The SNTRN has short and steady fault recovery time. And the network overhead increases linearly with the increase of the fault supernodes number. Hence, the SNTRN has good scalability.Secure transmission of trust data. A CL-PKC-Based secure transmission protocol is proposed in this thesis. The supernodes of hybrid P2P network are selected to be Key Generation Center (KGC). The protocol avoids the escrow problem in identity-based cryptosystem, and the identity binding technique is used to ensure secure delivery of private keys. The strand space model is used to prove the security of trust query schemes.
Keywords/Search Tags:hierarchical fuzzy system, trust model, trust data, bilinear pairins, certificateless public key cryptography (CL-PKC), strand space, Peer-to-Peer
PDF Full Text Request
Related items