Enhancements to reputation based trust models for improved reliability in grid computing

• Authors: P. Srivaramangai;Rengaramanujam Srinivasan
• Venue: WSEAS Transactions on Computers
• Year: 2011

A Grid integrates, coordinates resources and users from different domains. Grid computing is an interconnected computer system, where machines share resources that are highly heterogeneous. Grid computing and its related technologies will only be adopted by users, if they are confident that their data and privacy are secured, and the system is as scalable, robust and reliable as of their own, in their places. Trust and reputation systems have been recognized as playing an important role in decision making on the internet. Reputation based systems can be used in a Grid to improve the reliability of transactions. Reliability is the probability that a process will successfully perform its prescribed task without any failure at a given point of time. Hence, ensuring reliable transactions plays a vital role in grid computing. To achieve reliable transactions, mutual trust must be established between the initiator and the provider. Trust is measured by using reputation, where reputation is the collective opinion of others. The main purpose of security mechanisms in any distributed environment such as the Grid is to provide protection against malicious parties. There is a whole range of security challenges that are yet to be met by traditional approaches. Traditional security mechanisms such as authentication, and authorization, typically protect resources from malicious users, by restricting access to only authorized users. However, in many situations users have to protect themselves from those who offer resources so that the problem, in fact, is reversed. Information providers can deliberately mislead by providing false information; traditional security mechanisms are unable to protect against this type of security threat. Trust and reputation systems, on the other hand, can very well provide protection against such threats. Reputation models can be modeled in such a way they it could provide reliability for both users and providers. Reputation systems provide a way for building trust through social control, by utilizing community based feedbacks about past experiences of peers to help making recommendations and judgments on the quality and reliability of the transactions. Reputation and trust systems are soft security mechanisms which can assure behavior conformity. In this paper two new reputation based trust models are proposed. The first, model, Model 1, uses a new factor called compatibility, which is based on Spearman's rank correlation. The feed backs of the recommenders which are incompatible with those of the initiator are eliminated by using the compatibility factor. Model 2 is an improvement over the Model 1. In this model, new factors are included for measuring the direct trust. In order to effectively evaluate the trustworthiness of different entities and to address various malicious behaviors, this comprehensive trust model based on reputation, is proposed. Two important factors - context and size, are incorporated in evaluating the trustworthiness of entities.