            ! " #"#$ %%# & ’  ((() © 2010, IJARCS All Rights Reserved 474 ISSN No. 0976-5697 An Approach to find Trustworthiness among Different Domains in a Grid Environment Dolly Sharma* Deptt. of IT, BUEST Baddi, HP, India dollysharma83@ymail.com Sarbjeet Singh CSE, UIET, Panjab University Chandigarh, India sarbjeet@pu.ac.in Seema Dept of CSE, Amity Unversity, Gurgaon, Haryana, India mudgil.seema@gmail.com Abstract: The goal of Grid computing is to create illusion of a simple yet large and powerful self managing virtual computer out of a large collection of connected heterogeneous systems sharing various combinations of resources. Such an environment introduces challenging trust related issues as both service providers and users can come from mutually distrusted administrative domains and any of them can behave maliciously. The use of trust evaluation simplifies the security architecture and is evaluated on the basis of a number of parameters like trust decay, reputation, trust updation, transitivity etc. A number of models have been proposed by different researchers for the evaluation of trust but many of them have missed one or the another required parameters that are necessary for the evaluation of trust in a comprehensive way. In this paper a novel approach for evaluation of trust has been proposed that insists on the use of a number of imnportant parameters to calculate trust in a comprehensive way. Keywords: Grid computing, trust, reputation, trust model, feedback, trust evaluation, trust decay. I. INTRODUCTION Grid technology brings together a set of resources distributed over wide-area networks and supports large-scale distributed applications by coordinating resource sharing and problem solving in dynamic, multi-institutional, virtual organizations [1]. Security requirements are fundamental to the grid design [2]. Rasmusson and Jansson [3] categorized security as hard security and soft security. Hard security is achieved through cryptographic mechanisms, encryption techniques etc. But it overshadows the essence of soft computing that the resources can be accessed directly. Integrating trust in grids is one of the ways to achieve soft security in grid environment. Applying trust to grid computing provides a mechanism for entities to manage risk arising due to interactions taking place between different entities. Trust is a social phenomenon, and can be defined as a firm belief in the competence of an entity to behave as expected such that this firm belief is a dynamic value associated with the entity and is subject to the entity’s behavior and applies only within a specific context at a given time [4]. A lot of attempts to evaluate trust in the field of distributed systems have been initiated. Important among them include [4, 5, 6, 7, 8, 9]. A survey of these trust models has been presented in [10]. Some of these models lack strong mathematical foundations whereas others have missed out one or the other required parameters for trust evaluation. Based on the literature survey a list essential trust related parameters has been identified. These parameters have been described in Section III. Finally a trust model has been proposed in section IV that insists on the use of all the trust related parameters identified in section III to calculate trust in a comprehensive way. II. RELATED WORK A number of models have been proposed by different researchers for the evaluation of trust in grid. A summary of classification of these models has been given in Table II. A. Abdul-Rahman and S. Hailes proposed a Trust- Reputation Model [1] based on trust characteristics from social sciences. Trust is context-dependent and based on prior experiences. Trust supports negative and positive degrees of belief of an agent’s trustworthiness. Trust is not transitive but subjective, dynamic and nonmonotonic. Trustworthiness is evaluated on the basis of experiences and reputation. An experience results from direct interaction. A reputation is an expectation about past behavior of an agent and is calculated from a trusted set of recommenders. F. Azzedin and Muthucumaru proposed a Trust model for Grid Computing Systems [4] which is an extension to [1] and [11]. They insist that direct trust weighs more than recommender trust. The model also lets a newcomer to build its trust from scratch by enforcing enhanced security. Here trust is dynamic, context specific, based on past experiences and spans over a set of values ranging from very trustworthy to very untrustworthy. Trust is evaluated on the basis of direct trust and reputation. A Recommender trust factor is introduced to prevent cheating via collusions among a group of domains. Farag Azzedin and Muthucumaru proposed a Trust Model [7] for peer to peer computing systems also. In [4], an accuracy measure has been associated with each