Organizing large-scale network computing systems in a peer-to-peer (P2P) fashion is a manifestation of one of the fundamental design principles on the Internet. Current research is focusing on improving P2P systems and one of the future directions is to combine P2P and Grid technologies. One of the key issues identified in the evolution of P2P technologies is the trust issue.; This thesis presents a trust model for P2P structured large-scale network computing systems. The most widely used trust modeling approach is to use a network of recommenders to obtain references and use these to predict the trust between two entities. This approach is known to suffer from drawbacks such as trustworthiness of the recommenders and scalability.; To address this problem, a solution is proposed where a recommender is independently evaluated using accuracy and honesty measures. This thesis explains using simulation results how the separation of accuracy and honesty helps in addressing the above issues. To demonstrate the utility of the trust model, a trust aware resource allocation model is developed such that it can be used to make trust cognizant resource allocations. To the best of our knowledge, this is the first study to integrate trust into resource management systems. The simulation results indicate that significant preferences gain can be obtained through this integration.
展开▼
