Reputation-Aware Contract-Supervised Grid Computing

Abstract

Service-related information available in current Grid is usually static, limited and self-stating. The lack of evaluation from service cooperators and record of past transactions hinders information sharing, isolates entities and separates related transactions. As a result, QoS (Quality of Service) and QoP (Quality of Protection) are hard to be guaranteed. Inspired by observations from human society, we propose a reputation-aware contract-supervised Grid computing model. The adoption of reputation targets at providing an evaluating mechanism, enhancing predictability from past experiences, promoting efficient, secure and reciprocal cooperation between entities, and forming a benign cycle in Grid computing. And contract is introduced to be a supervising mechanism. By this means, a context sensitive, deception detectable, criteria clear, and bias correctable reputation computing will be achieved. Besides, this scheme also develops some kind of CBR (Case-Based Reasoning) capabilities.

This work is partially supported by the National 863 High Technology Plan of China under the grant No. 2002AA1Z201, China 863 OS New Technologies Research under the grant No. 2003AA1Z2060 and the Grid Project sponsored by China ministry of education under the grant No. CG2003-GA00103.