Research On Service-Oriented Automated Negotiation In Grid Environment

Grid denotes an advanced distributed computing infrastructure. It is an integrated computation and resource environment. It contains various computational resources and converts them into standard and economical computing power. Grid technologies support the sharing and coordinated use of diverse resources in dynamic, distributed virtual organizations. The geographically and organizationally distributed components of grid are sufficiently integrated to deliver desired QoS.The establishment of grid infrastructure enables the worldwide cooperation. The participants of cooperation form a virtual organization. During the cooperation course, members of the virtual organization provide service to others, and consume the service provided by the other members. Thus the responsibilities and rights of a member should be determined before the cooperation begins. This requires the parties (represent by agents) forming the virtual organization to negotiate these issues.In this thesis, the characteristics of grid environment are mentioned. The theories and models of automated negotiation in MAS field are studied. This thesis focuses on multi-agent multi-issue negotiation. The main results of the research are as follows.1. A new many-to-many multi-issue negotiation model, which allows agents to send their proposals simultaneously, was proposed. The denotation of the outcome of negotiation was presented and the interaction behavior of agents was standardized by defining negotiation protocol.2. The initial proposal generation algorithm and trade-offalgorithm were developed based on genetic algorithm.3. A security mechanism based on encryption algorithm and digital signature technique was design to ensure that the negotiation process smoothly.The negotiation model proposed in this thesis can be applied to occasions where many-to-many multi-issue negotiation occurs. This model shows its advantage in strict time-constrained situation because it is based on genetic algorithm. Experiment results show that the algorithm developed in this thesis is effective and feasible.