Research On Multi-Agent System Collaboration Based On Trust Mechanism

With the development of artificial intelligence, collaborative research on Multi-Agent System has become an important issue. Collaboration is the process when multi-Agent faced problem and tasks how to settlement by a rational and effective method. It includes how to deal with conflict in tasks and which method can be choosing in solution process. At present, Collaboration research is focused on two areas: the theory-oriented research and application-oriented research. Theoretical study contains building a collaborative framework and cooperation mechanisms in large-scale autonomic system. Applied research focused on information collection, system optimization, Mobile Agent, robot soccer and so on.This paper studies the trust mechanism based on Multi-Agent System (MAS) collaboration problem. It is mainly to solve how to establish MAS collaboration system by biological intelligent method, and to achieve the optimal allocation of system resources in a dynamics open MAS environment. And it will be introduced into the trust mechanism in the MAS to achieve the Agent for the treatment of the task more efficient, and saving system cost.Currently, researchers using a variety of methods for multi-agent collaborative research, some of the more well-known game theory on the Agent coordination and collaborative has been researched. The method has gained a lot of research results, but they are primarily designed for the collaborative model reflects the in the formal description aspects. The model itself does not address what kind of optimization algorithms used is explained. In this paper, the artificial immune algorithm for design of a collaborative model for multi-Agent systems, from the algorithm level, collaborative model for multi-Agent systems have been studied.Distributed artificial intelligence research in other areas progress the MAS collaborative research. In the 21st century, researchers found that in open MAS, as Agent may at any time enter or leave, and because of Agent's own self-serving nature, the traditional method is no longer suitable for the open environment. Because in the open MAS system, each Agent's own characteristics, as well as the relationship between the Agents on the system affected is very obvious. This need to introduce a new solution. And how to deal with these relationships will be improving the problem solving abilities of MAS system. Through researching found that Agent characteristics, the introduction of the system of trust and reputation mechanisms of Agent can be more and better to dealing with Agent collaborative problem. Thus for the Agent's trust and reputaion mechanisms become the important issue in the MAS collaborative system.In this paper, MAS collaboration and trust mechanisms are fully studied and discussed, and we find in the collaborative process can be introduced trust mechanisms, and make system more rational and more effective.The main contribution of this paper is as follows:(1) Through the researching current MAS collaboration system, proposed a collaborative approach based on artificial immune system model. Through a multi-Agent interaction in the relationship between the various Agents provides the main parameters of the system, so that system can better solution Multi-Agent System in the task decomposition problem.(2) Proposed trust model (E-FIRE) applying in open Multi-Agent Systems. The model introduces personality characteristics of consumers and service providers. The FIRE model was modified to more effectively reflect the interaction of Agent difference between the personality traits of trust models. Meanwhile, when consumer choicing providers, provider of integrated computing confidence rating, the greater reliance on evidence of a direct provider of consumer confidence, thereby reducing the amount of communication between Agent. The experimental results show that when there are fewer numbers of times in the interactive, E-FIRE model performance and performance of the model rather than FIRE; with an increase in the number of interactions, E-FIRE performance of the model is better.(3)In the process of Multi-Agent Systems collaboration research, the trust mechanism will be introduced into the integrative negotiation It solved the problem of attitude about two negotiating sides facing the same problem or task in the integrative negotiation process.(4) In the process of Multi-Agent Systems collaboration research, the trust mechanism will be introduced into the contract net protocol (CNP). For the CNP application can effectively avoid the honesty or dishonesty environment on systems. It has a certain theoretical and practical significance.