Research On The Local Structure In Mas

Along with the rapid development of the information technology and the networktechnology, Scale and complexity of the complex Systems is also increasing sharply. Tosolve this problem in MAS theory and method is getting more and more concerned.So it isnecessary to find the solutions with improving efficiency and profits of MAS, Internet’sinformation Transfer and control, self-adaptation in Agent organisations, which have beenthe hot topics. Most of the traditional study are int a global control train of thought, whichis not adaptive for the large-scale MAS. This paper analyzed the local structure of theMAS, put forward a Distributed coalition structure generate algorithm based on localinformation. Also give a method to solve the problems mentioned above, such as thecontrol of the Internet’s information and the self-adaptation in Agent organisations.This paper mainly include the following aspects:Numbers of coalitions in the MAS increase exponentially along with the number ofAgents,So it is not realistic to search all the coalition structure. In view of the aboveproblem, we give a distributed coalition structure generate algorithm based on individualagent local information and the systems global income. The algorithm follows the localoptimum principle and the characteristics is n Agents’ top-down parallel search in thecoalition structure graph, thus we can short the search time.The structure of the MAS is getting more and more complex, the dissemination ofinformation and the self-adaptation in Agent organisations are facing new challenges,which have influence the social harmony and stability.Such as the bad news spreading inthe internet often get out of control. Effectively mining the structure characteristics incomplex network play a very important role in the Information transfer and control on theinternet. Accessing local network structure is much more easier in the reality of the socialnetwork, And the local network structure is more useful to the local node. In this paper,we give a new concept called neighhoods-degree which is based in a two layer depthnetwork. Through the analysis, neighborhoods network can be applied in Internet’s information Transfer and control, self-adaptation in Agent organisations and network’srobustness.