Research On Multi-agent Networking For Information Detection

Information accessing is an important part of information processing. The higher theaccuracy of the data acquisition is, the better for promoting the effect of the system.High-precision data can be obtained by two ways. One is to use high-precision instruments,but the cost of the research and application may be too high and the technical bottlenecksare liable to emerge; the other tends to use plurality of low precision instruments and getintegrated information with fusing algorithms. It needs the management of variousinstruments.According to the actual need, the research background is based on the latter method.Each detection node is built as an agent. The technology of synergetic agent network isdiscussed. First, the agent model is constructed. Then, in order to make agent interact moreeasily, the task allocation and path planning is studied. Finally, a presentation is designed todemonstrate the systematic functions. The work accomplished is as follows:(1) The agents are expressed formally. The agents’ various properties and theirbehavior are defined with formal representation. According to the agent hierarchy in thesystem, agents are constructed with different structure; Agent communication language isconcerned for agents’ better cooperating. Communication primitives are extended to adaptto our system.(2) The widely-used contract net protocol also has shortcomings. For example thecommunication scale may be too large and it may not be so efficient in dynamicenvironment. Thus, an improved contract net protocol is proposed, by introducing loadbalancing degree and trust degree to limit the scope where task was announced and tochange agent’s evaluation strategy. The proposed algorithm improved the task-allocationefficiency;(3) Multi-Agent collaborative algorithms are designed and implemented. With theartificial potential field method for path planning, the executive agent gains the best pathplanning; in order to make agents have optimized actions, learning methods are discussed.(4) The netlogo platform for information detection based on multi-agent modeling iscompleted. Firstly the agent environment is built. Then a variety of agents are added to theenvironment. Each agent has its attributes and functions. After the multi-agent system isinitialized, tasks and agents can be manually added to the system. The simulation experiment verifies the effectiveness of the proposed method. Finally,the paper analyzes the shortcomings in this study, and points out further research directions.