Research On Optimization And Fault Diagnosis Of Wireless Sensor Networks Based On Artificial Immune System

Recently, wireless sensor networks have been a new technology for information acquisition andhave wide potential applications in military, defense, environment monitoring and so on. Because oflittle size and monitoring distribution, wireless sensor networks include the features: the nodes carrylimited energy, the topology is dynamic and the fault is easily caused by communication interruption.For the applications of wireless sensor networks, however, it is important how to reduce the energyconsumption of node, optimize the performance of network and diagnose the faults. The naturalimmune system is a parallel and self-adaptive system and has characters of diversity, dynamics,self-adaptability, study, memory and so on. As a new branch of artificial intelligence, artificialimmune system generated by imitating the natural immune system has strong ability to processinformation and solve complex problems. Based on the introduction to the natural immune system,immune mechanism, artificial immune system and its applications status, the approaches inspired byartificial immune algorithm are presented for performance optimization and fault diagnosis ofwireless sensor networks. Contributions are summarized as follows.(1) It is summarized about research background and significance of wireless sensor networks foroptimization and fault diagnosis techniques. And the research direction is confirmed;(2) Inspired by Markon evolution process and self/non-self discrimination mechanism of naturalimmune system, the mapping relation between natural immune system and optimization as well asfault diagnosis for wireless sensor networks is built. An optimization and fault diagnosis immunemodel in wireless sensor networks are designed, respectively.(3) On the base of immune stimulation mechanism, an optimization approach for the nodes ofwireless sensor networks is presented. The similar relationship between artificial immune system andwireless sensor networks is built by imitating the B-cell and T-cell respond mechanism of naturalimmune system. According to spatial-temporal correlation and adaptive least mean square filter, themathematics model between distortion and number of communication nodes as well as distortion andreporting frequency are constructed. An adaptive adjustment algorithm based on immune stimulationmechanism is proposed to achieve the minimum number and suitable reporting frequency ofcommunication nodes. The location of communication nodes in the monitoring area is decided byvector quantization technology. The simulation result in different scenes shows that the approach hasthe capability of improving node redundancy and saving network energy consumption.(4) According to immune clone and selection, a routing optimization algorithm for wireless sensor networks is proposed. Combined with multicast theory and the global optimization performance ofartificial immune system, the relationship between routing optimization and artificial immune systemis built. The regulation of routing optimization algorithm is presented such as antibody expression,clone multiplication, clone selection and gene maturation. The procedure and step of algorithm isdesigned. Then the algorithm characteristic is introduced from the respects of calculation method,premature convergence prevention as well as robustness. The simulation result shows that theapproach can save network energy and improve routing protocol robustness in wireless sensornetworks.(5) Based on immune collaboration mechanism, a multi-agent fault diagnosis model and itsimplement method for wireless sensor networks is proposed. The functions of the agents and theircollaboration mechanisms in the model are analyzed. The agents constructed by immune algorithmare emphatically presented. The performance of immune algorithm is analyzed from the respects ofantibody affinity, convergence as well as adaptivity. The simulation proves that the model can meetthe self-adaptive diagnosis demand of changeable and dynamic wireless sensor networks.（6）A fault diagnosis monitoring software for wireless sensor networks is designed. Thecomponent, function and implementation step of software are analyzed. The validation is also verifiedby using simulation results.（7）Summary of the whole works is drawn. The problems meted in the research and the continuingdirections are presented.