Research On Particle Swarm Optimization-based And Neural Network-based Nodes Positioning Algorithm For Wireless Sensor

Wireless Sensor Networks, which is ranked first among the ten emerging technologies that can change the future of world, has received more and more attention since the 90s of last century. U.S, EU, Japan and other developed countries have made plans in wireless sensor network. In recent years, China also attaches great importance to research on wireless sensor networks, November, 3, 2009, Premier Wen Jiabao instructed to make breakthroughs in sensor network and internet of things.Traditional positioning system, such as GPS, is important both in military affairs and civilian use while node positioning system, which provides protection for the collected information, is one of the key technologies in sensor network. In this paper, method for node localization and its improvement based on Particle Swarm Optimization (PSO) and method for node localization based on geometric constraints and intelligent search are proposed to further improve the accuracy and robustness of node positioning method. In addition, basic research on the application of Back—-propagation (BP) artificial neural network is made and the model of node localization based on BP neural network is put forward.The main contents of this paper are as follows:(1)Explore basic principles of node positioning , analyze classical methods of node positioning.(2)Depth research is made into the ranging technology based upon Links Quality Indicator. The data between LQI and distance is collected through a large number of experiments and the mapping relationship between LQI and distance is concluded through mathematical means, such as fitting. According to practical test, the actual ranging error of this method is less than 20%, which is better than the ranging method based on RSSI.(3)By studying basic principles of PSO, optimal solution to PSO search and practical application of node positioning are analyzed respectively, so the node positioning method based on PSO search is proposed. In order to verify the effectiveness of this method in node positioning, firstly, the search of unknown node is made under the circumstances of no ranging error, and stimulation result shows that whether the anchor node is put in special place or not, the precise positioning of unknown node can be achieved. Then the same search is made adding different ranging errors, stimulation result shows that positioning accuracy under 20%ranging error is 19.01% higher than that of least squares method, which indicates that the application of PSO search in node positioning is practicable.(4)To further improve the positioning accuracy of unknown node, after the analysis of the satisfying position that PSO has searched and constraint domain that anchor node has on the unknown node, node positioning method based on geometric constraints and intelligent search is proposed. Stimulation result shows that positioning accuracy under 20%ranging error is 30.33% higher than that of least squares method. Furthermore, this node positioning method can be applied not only in localization of static node but also in mobile node.(5)Preliminary exploration also has been made upon the application of BP neural network in node positioning, and positioning model based on BP network is finally established after many experiments and 160,000 trainings.In summary, this paper shows that after the introduction of PSO into node localization method, both positioning accuracy and robustness are improved.