Research On Wireless Sensor Network Location Algorithm For Forest Fire Monitoring

Forest fire detection has important ecological environment significance and social economy significance. Advanced technologies are utilized to obtain and describe the forest environment variable information quickly and effectively. These technologies can realize the automatic detection of forest fire, which is the precondition and guarantee for the forest fire prevention.Wireless sensor network is a kind of integrated intelligent information processing system, including sensor technology, wireless communication technology, micro-electromechanical technology, and modern network technology. It has a very wide range of application, and its further development and application will deeply affect many areas of the human society. Due to the wireless sensor nodes generally have random deployments, the position of the majority of nodes in the network cannot be determined in advance. Howere, many applications of the wireless sensor technology require the nodes position information to determine the exact location of information. Therefore, the node localization is one of the key technologies for wireless sensor network applications. This dissertation is focused on using wireless sensor network technology in the forest fire detection, and investigates the node localization algorithms to improve the accuracy. The main work is summarized as follows:At the beginning of this dissertation, the present research situation, tendency and the main problems for the node localization technology in wireless sensor network is firstly analyzed. The present situation for wireless sensor network is expatiated, and the main application foundation and the typical algorithms of the node localization technology are introduced. Combining with the characteristics of application technology of the forest fire detection, this dissertation researches the range-free positioning technology (DV-Hop algorithm) in wireless sensor network, and analyses the advantages and disadvantages of the DV-Hop localization algorithm, as well as the main factors affecting the localization accuracy. Among them, an anchor node is an extremely important factor in the localization algorithms, as anchor nodes are required to realize localization in every localization algorithm. The effect of anchor node upon the location accuracy is embodied in the number of anchor nodes.Secondly, the typical DV-Hop algorithm is further improved for the range-free localization techniques. The DV-Hop algorithm can get more reasonable location accuracy in isotropic density sensor network. However, its location errors are restively large in randomly distributed networks. In this dissertation, entropy is proposed in the estimation of the average single hop distance. This improvement can be used to replace the corresponding steps in the DV-Hop algorithm. The theoretic and simulation results show that the proposed improvements can greatly enhance the localization accuracy of the unknown nodes. In addition, the proposed algorithm do not change the localization process of the DV-Hop algorithm, and it requires no further communication task and additional hardware requirement in wireless sensor network.Thirdly, this dissertation proposes two method including the estimation of average single hop distance and the estimation of node positions based on grey correlation analysis and mix-max method. In anisotropy network, grey correlation analysis is applied in the DV-Hop algorithm. The location accuracy of the DV-Hop algorithm is improved with stronger adaptability. Considering the problems of heavy computational complexity, low precision, and algorithm invalidation caused by methods of the maximum likelihood and the trilateration in the DV-Hop algorithm, mix-max method is adopted to computer node position. Simulation results show that the proposed method has features of a small amount of calculation and high location accuracy under the same network parameters.Fourthly, in this paper, entropy and grey correlation analysis is applied to the DV-Hop algorithm. The applicability and location accuracy of the method are improved. Compared with the typical DV-Hop algorithm, the proposed algorithm has low computational complexity and good localization accuracy. All the localization methods proposed in this dissertation have been verified through simulation experiments on Matlab platform. The corresponding performance analysis is also investigated.Finally, the forest fire detection system is presented based on the proposed localozation algorithms for wireless sensor network. The main research results are verified by the simulation of the system.In summary, through the verified research results, the suitable and reliable location theory is provided for wireless sensor network in the forest fire detection. The respond speed of the forest fire prevention and fire information feedback are improved in the application.