The Research And Applications Of Data Fusion And Target Tracking In Wireless Sensor Networks

Wireless Sensor Network (Wireless Sensor Network, WSN) currently is a cutting-edge hot research field, it be concerned by international area, involved by cross-cutting multi-disciplinary and high integrated by knowledge. Wireless sensor network is make up by a large number of cheap sensor nodes which be deployed in the region, these sensor node is self-organized wireless sensor network by the formation of wireless communication. The aims of these sensor node is perception and understanding the phenomenon in the network coverage area by the method of collaborate perception, acquisition and processing phenomenon data, and send it to the users of wireless sensor network. At the same time, wireless sensor network has the advantages of low-cost, small volume, convenient and flexible networking, it cab be used widely in national security, environmental monitoring, traffic management, medical care, manufacturing, anti-terrorism and other disaster areas. It has been widespread concern around the world because of the important research value and widely application prospects, be recognized as a important technology will have a great impact on the 21st century.Focusing on the key technologies of wireless sensor networks, sensor noise processing, network data fusion, as well as locating and tracking methods and technology is involved in this article, following the design requirements of energy priority, data centric in the wireless sensor network, Aiming at these technology, this article has do an intensive research on it, and carrying on detailed analysis on the issues involved in the these area, proposed corresponding resolution control algorithms, make foundation of future further research and application. As a innovative work, methods and ideas of research seeks to achieve a breakthrough in the article, the main research achievements include the following aspects:The fundamental task of wireless sensor network is accurate to obtain valuable information in the physical world, due to environmental impacts, noise in the measurement signals; it will affect the accuracy of nodes measurement results, and even affect the entire network results. To solve these problems, a lightweight, improved Kalman algorithm for signal processing of sensor signals is proposed in this paper, the traditional Kalman filter need a precise measurement noise variance values and the process noise variance values be pre-establish, these values are generally get by empirical analysis to be achieved or get its approximation by the probability. In order to overcome the limitations of traditional Kalman filter algorithm, a "sliding window" is used in this paper to estimate the measurement noise variance value, achieve the Kalman filter real-time estimation for the sampling data. Not only Kalman parameters can be adjusted adaptively,but also the noise signal which bring from sensor itself and the sampling circuit can be reduced .by using this method.In the applications of wireless sensor network, the network is made up by a large number of sensor nodes; and an enormous data flow is produced by the nodes. The contradiction between the limited bandwidth and the enormous data transmission will be resolved by using a accommodate data processing algorithms. The method of data fusion is discussed in this paper, and a algorithm of adaptive weighted data fusion is proposed, adaptive optimal weighted value is produced based on measurement noise variance estimated and constructed a judgment matrix, the sensor data is adaptive, fast, reasonable grouping weighted, a robust and exact results can be achieved. The method is a simple, practical algorithm, and greatly reduces the redundancy of data within the network, saving a lot of storage resources and network bandwidth.Usually, location information is a part of the sensor data in the wireless sensor networks, the location information is contained monitoring information, it is a key features of the monitor node. The basic principle of node localization of wireless sensor network is discussed in this article, a new node fast location algorithm based on anchor node clustering is proposed (localization by anchor clustering, LAC). nodes ranging technical and multi-dimensional scaling analysis is combined in LAC algorithm, the sensor network is divided into many clusters by anchor node as the center of cluster, the fake-anchor node is established by multilateral position algorithm. The measurement error is reduced and the precision of the coordinate reconstruction is improved by using anchor clustering; the absolute coordinate's position is realized intelligently by using the combined the fake-anchor and anchor position; positioning complexity is reduced and computational efficiency is improved, by distributed positioning, the overall network scalability and robustness is improved, simultaneously.Focused the application of the moving object tracking in the wireless sensor network, a algorithm of target tracking based on the wireless sensor network is discussed. based on connectivity of the network and continuity of the movement objects, and combined the law of motion prediction with particle filter algorithm, a algorithm of parallel collaborative particle filter is presented for moving object tracking in wireless sensor network, the algorithm is implemented by multiple particle filter running in the every sensing nodes, local estimations are formed in these nodes, information fusion is implemented in the cluster head, the final state estimation is formed. This algorithm have the advantages of high precision, fast response and a balanced energy consumption, it is a better solution for the moving object tracking in the wireless sensor network. and experimental simulation results show that correctness, validity and accuracy of the algorithm.As an application of wireless sensor network, a gas monitoring system is designed and development, the algorithms is simply inspected and verified by using the experiments of real-time gas concentration monitoring and gas source moving spread tracking. Experiments showed that the coverage of space and time is expanded and time and the resolution and coverage of space in the monitoring system is improved by using this algorithms are proposed in the article.