Research And Design Of Localization System Based On Kalman Filter For WSNs

It is now the direction of intelligent knowledge to let every corner of the world have a perception. As a new data collection and information transfer technology, wireless sensor networks have a huge potential on all aspects of life. Position sensors have become important and foundational information for the service experience. Classical localization algorithms are divided into two categories:range-based measure and non-range measure. Inevitably, the sensors in certain areas are so scarce that people have to use range-based measure localization algorithm. Obviously, the accuracy of the range-based measure directly affects the localization precision The fluctuation of the RSSI values can affect range measurement a lot if one uses the RSSI-based range measure method. Therefore, the thesis focuses on the problems of how to improve the localization precision from the collected noise RSSI values by filtering methods.Firstly, the thesis made a brief investigation on the research background of the wireless sensor networks. After that, the principles of localization, the range-based measure technology and coordination calculation algorithm in wireless sensor networks together with their advantages and disadvantages are analyzed. The key factors of localization algorithm are summarized, which should be careful considered when designing. Based on the real data gathered in devices, the thesis analyzed the noise features of measuring error and presented the Kalman filter method to smooth the data. It is found the last few samples have greater impacts on the final localization precision through simulation. In order to eliminate this kind of interference, the thesis further proposed to use the median filter after using the Kalman filter. The feasibility and improvement are testified through comparing the simulation results with that of the means algorithm and the median algorithm. The results show that the Kalman-Median algorithm can smooth fluctuations RSSI values very well and can get good stability. In order to simplify the algorithm and adapt the environment changes, a localization method which don’t require to compute A and n is presented, which are the key parameters when calculating the distance between reference nodes and the unknown nodes.Then, based on the algorithm presented above, a localization prototype WSN system is constructed, in which the sensor nodes are designed by the chip of CC2530. In the system, the Kalman-Median algorithm is implemented in a prototype system for localization Experimental results show that the fixed parameters in the signal attenuation model will cause huge measurement errors and the huge localization errors. To eradicate the influence of hardware environmental and other factors, a feedback idea is carried in the system, which eliminates the errors caused by unreasonable attenuation model parameters. Combined with the Kalman-Median algorithm, we effectively reduce localization errors and solve the beat phenomenon in the static localization algorithm.Finally, we summarized the entire thesis and discussed the restrictions of this system. Also, the future research directions and a general plan are given.