关键词: 传感器网络, 接收信号强度, 贝叶斯, 高斯拟合

Abstract: Most of the traditional positioning methods in wireless sensor network mainly study the matching relationship between the Received Signal Strength（RSS） and do not give full consideration to the physical environment and the use of off-line training database information. As to the above problem, this paper puts forward a new positioning method based on Gaussian fitting of deep similar domain. This method establishes a mapping relationship between distance of RSS in the offline phase, and uses the information of the node which to be located and its surrounding physical neighbor points in the domain of logic deep similarity. At the same time, it uses the methods of Bayesian algorithm and Gauss curve fitting to get the measured values of RSS. It can improve the accuracy of the fingerprint locationing method. Experimental results show that the method improves the precision of positioning effectively.

Key words: sensor network, received signal strength, Bayes, Gaussian fitting