Abstract: Currently wireless sensor network localization algorithms can hardly meet people’s requirements of positioning accuracy and computational complexity at the same time. A multidimensional Gaussian approximation based algorithm is proposed to deal with the dilemma. According to the received signal strength map, the algorithm establishes the fuzzy attribute matrix between the reference points and the unknown node, then sets an ideal point and calculates its attribute coordinates. After that, the distance relationship of the ideal point and the different reference points is estimated based on the multidimensional Gaussian model. Then an adjustment function is introduced to amend the position deviation between the ideal point and the unknown node. And the similar degree between different reference points and the unknown node is calculated for positioning. The algorithm has a simple operation process and the experimental results show that the accuracy of the positioning result is high and the error fluctuation range is small.

Key words: wireless sensor network, fingerprint localization, fuzzy attribute, multidimensional Gaussian model, similar degree

摘要： 目前无线传感器网络定位算法在定位精度、运算复杂度之间往往顾此失彼，针对该问题提出一种多维高斯近似指纹定位算法。该算法根据信号接收强度地图建立参考点与未知节点间的模糊属性矩阵，然后构造理想点并计算其属性坐标，基于多维高斯模型估测理想点与不同参考点间的距离关系，引入调节函数修正理想点和未知节点在位置上的偏差，进而计算不同参考点与未知节点的相似度进行定位。该算法的运算简单，且实验结果表明定位结果的准确性高，误差波动范围小。

关键词: 无线传感器网络, 指纹定位, 模糊属性, 多维高斯模型, 相似度