无线传感器网络中基于BP-UKF的室内定位算法

Indoor Localization Algorithm Based on BP-UKF for Wireless Sensor Network

针对复杂室内环境中的定位噪声问题,文中提出一种基于BP-UKF的室内定位算法。利用BP神经网络拟合任意非线性函数的特性训练样本集,以确定接收信号强度指示(RSSI)和距离之间非线性关系,进而估计待定位节点与锚节点间的距离;采用三边定位法计算待定位节点的坐标初始值,并利用无迹卡尔曼滤波(UKF)处理非线性系统方程;通过设距离的估计值为观测变量,对待定位节点坐标的初值进行精确化处理。仿真试验结果表明,对于较复杂环境的室内定位,与传统的定位算法相比较,文中所提BP-UKF算法实现达到更可靠和准确的位置估计。

In order to solve the localization problem at the complex interior environment,indoor localization algorithm based on BP-UKF algorithm is proposed. This algorithm uses the BP neural network algorithm to obtain the relationship between the distance and Received Signal Strength Indication（RSSI） accurately by use of the characteristic in fitting any no-linear function. Then the distances between the unknown nodes and anchor nodes can be calculated.The coordinates of the nodes are initialized by using the trilateration with those distances.It uses the unscented kalman filter algorithm to deal non-linear system equation.The initial value of the coordinates of the positioning node is treated accurately by the estimated value of the distance as the observation variable. The result of simulation shows that BP-UKF algorithm can achieves reliable and accurate solution in interior environment than the traditional positioning algorithm.