粒子滤波实现无线传感器网络目标跟踪预测

Realization of Target Tracking Prediction in WSN Using Particle Filter

为减少无线传感器网络(WSN)目标跟踪预测误差,提出一种粒子滤波实现WSN目标跟踪预测方法;该方法采用粒子滤波获得目标运动状态,联合当前时刻目标的本地估计位置、预测速度和加速度获得下一时刻目标预测位置,预测位置可作为当前头节点唤醒所述下一时刻传感器节点的依据;结果表明,上述粒子滤波预测方法预测准确度相比线性预测方法明显提高,均方根误差RMSE减少49%;相比基于二次多项式运动建模的WSN目标跟踪预测方法,均方根误差RMSE减少6%。

Aiming at reducing then target tracking prediction error in wireless sensor network, a novel method of realizing target tracking prediction in WSN using particle filter is proposed. According to the method, target state can be obtained by particle filter. The local estimation position, the predicting speed and acceleration of current time instant are combined to achieve the predicting position of next time instant, which is used for the current head node to awake sensor node of next time instant. Experimental results show that the general prediction accuracy of particle filter method is improved highly compared with linear prediction method while RMSE decreased by 49%, and compared with target tracking prediction in WSN based on quadratic polynomial motion modeling method （PQPMM） while RMSE decreased by 6 %.