中文摘要：

为了实现基于无线传感器网络的高精度目标跟踪,提出一种对物理层透明的基于信号特征序列的粒子滤波跟踪算法.采用信号特征序列实现由物理层信号到抽象信号的转变,利用信号特征序列的距离作为粒子滤波权值的更新依据;针对经典粒子滤波算法容易退化、贫化、粒子偏离后验分布等问题,综合运用轨迹预测、遗传交叉、重采样等方法使粒子分布有效的表征目标后验分布;同时,采用多维滤波对数据进行处理改善算法的噪声抑制能力.仿真结果表明算法的各种措施有效地提升了跟踪精度.

英文摘要：

In order to realize accurate tracking task with wireless sensor network,a signal character sequence particle filter algorithm is proposed which is generic and compatible with different physical layers.The signal character sequences are used to translate the signal from the physical layer to the abstract layer and realize the update of the particle weights.To overcome the degeneration and impoverishment problems of the classical particle filter algorithm,trace prediction,cross operation and resample method are proposed to sample particles from the posterior distribution.In addition to the basic algorithm,multidimensional filtering is also adopted to deal with the noise.Simulation results show the effectiveness of the improvement measures of the algorithm.