【摘要】 针对无线传感器网络目标定位过程中存在的测量矩阵维数高、运算量大以及测量向量不准确的问题，文章提出一种优化压缩感知(compressive sensing, CS)定位算法。首先对传感器采集的接收信号强度值进行卡尔曼滤波保证压缩感知测量向量的准确性；然后引入bounding-box方法估计位置区域，缩小节点定位范围从而降低后一阶段压缩感知测量矩阵的维数；最后在节点估计区域进行压缩感知定位，并提出基于原子相关度阈值的回溯匹配追踪算法，通过原子相关度阈值控制对候选集原子进行二次筛选剔除低相关度原子，在支撑集中保留系数较大的原子，提升重建精度。实验结果表明，在信噪比为5 dB,目标数为8时，相较于传统的OMP算法、GMP算法、CoSaMP算法，所提优化定位算法的定位精度分别提升61.21%、51.53%和45.12%。更多还原

【Abstract】 An optimized compressive sensing(CS) localization algorithm is proposed to address the problems of high dimensionality of measurement matrix, large computation and inaccurate measurement vectors in the target localization process of wireless sensor networks(WSN). Firstly, Kalman filtering is applied to the measured receiving signal strength indication to ensure the accuracy of the CS measurement vector. Secondly, a bounding-box algorithm is introduced to estimate the location area, narrowing the node localization range and thus reducing the dimensionality of the CS measurement matrix in the later stage. Finally, the CS localization is performed in the node estimation region, and a backtracking matching pursuit algorithm based on the atomic correlation threshold is proposed to eliminate the low correlation atoms by secondary screening of the atoms in the candidate set through the atomic correlation threshold control, and retain the atoms with larger coefficients in the support set to improve the reconstruction accuracy. The experimental results show that the localization accuracy of the proposed optimized localization algorithm is improved by 61.21%, 51.53% and 45.12%, respectively, compared with the traditional orthogonal matching pursuit(OMP) algorithm, greedy matching pursuit(GMP) algorithm and compressive sampling matching pursuit(CoSaMP) algorithm when the signal-to-noise ratio is 5 dB and the target number is 8.更多还原