A PARAFAC Decomposition Based Direct Tracking Algorithm for Wireless Sensor Networks

In this paper, we investigate a localization system for wireless sensor networks (WSNs) that effectively addresses localization errors and system failures caused by obstacle occlusion. First, we establish a time-delay-based Parallel Factor (PARAFAC) signal model for sensor networks. To reduce computational complexity and enhance real-time processing capability, the Randomized OnLine CANDECOMP/PARAFAC (ROLCP) method is adopted to achieve rapid decomposition of the time-delay matrix. Based on the estimated delay matrix, a localization cost function directly correlated with emitter location is formulated, and grid search is employed to determine the emitter position, enabling tracking of moving emitters. Furthermore, simulation experiments validate the efficacy of our proposed method, demonstrating its superiority in tracking accuracy and real-time performance.