Energy and bandwidth efficient multi-target tracking in a two-tier hierarchical wireless sensor network

Abstract

Target tracking is an important application of wireless sensor networks (WSN). In order to deal with sensor nodes with limited energy supply and communication bandwidth, two-tier hierarchical WSNs are proposed for solving the problems of single-target tracking without clutter and multi-target tracking in cluttered environment respectively. To handle the first problem, the Base Station (BS) selects one active cluster with highest expected information gain at each time step and the Cluster Head (CH) of the active cluster selects combination of several active sensor nodes within its own cluster to achieve lowest energy consumption. The active sensor nodes send mw measurements to their cluster head, which processes the information locally and estimates the target state using Auxiliary Particle Filtering (APF) method. The efficiency of this proposed two-tier hierarchical WSN is demonstrated in the application of signal strength tracking. To address the Multi-Target Tracking (MTI) in cluttered environment, the base station selects multiple active clusters of sensors which have a better view of the targets at each time step and determines their order for the sequential data fusion in the second level of hierarchy. Within each active cluster, multiple sensor nodes are activated, and transmit the measurement-sets to their cluster head. The cluster head only fuses some of the measurement-sets with higher information gain, and a Particle Probability Hypothesis Density (PHD) filtering method is performed for estimating the number of targets and their states. The proposed two-tier hierarchical WSN works well for multi-target tracking in cluttered environment even when target dynamics and/or measurement process are severely non-linear. Since the PHD filter keeps no record of target identities, it does not provide the temporal association of estimates over time. To address this issue, a data association scheme is developed to produce track valued estimates of individual targets. This scheme is computationally efficient and works well even when target dynamics and/or measurement process are severely non-linear. The performance of the proposed two-tier hierarchical WSN in multi-target tracking is demonstrated in the application of bearing and signal strength tracking.