Sensor Self-Organization for Mobile Multi-Target Tracking in Decentralized Wireless Sensor Networks

We propose the self-organization based sensor collaboration scheme for the Mobile Multi-Target Tracking (MMTT) in our developed Decentralized Wireless Sensor Networks (DWSN). Our developed DWSN has three-tier hierarchical structure. At each time step, the Cluster Heads (CH) having the more information on the tracked targets are activated. The activated CHs receive the measurement-sets from their own cluster members, and obtain the local multi-target estimates based on these measurement-sets. Using the local target-state estimates, the CHs quantize the measurement-sets, and share with other activated CHs. The CHs that have the sufficient detection capabilities with respect to the same tracked targets are included into the same group and collaborate with each other to achieve the final target-position estimation. Since only the quantized measurement-sets are communicated between sensor-nodes, we also propose the Decentralized Cardinality Balanced Multi-Bernoulli (DCBMeMBer) filtering algorithm, which is based on the adaptive Huffman-tree scheme and is implemented by the Sequential Monte Carlo (SMC) method. The obtained extensive simulation results validate and evaluate our proposed self-organization-based sensor collaboration scheme and our developed DCBMeMBer filtering algorithm.