Entropy-aware cluster-based object tracking for camera Wireless Sensor Networks

The paper presents entropy-based mechanisms to improve energy efficiency and robustness to transmission errors in camera-based object tracking systems. The main mechanisms are: an entropy-based algorithm that dynamically activates/deactivates nodes; a method that dynamically selects the cluster head using entropies and transmission error rates between the cluster nodes; and a distributed Extended Information Filter (EIF) that integrates measurements gathered within the cluster. These mechanisms have been integrated and implemented in a camera-based Wireless Sensor Network, being each camera node comprised of a TelosB WSN node connected to a CMUcam3 module. The proposed mechanisms and tracking system have been experimented and validated in the CONET Robot-WSN Integrated Testbed.