Efficient Cluster-Based Tracking Mechanisms for Camera-Based Wireless Sensor Networks

Abstract

This paper proposes mechanisms to efficiently address critical tasks in the operation of cluster-based target tracking, namely: (1) measurement integration, (2) inclusion/exclusion in the cluster, and (3) cluster head rotation. They all employ distributed probabilistic tools designed to take into account wireless camera networks (WCNs) capabilities and constraints. They use efficient and distribution-friendly representations and metrics in which each node contributes to the computation in each mechanism without requiring any prior knowledge of the rest of the nodes. These mechanisms are integrated in two different distributed schemes so that they can be implemented in constant time regardless of the cluster size. Their experimental validation showed that the proposed mechanisms and schemes significantly reduce energy consumption (>55 percent) and computational burden with respect to existing methods.

An Effective and Robust Decentralized Target Tracking Scheme in Wireless Camera Sensor Networks

“…

M1. The CH activates or deactivates a camera analyzing the usefulness of its measurements and the resources as given in [23], which also adopts an on-line decision making approach that maximizes the trade-off between sensing gain and resource consumption. This work considers the remaining energy of each node instead of the relative energy in current candidate cluster to weight the resource consumption.

…”

“…Sensor nodes in the alert state could receive data from their neighbours and periodically acquire and process data. Meanwhile, sensor nodes spend most of their time in sleep state during which nodes only periodically sense the target, which consumes the least energy among the three states [23]. …”

“…Different metrics have been proposed to gauge the tracking performance of the information filter [7,14,22,23]. Among them, the trace (sum of diagonal elements) of the predicted information matrix $Y_{k+1}$ computed at timestep k using Equation (33) corresponds to the mean squared error (MSE) of the updated state.…”

“…In traditional schemes, the absolute quantity of energy usually conducts as an indicate of β (like in work [23]). However, in this work, we set ${\beta }_{i,k+1}$ dynamically according to the current remaining energy of itself and other candidate nodes.…”

“…Nevertheless, entropy is a nonlinear metric and its computation is inappropriate for a decentralized approach. In [23], novel camera activation and CH selection mechanisms that consider transmission errors and use the trace of information matrix as uncertainty metric are developed, where all camera nodes that their rewards overtake the cost will be activated to form the tracking cluster. However, both of the above works assume an omnidirectional sensing model, i.e., the target can be viewed within a circular area whose radius equals to the sensing range of camera nodes.…”