Cluster-based Position Tracking of Mobile Sensors

TL;DR

This paper proposes a cluster-based cooperative position tracking algorithm for mobile sensors that reduces energy consumption and improves accuracy by central coordination and using Kalman filters, especially effective in periodic GPS sampling.

Contribution

It introduces a novel cluster-based cooperative tracking method with Kalman filtering, optimizing energy and accuracy tradeoffs for resource-constrained mobile nodes.

Findings

Over 25% energy savings in periodic sampling

More than 33% improvement in position accuracy

Comparable results in dynamic sampling with existing schemes

Abstract

Tracking movement of mobile nodes has received significant scientific and commercial interest, but long term tracking of resource-constrained mobile nodes remains challenging due to the high energy consumption of satellite receivers. Cooperative position tracking has been proposed for energy efficiency, however, all the cooperative schemes use opportunistic cooperation and optimize for either energy or accuracy. Considering the existence of a reasonably stable group of mobile nodes like animals, birds, and mobile assets, we propose a cluster-based cooperative tracking algorithm, where cluster head centrally coordinates resource usage among cluster members. Variants of this strategy include the use of a cooperative Kalman filter with and without inertial sensor inputs to estimate nodes positions. We use the Boid flocking algorithm to generate group position movements in 3D and perform various experiments to compare the energy and position accuracy tradeoff of our proposed scheme with individual-based tracking and existing cooperative schemes. We perform the experiments for fixed periodic GPS sampling and dynamic GPS sampling triggered by node position error uncertainty tolerance limit. Experiments results show that in periodic sampling scheme cooperative tracking yields more than one-quarter reduction in energy consumption and more than one-third improvement in position accuracy over individual-based tracking, however, results for dynamic sampling scheme are comparable with existing cooperative scheme.