Identification and Localization on a Wireless Magnetic Sensor Network

This thesis presents the design and implementation of an Identification, Localization, and Sequential localization (ILS) system based on a Wireless Magnetic Sensor Network (WMSN). The system leverages magnetic sensors (Honeywell HMC1002) mounted on MICAz motes to detect, identify, and track small ferromagnetic objects within a confined sensing area.

Unlike prior work focused on detecting large metal objects (e.g., vehicles, aircraft), this research targets small-scale targets such as iron bars, and investigates the limitations of magnetic sensing in close-range (within 30 cm). The system uses a sparse deployment strategy to minimize energy consumption and cost, while still achieving accurate sensing.

A testbed with 9 sensors for tracking a target in discrete movement

Key contributions and components include:

A centralized tree-based WSN with MTS310CB sensor boards.
Development of techniques to calibrate sensor sensitivity, determine sensor axis orientation, and mitigate environmental effects like temperature, sunlight, noise, and power fluctuations.
Application of the Minimum Euclidean Distance (MED) algorithm to identify and locate targets based on magnetic signature differences.
Use of a dictionary-based approach for both identification and localization tasks, supported by MATLAB-based visualization tools.
Real-time tracking of moving objects through discrete and continuous movement experiments.
Proposal of Orthogonal Matching Pursuit (OMP) as a future method for handling multi-target and multi-sensor scenarios.

The thesis also introduces practical improvements such as threshold and counter checking to enhance decision reliability, and discusses future extensions including energy harvesting, custom communication protocols, and Kalman filtering for robust tracking.

Identification and localization on a wireless magnetic sensor network

For more information, please click here!