Multiple Target Tracking with RF Sensor Networks (1302.4720v1)

Abstract: RF sensor networks are wireless networks that can localize and track people (or targets) without needing them to carry or wear any electronic device. They use the change in the received signal strength (RSS) of the links due to the movements of people to infer their locations. In this paper, we consider real-time multiple target tracking with RF sensor networks. We perform radio tomographic imaging (RTI), which generates images of the change in the propagation field, as if they were frames of a video. Our RTI method uses RSS measurements on multiple frequency channels on each link, combining them with a fade level-based weighted average. We describe methods to adapt machine vision methods to the peculiarities of RTI to enable real time multiple target tracking. Several tests are performed in an open environment, a one-bedroom apartment, and a cluttered office environment. The results demonstrate that the system is capable of accurately tracking in real-time up to 4 targets in cluttered indoor environments, even when their trajectories intersect multiple times, without mis-estimating the number of targets found in the monitored area. The highest average tracking error measured in the tests is 0.45 m with two targets, 0.46 m with three targets, and 0.55 m with four targets.

• The paper presents a novel device-free localization method that uses RSS changes to generate RTI images for tracking multiple targets.
• It employs Gaussian filtering, hierarchical clustering, and Kalman filtering to achieve a tracking accuracy of 0.55 meters in complex indoor environments.
• The approach offers robust, real-time tracking without strict preconditions on target numbers, enhancing applications in surveillance and smart building systems.

An Analysis of Multiple Target Tracking with RF Sensor Networks

The paper "Multiple Target Tracking with RF Sensor Networks" explores a novel approach to device-free localization (DFL) through the use of radio frequency (RF) sensor networks. In contrast to traditional localization methods requiring devices or tags for tracking, RF sensor networks infer targets' positions by monitoring changes in received signal strength (RSS). The paper introduces advancements in RF-based multiple target tracking, specifically utilizing radio tomographic imaging (RTI) to create images analogous to video frames, enabling real-time tracking of multiple targets in diverse indoor environments.

Methodology Overview

RF sensor networks deployed in this paper capture RSS measurements across multiple frequency channels, which are processed using a fade level-based weighted averaging technique. This innovative approach considers channel diversity, allowing the system to leverage anti-fade channel measurements, which provide more localization-relevant data due to their stable RSS characteristics amidst human body obstructions.

RTI images are generated by interpreting RSS changes as a spatial integral of the propagation field, and these images undergo image processing akin to methods used in machine vision. Techniques like Gaussian filtering are employed to enhance image quality by reducing noise and isolating target-related signal variations. The tracking of multiple targets employs hierarchical agglomerative clustering, dynamic thresholding, and Kalman filtering to manage challenges posed by intersecting target trajectories and environmental multipath effects.

Experimental and Comparative Insights

The system's efficacy was validated across three varied environments: an open area, a cluttered office, and a furnished one-bedroom apartment. The experimentation involved up to four targets moving in intersecting paths, aiming to evaluate both tracking accuracy and system reliability. Integral to the research was the use of metrics such as the optimal mass transfer (OMAT) metric and the optimal subpattern assignment (OSPA) metric, particularly the latter's sensitivity to discrepancies between estimated and actual target counts.

Tracking accuracy exhibited a maximum average error of 0.55 meters with four targets in the office environment—a performance notably superior to prior works that often lacked real-time capabilities or assumptions made about target numbers and separation. When contrasted with other methods, such as particle filters or probabilistic data association, this approach demonstrated both efficiency in computation and robustness in environments with significant multipath interference.

Implications and Future Directions

This research signifies substantial progress in the field of indoor localization, indicating promising applications in surveillance, smart building systems, and assisted living environments. Its key strength lies in achieving real-time processing without cumbersome preconditions on the number or behavior of targets, a common limitation in existing systems.

The paper highlights areas for further exploration, including handling simultaneous entrances when individuals are side-by-side—a scenario complicating track association. Developing methodologies to maintain consistent tracking during such events without requiring participants to carry RF-emitting devices represents a significant future undertaking. Additionally, the integration of auxiliary sensors or algorithms, such as active badges or directionality-aware RF methods, could enhance system resilience against mis-associations during complex movement patterns.

In conclusion, RF-based multiple target tracking presents a compelling, non-invasive alternative to traditional indoor localization techniques. This advance in RF sensor networks opens avenues for refined, scalable tracking in environments where privacy and infrastructural simplicity are paramount. The practical applications of such systems are vast, holding transformative potential in fields requiring seamless human-computer interaction and situational awareness without the intrusive demands of device-based localization systems.