Random Finite Set Approach to Signal Strength Based Passive Localization and Tracking

Ossi Kaltiokallio, Hüseyin Yigitler, Jukka Talvitie, Mikko Valkama

Abstract: Abstract—Radio frequency sensor networks can be utilized for locating and tracking people within coverage area of the network. The technology is based on the fact that humans alter properties of the wireless propagation channel which is observed in the channel estimates, enabling tracking without requiring people to carry any sensor, tag or device. Considerable efforts have been made to model the human induced perturbations to the channel and develop flexible models that adapt to the unique propagation environment to which the network is deployed in. This paper proposes a noteworthy conceptual shift in the design of passive localization and tracking systems as the focus is shifted from channel modeling to filter design. We approach the problem using random finite set theory enabling us to model detections, missed detections, false alarms and unknown data association in a rigorous manner. The Bayesian filtering recursion applied with random finite sets is presented and a computationally tractable Gaussian sum filter is developed. The development efforts of the paper are validated using experimental data and the results imply that the proposed approach can decrease the tracking error up to 48% with respect to a benchmark solution. Index Terms—Received signal strength, RF sensor network, localization and tracking, random finite set, Gaussian sum filter
Published in: 2023 IEEE/ION Position, Location and Navigation Symposium (PLANS)
April 24 - 27, 2023
Hyatt Regency Hotel
Monterey, CA
Pages: 1215 - 1225
Cite this article: Kaltiokallio, Ossi, Yigitler, Hüseyin, Talvitie, Jukka, Valkama, Mikko, "Random Finite Set Approach to Signal Strength Based Passive Localization and Tracking," 2023 IEEE/ION Position, Location and Navigation Symposium (PLANS), Monterey, CA, April 2023, pp. 1215-1225. https://doi.org/10.1109/PLANS53410.2023.10140040
Full Paper: ION Members/Non-Members: 1 Download Credit
Sign In