Welcome to

Lancashire Online Knowledge

Image Credit Header image: Artwork by Professor Lubaina Himid, CBE. Photo: @Denise Swanson


Decimetre-Level Single- and Multi-Target 3D mmWave Sensing Using Mechanical Sensor Steering and Kalman Filtering

Raspopoulos, Marios orcid iconORCID: 0000-0003-1513-6018, Sesyuk, Andrey orcid iconORCID: 0000-0002-1908-7850 and Ioannou, Iacovos (2026) Decimetre-Level Single- and Multi-Target 3D mmWave Sensing Using Mechanical Sensor Steering and Kalman Filtering. In: 2026 International Conference on Indoor Positioning and Indoor Navigation (IPIN), 5-8 October 2026, Rome, Italy.

[thumbnail of AAM]
Preview
PDF (AAM) - Accepted Version
Available under License Creative Commons Attribution.

1MB

Official URL: https://doi.org/10.1109/IPIN72673.2026.11772549

Abstract

Millimeter-wave (mmWave) sensing has emerged as a promising solution for accurate indoor positioning due to its ability to provide range, azimuth, and elevation measurements. While prior work has demonstrated decimetre-level accuracy for single-target localization using fixed sensor deployments, performance degrades in dynamic and multi-target scenarios due to limited sensing coverage and challenges in measurement association. This paper presents a unified framework for single- and multi-target 3D positioning based on adaptive sensor steering, robust measurement association, and Kalman filtering. The proposed approach dynamically orients sensors towards predicted target locations, improving measurement geometry and visibility. For multi-target tracking, a prediction-guided gating and association strategy is introduced, enabling reliable separation of closely spaced targets without relying solely on clustering techniques. The proposed framework is evaluated in a realistic simulation environment that incorporates experimentally derived mmWave error models. Results demonstrate that sensor steering provides measurable improvements in accuracy, while the integration of Kalman filtering significantly enhances robustness. Overall, the proposed method achieves consistent decimetre-level accuracy, with approximately 25% reduction in RMSE compared to fixed sensor configurations, and improved resilience in dense multi-target scenarios.


Repository Staff Only: item control page