Drone localisation and tracking using a distributed FMCW MIMO radar network
Authors
Andersen, Rasmus Himmelstrup ; Best, Aske Barnaby Rasborg ; Nielsen, Malthe Raschke
Term
4. semester
Education
Publication year
2025
Submitted on
2025-06-03
Pages
82
Abstract
Small drones are becoming more common and bring both opportunities and risks, so reliable ways to detect and track them are needed. This thesis examines whether the MRBLaT algorithm (multiple radar Bayesian localisation and tracking) can run on real measurements from distributed, off-the-shelf FMCW MIMO radar systems. We develop a detailed signal model for FMCW radar and build the Python-based TMARS (Tracking and Mimo radAR Simulation) tool for simulation and analysis. MRBLaT is integrated into TMARS and evaluated in both simulations and field experiments. Measurements are collected with three synchronized Texas Instruments AWR1843BOOST radar modules placed apart and working together. The results show that MRBLaT can track a drone with high accuracy under real conditions, even when the signal is weak relative to noise (low SNR). We also highlight practical hurdles: multipath fading (signals bouncing off surfaces), clutter (unwanted echoes from other objects), synchronization across radars, and hard, non-convex optimization. Overall, the study validates MRBLaT as a promising approach to drone tracking with distributed radars and lays the groundwork for future improvements.
Små droner bliver stadig mere udbredte og giver både muligheder og risici, så der er brug for pålidelige systemer til at opdage og følge dem. Denne afhandling undersøger, om MRBLaT-algoritmen (multiple radar Bayesian localisation and tracking) kan bruges på virkelige målinger fra distribuerede, kommercielt tilgængelige FMCW MIMO-radarsystemer. Vi udvikler en detaljeret signalmodel for FMCW-radar og bygger det Python-baserede værktøj TMARS (Tracking and Mimo radAR Simulation) til simulering og analyse. MRBLaT integreres i TMARS og vurderes både i simulationer og i forsøg i felten. Data indsamles med tre synkroniserede Texas Instruments AWR1843BOOST-radarmoduler, placeret adskilt og arbejdende sammen. Resultaterne viser, at MRBLaT kan spore en drone med høj nøjagtighed under virkelige forhold, selv når signalet er svagt i forhold til støjen (lavt SNR). Samtidig peger vi på praktiske udfordringer: flervejsudfading (signaler, der reflekteres fra omgivelser), uønskede ekkoer fra andre objekter (clutter), synkronisering mellem radarer og svære, ikke-konvekse optimeringsproblemer. Samlet set bekræfter arbejdet, at MRBLaT er et lovende bud på dronesporing med distribuerede radarer og lægger grunden til fremtidige forbedringer.
[This apstract has been rewritten with the help of AI based on the project's original abstract]
Keywords
Other projects by the authors
Best, Aske Barnaby Rasborg: