Hypothesis Testing based Interference Mitigation in mmWave OFDM Radar
Author
Randers, Laurits
Term
4. semester
Education
Publication year
2025
Pages
43
Abstract
This thesis investigates how to mitigate interference in millimeter-wave OFDM automotive radar without coordination between radars. Motivated by increasing spectral overlap in the 77–81 GHz band, it introduces an OFDM sensing approach that uses only a subset of subcarriers. A channel model is developed for a radar scene with both targets and interferers, and an algorithm is proposed that combines directional beamforming and angle-of-arrival estimation with two hypothesis tests: first, to distinguish line-of-sight (LOS) paths from interferers versus backscattered target returns, and second, to detect which subcarriers are affected by interference. Target range and path gain are then estimated using only the non-interfered subcarriers. Numerical experiments indicate that, with a priori angle-of-arrival information, interference mitigation can improve range estimation performance by approximately 5–7 dB. Since the algorithm has not been implemented as a complete system, the overall impact on end-to-end sensing performance cannot yet be stated definitively. The study assumes stationary targets and interferers and focuses on validating hypothesis test-based interference detection in OFDM radar.
Dette speciale undersøger, hvordan interferens i millimeterbølgebaserede OFDM-radarsystemer til biler kan afbødes uden koordinering mellem radarenheder. Ud fra et motiv om stigende spektralt overlap i 77–81 GHz-området præsenteres et OFDM-baseret seneprincip, der kun udnytter et delmængde af delbærere. Der opstilles en kanalmodel for et radarmiljø med både mål og interfererende signaler, og der foreslås en algoritme, der kombinerer retningsvirkende beamforming og vinkel-estimation med to hypotesetest: først til at skelne line-of-sight (LOS) fra interferenter kontra tilbagespredte målsignaler, og dernæst til at identificere hvilke delbærere der er ramt af interferens. Rækkevidde og stibevinst estimeres efterfølgende kun fra de ikke-interfererede delbærere. Numeriske eksperimenter indikerer, at hvis indfaldsvinklen er kendt a priori, kan interferensmitigering forbedre rækkeviddeestimatets ydeevne med ca. 5–7 dB. Da algoritmen ikke er implementeret som et fuldt, integreret system, kan den samlede forbedring af den totale sensingsydelse endnu ikke konkluderes endeligt. Arbejdet antager stationære mål og interferenter og fokuserer på at validere hypotesetest-baseret interferensdetektion i OFDM-radar.
[This apstract has been generated with the help of AI directly from the project full text]