Channelized MUSIC with incoherent aggregation for DOA estimation of wideband LFM signals
Date
2020
Authors
Mulinde, R.
Attygalle, M.
Aziz, S.M.
Editors
Advisors
Journal Title
Journal ISSN
Volume Title
Type:
Report
Citation
Statement of Responsibility
Conference Name
Abstract
This report is in fulfilment of the requirement (Deliverable D2) of phase 3 of the collaborative research project conducted by University of South Australia (UniSA) and Systems Group of the Cyber and Electronic Warfare Division of the Defence Science and Technology Group (DSTG). Previous reports of the collaborative projects presented detailed review of relevant literature and suitable techniques for direction of arrival (DOA)) estimation of wideband pulsed nonstationary signals (linear frequency modulated (LFM) and binary phase-coded signals).
This report thoroughly investigates a channelized version of MUltiple SIgnal Classification (MUSIC) for DOA estimation of wideband pulsed LFM signals with high chirp rates with the aim of keeping the computational complexities within reasonable bounds. DOA estimation for wideband signals is highly important in modern communications, sonar and radar systems. Traditional algorithms such as MUSIC used for DOA estimation do not give accurate estimations for wideband signals. Additional complexities arise if the signals of interest are nonstationary, i.e., have a temporal variation of spectral content such as LFM signals with high chirp rates. These computational complexities imply that achieving real-time performance with good DOA estimation accuracy for wideband high chirp signals is a challenge. First, the technique presented here is validated using synthetic data. Further validation is performed using experimental data on pulsed LFM signals generated in an anechoic chamber and received using a uniform linear array (ULA). The pulsed LFM signals considered have bandwidths between 1 MHz and 500 MHz within a duration of 10 µs and have chirp rates of up to 50 MHz/µs.
Results show that an accurate DOA estimation with reasonable computational complexity can be achieved with frequency channelized MUSIC where the pseudospectra from individual channels are incoherently combined to obtain a single estimate. The DOA estimation improves (root mean-squared error (RMSE) reduces) as the chirp rate increases, the number of time segments reduces, and as the number of bins increases. To the best of our Knowleedge, this is the first time this DOA estimation technique for wideband LFM signals with high chirp rate has been validated using real experimental data.
School/Discipline
Dissertation Note
Provenance
Description
Access Status
Rights
Copyright 2020 The Authors