Pre-processing-based performance enhancement of DOA estimation for wideband LFM signals

The prevalence of wideband signals in today’s sensing environments demands accurate and efficient wideband direction of arrival (DOA) estimation techniques. These estimation techniques typically perform complex computations across large operational bandwidths, which leads to high computational overheads. The channelised MUltiple SIgnal Classification (MUSIC) technique achieves an acceptable trade-off between DOA estimation accuracy and computational complexity. This paper investigates several blind pre-processing techniques, that can further reduce the computational complexity of channelised MUSIC while maintaining acceptable DOA estimation accuracy. Using linear frequency modulated (LFM) waveforms, we examine the comparative performance of these blind pre-processing techniques. These techniques are then applied to channelised MUSIC with incoherent aggregation of spatial pseudospectra. The results obtained reveal that the majority of the presented pre-processing techniques achieve significant speedups without sacrificing DOA estimation accuracy. Energy detection (ED) pre-processing reveals superior performance, followed by eigenvalue-based techniques that involve single functions such as maximum eigenvalue (ME), while eigenvalue-based detectors with aggregated functions, such as maximum-minimum eigen-value (MME), can significantly increase computational loads.