Measurement information assisted soft decision line spectrum detection

To improve the detection capability of weak line spectra, a measurement information assisted track-before-detect algorithm (MIA-TBD) is proposed. The method applies the track-before-detect (TBD) framework to line spectrum detection using the single vector hydrophone. First, multi-channel signals acquired by the vector hydrophone are coherently combined to obtain the active acoustic intensity. To address the difficulty in accurately obtaining the likelihood function of the active acoustic intensity, measurement information is introduced to assist in constructing the likelihood function. In the MIA-TBD algorithm, the active acoustic intensity is further employed as the detection statistic for line spectrum detection. Simulation analyses and experimental results demonstrate that, under low signal-to-noise ratio conditions, compared with conventional TBD-based detection methods and constant false alarm rate (CFAR) detectors, the proposed MIA-TBD algorithm can improve the line spectrum detection probability by more than 20%, significantly enhancing the detection performance of weak tonal signals.