Blind channel estimation for whistle-mimicking signals based on signal segment extraction and orthogonal matching pursuit

Blind channel estimation (BCE) is crucial for identifying modulation schemes and estimating the modulation parameters of non-cooperative whistle-mimicking communication signals in underwater acoustic environments. However, whistle-mimicking signals, a common type of bio-inspired underwater acoustic communication signal, exhibit long duration, wide bandwidth, and non-stationary characteristics, which limit the performance of cross-relation model-based methods and source signal reconstruction methods. To address this, this paper proposes a blind channel estimation method that combines signal segment extraction with orthogonal matching pursuit (OMP). A selection approach for the optimal channel signal is introduced based on time–frequency contour masking filtering and cross-correlation among multiple received signals. A whistle-mimicking signal extraction method is developed by integrating energy-peak denoising with region-growing-based time–frequency contour tracking. The time–frequency contour of the segment that is less affected by multipath effects is extracted through the signal extraction method. This contour is converted back via inverse short-time Fourier transform (ISTFT) to serve as the OMP reference signal. Furthermore, minimum mean square error (MMSE) equalization mitigates multipath interference, and the extraction method is reapplied to reconstruct the whistle-mimicking signal. Simulation and sea experiments validate the effectiveness of the proposed method, achieving correlation coefficients of 0.8 at 5 dB and exceeding 0.85 in real underwater environments.