Optimization-Based Mismatched-Channel Filtering Using ADMM for Continuous Active Sonar

Generalized Sinusoidal Frequency Modulation (GSFM) signals can enhance Continuous Active Sonar (CAS) performance by providing high sub-signal processing gain while achieving high target update rates. However, conventional processing methods for GSFM often exhibit high sidelobe levels arising from the waveform’s autocorrelation which degrade detection performance, especially in severe multipath environments. To address this issue, a Mismatched-Channel Filtering (MMCF) method for GSFM in CAS is proposed to focus multipath energy while suppressing sidelobe levels. Adopting the sub-pulse processing scheme, we incorporate the orthogonality of GSFM sub-signals (optimized via a genetic algorithm) and sparse channel estimates into the MMCF design for each sub-signal. The design is formulated as a Quadratically Constrained Quadratic Program (QCQP) and solved iteratively using the Alternating Direction Method of Multipliers (ADMM) for long-duration signal processing in CAS. Numerical simulations demonstrate that, compared with the matched filtering and matched channel filtering methods, the proposed MMCF method effectively suppresses sidelobe levels by approximately 20 dB and produces a Dirack-like main-lobe peak, while efficiently focusing multipath energy. The method’s effectiveness is further validated using experimental data from a lake trial. Therefore, this algorithm has distinct advantages for signal processing in multipath environments.