Robust LFM Target Detection in Wideband Sonar Systems

Abstract

Linear frequency modulation (LFM) signals have been widely used for target detection in active sonar systems due to their robustness to reverberation. However, LFM active sonar requires a large number of reference signals for detecting targets moving with unknown speeds. To obtain more accurate detection results, more reference signals are required, resulting in an increased computational burden and memory size. To cope up with this problem, we propose a new fast target detection method that is robust to the variation of unknown target speed. A large number of reference signals come into a single reference signal by aligning them with precalculated time-shifts, which is followed by a summation process. Both narrowband and wideband cases are considered. The proposed method secures a signal-to-noise ratio (SNR), approaching that of the optimal matched filter output, that is also robust to the variation of target speed and thus it is very useful for the practical use in antitorpedo torpedoes or supercavitating underwater missiles that need to equip low-complexity and robust signal processing systems. Moreover, a rough Doppler estimation is presented using the proposed replica design method. Performance analyses show that the proposed method provides output SNR close to the optimal performance and that the computational load is extremely reduced as compared to the conventional LFM target detection method.