Detection of Weak Sonar Signals in Harsh Noise Conditions Using Modified Classical Adaptive Algorithms

Malyshkin G. S., Kuznetsova A. S.

The paper considers optimization of receive path for the case of detecting weak local sources in the presence of stronger local sources. The initial classical problem statement based on spectral decomposition with sample estimate of correlation matrix is complemented with matrix correction before its future use using for example Capon algorithm. The estimate of correlation matrix is corrected as follows: based on analysis of numerical values of eigenvalues and analysis of situation using nonadaptive algorithms, the number M1 of interfering signals, which provide the most significant masking action on the weak signals is determined; M1 largest spectral numbers are modified by certain decreasing their values to mitigate the interfering action. Further, the corrected estimate of correlation matrix is used for example in Capon algorithm with corrected matrix used instead of initial one. As a result, a new algorithm is generated, intended for detection of weak signals (hereinafter referred to as CaponW, W meaning weak), with strong signals presents in its bearing plots in suppressed form. Simulation studies have been performed for the tactical situation with 14 signals, with strong signals being multipath and containing a significant dissipated component. Simulation has shown that the modified CaponW algorithm features better resolution and reduced contact loss time as compared with classical Capon algorithm.

Download original text