A robust detector based on the principal component analysis in uncertain ocean

Environmental uncertainty can cause severe performance degradation to sonar detection algorithms that rely on precise knowledge of the environmental parameters. The optimal uncertain field processor (OUFP) can detect the signal robustly. However, it suffers a big computation burden. A new robust detector is proposed in this paper which uses the Bayesian theory to utilize the a priori information of the environmental acoustic parameters and the principal component analysis to reduce the computation cost. The detector can detect the signal robustly and efficiently. Computer simulation and experimental data verification show that, (1) the proposed detector has an equal detection performance to the OUFP and outperforms the conventional used mean ocean detector and energy detector; (2) The computation complexity is typically 1/8-1/5 of the OUFP; (3) Assuming that the ocean environmental uncertainties range is somewhat greater than truth can get a better performance.