Narrowband Detection of Underwater Acoustic Signal Under Noise Uncertainties

In this work, the problem of narrowband acoustic signal detection in an ocean using an array of sensors is investigated. The associated hypothesis testing problem is formulated as a goodness-of-fit test (GoFT) due to unavailability of the prior knowledge of the signal vector and the noise variance. A new GoFT, based on the statistics of the number of zero-crossings in the received observations (called the zero-crossings based detector or ZCD) is proposed. ZCD offers many advantages over existing detectors such as the correlator detector (CD), the subspace detectors (SD) and the energy detector (ED). First of all, the ZCD does not require the knowledge of the signal vector. Unlike the SD, the ZCD does not require any knowledge of the channel; and unlike the ED, the ZCD is immune to noise variance uncertainties. Moreover, the computational complexity of ZCD is the same as that of the simple ED. Through Monte Carlo simulations, it is shown that ZCD outperforms ED under practically relevant scenarios such as noise variance uncertainties and low SNR, in heavy-tailed noise and low false-alarm requirement regimes.