Distributed array of cooperating acoustic sensors using local time-frequency coherence analysis

The distributed array of sensors is an interesting concept in the context of underwater environment monitoring. The signal processing distributed at sensors level has to have a low consumption (much less than the data transmission) and, in the same time, the processing efficiency needs to be satisfactory compared with a processing without any algorithmic complexity constraint. In this paper, we define a network of sensors which individually detect signals of interest based on the concept local time-frequency coherence analysis. This concept consists of estimating, in each analysis window, the local linear frequency modulation that matches the best the analyzed signal. This concept is much less sensitive to the time-varying energy of the signals, providing better detection performances than an energy-based detector. In addition, the sensors are organized in local sub-networks that will provide a joint detection stage. This operation consists in comparing the local time-frequency information obtained by the sensors of the sub-network and if they are the same, at the sub-network level, we decide that the detected signal is a useful one and its parameters has to be sent to the central processing level. This cooperating capability improves the performances as it is proved by the results on real data and quantified in terms of receiving operating characteristics.