This paper outlines novel approaches to design a new generation of distributed networks of acoustic sensors. The key-concept is marked by the use of time-frequency signal analysis tools directly embedded at the sensor node. The general framework for analysis at the acoustic sensor level is based on spectrogram which is very simple and easy to implement. The choice of the detection threshold in time-frequency domain is always a difficult task, even more complicated in the embedded computing configurations where eventual threshold adaptive selection algorithms might not have the required complexity.
That is, the performances in terms of probability of detection classification and localization accuracy are generally strongly depending of the operational conditions. As such, this paper proposes a novel approach of local signal processing based on phase coherence criterion, designed to track the time-frequency multi-components and which is implemented in a distributed network with a minimum number of acoustic sensors.