LPI waveform design using chirplet graphs

The Low Probability of Intercept (LPI) signals are of great interest in underwater acoustics. LPI signals are designed such that their presence is difficult to detect, hence taking countermeasures against them are less likely. The traditional method of LPI waveform design is to spread the signal energy in time and/or frequency thus making it indistinguishable from ambient noise. In this paper, we propose a method of energy dispersal that is based on the formalism of chirplet graphs. The parameters of the chirplet graph constitute the “secret key” for the detection of the signal. Detection is accomplished by accumulating the outputs of a filter bank along the same path used by the source. The idea is that although the signal to noise ratio for any one chirplet may not be enough for reliable detection, their sum along the path used by the source will reach the threshold of detectability based on a required false alarm rate. This accumulated local correlations is then tested in a likelihood ratio test. Unauthorized detectors will most likely follow a wrong path and their accumulated energy will fail detection. We show the strength and limitations of the proposed LPI waveform at various ranges using a simulated underwater acoustic channel.