CFAR analysis of the multicoset-thresholding detector: Application to the low complexity sub-Nyquist Radar Electronic Surveillance

Multicoset sampling scheme is a technique to achieve high-speed sampling rate, using a bank of lower-rate sampling channels. In this technique, each channel samples with a small delay with respect to the other channels. As a result, we often can reconstruct the high bandwidth input signal, by wisely combining the information from different channels. However, in many applications, the reconstruction is not the goal. Here, we consider an application, i.e. Radar Electronic Surveillance, in which the aim is the detection and identification of the incoming Radar pulses. As the sampling rate is very high, e.g. up to tens of Giga samples per second, we also need a fast detection scheme. We have recently proposed an efficient multicoset sampling technique, called LoCoMC, which is based on the thresholding for the detection and combining the information from different channels, to extract pulses. We here present an analytical investigation of the thresholding based detection and demonstrate how to choose the thresholding parameter. We then show that the algorithm can be competitive with a state of the art algorithm in performance, while it is computationally very cheap.