A Group Invariance Approach to a Very Weak LFM Signal Detection

摘要

This paper considers the detection of a very low signal to noise ratio (SNR, linear frequency modulated (LFM) radar waveforms from data received by a wideband receiver. The optimal method involves a computationally intensive two dimensional search. Faster alternatives, include the discrete ambiguity approach to LFM detection/estimation which is computationally efficient but is well known to be applicable only at moderately high SNR, while the windowed Fourier transform can be used to detect low SNR signals, but only for quite small chirp rates. Here we utilize multiple time and frequency shifts applied to the received data to structure the problem as one of detection of a multi-channel unknown rank-one component in noise. Our method which involves only a one dimensional search over chirp rate, works at very low SNR and can handle multiple signals and interferers The generalized-likelihood ratio test (GLR, the Bayesian test are discussed and compare with the generalized coherence test. The detection performance are demonstrated through numerical simulations.