A novel threshold detection method for FMCW radar

In frequency-modulated continuous-wave (FMCW) radar, fast Fourier transform (FFT) is applied as a spectrum analyser for beat signal and threshold detection is usually performed on the spectral magnitudes. The picket fence effect of FFT can cause a magnitude loss of the spectral line corresponding to a target, and therefore degrades the detection performance. To analyze this phenomenon in detail, this paper examines the statistical properties of the discrete spectrums of Gaussian noise and beat signal embedded in Gaussian noise, and derives mathematical expressions of false alarm probability and detection probability for FMCW radar. To alleviate the loss caused by the picket fence effect, this paper presents a novel threshold detection method that combines FFT and chirp ztransform (CZT). Simulation results indicate that the proposed method can improve detection probability while reducing false alarm rate, with high operation speed when the number of targets is low.