Abstract :
Aiming at frequency domain Constant False Alarm Rate (CFAR) detection, the influence of noncoherent integration on frequency domain Peak-to-Average Ratio (PAR) under H0 hypothesis (pure noise) and H1 hypothesis (signal and noise) is quantitatively analyzed. The theoretical analysis and simulation tests show that the influence of noncoherent integration on PAR can be ignored under H1 hypothesis, while under H0 hypothesis, the PAR loss increases with the increasing number of noncoherent integration. The work of this paper reveals that the probability of detection can be improved with the probability of false alarm invariant by noncoherent integration between FFT frames and decreasing the detection threshold. Or for the predetermined probability of detection and false alarm, noncoherent integrations can reduce the requirement of minimum SNR, which is achieved by decreasing the detection threshold.
