CFAR based NOMP for Line Spectral Estimation and Detection

The line spectrum estimation and detection problem are considered in this paper. We propose a CFAR-based Newtonized OMP (NOMP-CFAR) method which can maintain desired false alarm rate without the knowledge of noise variance. NOMP-CFAR consists two steps, namely, an initialization step step. In step, NOMP is employed to obtain candidate sinusoidal components. CFAR detector applied detect each frequency, provides “soft” thresholds. After removing most unlikely target, Newton refinements used refine remaining parameters. relationship between required threshold established. Compared with NOMP, has only 1 dB performance loss additive white Gaussian scenario probability $10^{-2}$ 0.8 For varied variance scenario, still preserves its property, while violates CFAR. Besides, real experiments also conducted demonstrate NOMP-CFAR, compared NOMP.