A Bayesian-Based CFAR Detector for Pareto Type II Clutter

Interference control in sliding window detection processes using a Bayesian approach

“…and since it is difficult to invert (14) to solve for τ one can instead reject H 0 if P FA (z 0 ) is smaller than the design Pfa, as explained in [1], [4]. It is interesting to observe that with the choice of π N = 1 and all other π j = 0, the Pfa reduces to that derived in the previous section, as expected.…”

“…Variation of the Pfa is a disaster from a target tracking prespective [3]. Although the Bayesian approach has been developed for the case of Pareto Type II clutter in [1], here the principles are illustrated in the simple case of exponentially distributed clutter, corresponding to Gaussian statistics in the complex domain. It is worth observing that the basic design paradigm will extend to more useful model for clutter, including the Pareto, Weibull, K and KK families of distributions…”

Mitigation of the Effects of Unknown Sea Clutter Statistics by Using Radial Basis Function Network

“…from which one can derive the detection threshold T θ n . In [41], detector for Pareto Type II clutter using sliding window based approach in Bayesian framework was proposed. Although true CFAR detector, it shows lower detection performance in comparison to techniques based on geometric transformation or ordered statistics that require prior knowledge of shape parameter in the former and scale parameter in the latter case [42].…”

“…Although true CFAR detector, it shows lower detection performance in comparison to techniques based on geometric transformation or ordered statistics that require prior knowledge of shape parameter in the former and scale parameter in the latter case [42]. Another drawback of the CFAR detector proposed in [41] is the need for numerical integration, in contrast to the method of moments which can be used in geometrical transformation and ordered statistics approach. Hence, for simplicity, in this paper we adopt traditional [zlog(z)] method as the reference method for shape parameter estimation of both the Pareto Type II (this is density (20) under transformation y = x 2 ), and the K-distributed clutter (18), although it is difficult to capture statistics of such low values of shape parameter as given in example in Fig.…”

Bayesian framework for detector development in Pareto distributed clutter

“…Unfortunately, the complete sufficient statistic approach in [9], nor the invariance approach in [10], could rectify this issue. Hence a Bayesian approach was introduced in [20], where it was shown that a CFAR detector could be developed for the Pareto Type II case.…”