Abstract
This paper addresses the problem of range-spread target detection in non-Gaussian clutter, which can be modeled as a spherically invariant random vector. First, the covariance matrix structure is assumed to be known, and a normalized matched filter integrator (NMFI) is proposed. The formula relating the false alarm probability to the detection threshold of the NMFI is then deduced. The NMFI statistic is the sum of the NMF statistics, and thus the NMFI outperforms the NMF for range-spread target detection. Moreover, it is assumed that the power spectral density of the baseband equivalent of the clutter is symmetric about <italic>f</italic>=0, and the modified recursive estimator (MRE) of the clutter covariance matrix structure is proposed. The MRE makes full use of the prior clutter information. As the initialization estimation matrix and the recursive progression require only real number operations, the MRE has a lower computational complexity than the adaptive estimator with recursive estimation (AE-RE). Finally, an adaptive NMFI (ANMFI) with MRE (MRE-ANMFI) is obtained by substituting the estimated matrix for the known covariance matrix structure of the NMFI. The constant false alarm rate properties of the MRE-ANMFI are theoretically proved. Simulation results show that the MRE has higher estimation accuracy than the AE-RE, and the performance of the MRE-ANMFI is better than that of the ANMFI with AE-RE.
Published Version
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call