Coherent integration for maneuvering target detection via fast nonparametric estimation method

Maneuvering target presents a great threat for radar detection applications. The complex unknown mo-tions between radar and maneuvering target seriously affect the performance of target coherent integra-tion. Existing methods utilizing parameter search or estimation are limited by the high computational burden and the propagation errors of different parameter estimations. In this study, a fast nonparametric estimation method is proposed for coherent integration of maneuvering target. Firstly, the second-order Keystone transform is adopted to eliminate the quadratic range cell migration (RCM). Secondly, the mod-ified range frequency reversal process is proposed to simultaneously remove the complex low-and high -order Doppler frequency broadenings (DFBs) by a one-step matrix multiplication operation. Subsequently, the residual RCM is corrected by the proposed improved time-scaled transform without searching for the Doppler ambiguity number, and then a well-focused result can be obtained. The proposed method is simple to implement and computationally efficient given that searching and estimation of parameters are avoided. It also addresses the high-order DFB and blind speed sidelobe. The processing results of simu-lated and real data are presented to verify the performance of the presented approach. (c) 2022 Elsevier B.V. All rights reserved.

Automated summary

Maneuvering target poses a significant threat to radar detection applications, as the complex unknown motions between the radar and the target adversely affect target coherent integration. This study proposes a fast nonparametric estimation method for coherent integration of maneuvering targets. The method eliminates quadratic range cell migration using the second-order Keystone transform, removes complex low- and high-order Doppler frequency broadenings through a modified range frequency reversal process, and corrects residual range cell migration using an improved time-scaled transform without parameter search or estimation.