Dimension-Reduced Robust Adaptive Beamforming via Group Cyclic Minimization

Adaptive digital beamforming (ADBF) is a widely used anti-interference technology for radar and communication systems. However, as the scale of receiving array increases and the fast time-varying interference is applied, reducing the computational complexity of ADBF is demanded to meet realtime requirements. To that end, a dimension-reduced ADBF algorithm based on group cyclic minimization, named GCM-ADBF, is proposed for large-scale array systems. By grouping the weights to be optimized and introducing the groupwise blocking matrix, GCM-ADBF enjoys fast implementation with parallelizable low-dimensional matrix inversion and iterations involving only matrix multiplications and additions, as well as superior robustness against few training samples. Moreover, GCM-ADBF is free from pre-estimation of the number and directions of interferences. The proposed algorithm is available for arrays with arbitrary configurations, and is compatible with fully digital and hybrid analog-digital beamforming systems. The superiority is verified by numerical simulations.

I. Introduction

As the electronic warfare technology develops and the electromagnetic spectrum becomes increasingly crowded, interference suppression is playing an essential role in radar and communication systems [1], [2]. Owing to the extensive application of digital arrays, adaptive digital beamforming (ADBF), especially the minimum variance distortionless response (MVDR) beamformer [3], [4] and the linear constraint minimum variance (LCMV) beamformer [5], [6], has been widely adopted for interference mitigation in spatial domain. However, the closed-form optimal solution for the weight vector of MVDR or LCMV involves matrix inversion, which is computationally expensive for large-scale arrays. Reducing the complexity of ADBF is thus particularly significant in practical applications, especially for synthetic aperture radar (SAR) [1] due to the limitation on the load of platforms.