Subarray partition based on sparse array weighted K-means clustering

Abstract

This paper introduces the subarray partition based on the sparse array weighted K-means clustering method, which extends the conventional K-means clustering method through the inclusion of a weight matrix approach. This matrix is derived by recording the frequency of each element's occurrence across multiple independent sparse arrays, thereby generating a frequency matrix. The performance of SWKCM is demonstrated through simulations and comparisons with four similar methods. To assess the effectiveness and superiority of the SWKCM, it is applied to the subarray partition of a 40×40 uniform planar phased array and compared with the other four methods. The simulation results show that the proposed SWKCM method maintains comparable sidelobe suppression capabilities to those of KCM, achieving a normalized peak sidelobe level of -43.1076 dB. Furthermore, compared to the K-means clustering method, the sparse array weighted K-means clustering method significantly enhances the stability of subarray partition outcomes, as evidenced by a reduction in the peak sidelobe level standard deviation from 1.0991 to 0.8104, resulting in a 26.3% decrease in variability.