Range-dimensional frequency band extension hyper beamforming

Abstract

Range estimation is crucial in sonar systems. To enhance range detection capability of active sonar, we propose a range-dimensional frequency band extension hyper beamforming method that improves range resolution and sidelobe suppression performance. The received signal's band is divided into narrowband frequency points. The phase shifts of these frequency points induced by the target distance form a geometric series. A bidirectional linear prediction (LP)-based virtual frequency band extension method exploits phase shift correlations to generate the predicted band. By expanding signal bandwidth, range resolution is preliminarily improved. After extension, the frequency band is divided into two half-bandwidth sub-bands for scanning at different ranges to obtain the beamforming outputs. A “sum” beam is formed by summing the magnitudes of the two sub-band beams, while a “difference” beam is obtained by taking the magnitude of their difference. At the target distance, the “sum” beam reaches its maximum, while the “difference” beam reaches its minimum, both sharing the same sidelobe locations. The hyper beam exponent is chosen to exponentiate the “sum” and “difference” beams, adjusting their contribution weights in the calculation. By amplitude cancellation, the range-dimensional output is optimized. Through nonlinear processing, a range-dimensional power spectrum with low sidelobes and a narrowed main lobe is obtained. Furthermore, we present a fast implementation using complex operations, fast Fourier transform (FFT), and inverse fast Fourier transform (IFFT), improving computational efficiency. Simulations demonstrate the proposed method delivers excellent range resolution and sidelobe suppression capability, while dynamic and static experimental data further validate the effectiveness.