Multiband Radar Signal Phase Reference Synthesis Based on Matrix Pencil Method

Abstract

The multi-band signal phase reference fusion technique fuses several different sub-bands into one large bandwidth signal at the signal layer, so it can effectively improve the radar distance image resolution. Currently, most of the fusion techniques based on the full-pole model use root-MUSIC and its improved algorithm to achieve the estimation of the poles, and this method can obtain good fusion results at high signal-to-noise ratios. However, the root-MUSIC algorithm is susceptible to noise interference in low signal-to-noise conditions, which makes it difficult to extract the poles correctly and thus greatly affects the final effect of signal fusion. In order to reduce the influence of noise, we propose to use the averaging matrix bundle algorithm to achieve multi-band pole estimation, find out the linear phase term and compensate it, and use the coherence function method to find out the fixed phase term on this basis. Finally, the signal fusion is performed by combining the Matrix pencil method using the Two-segment echo data after alignment. The simulation experiments and the measured data show that the method effectively improves the multiband signal fusion effect at low signal-to-noise ratio, which verifies the feasibility of the method.