Mutual Coupling Error Correction Algorithm of MIMO Radar Based on Deep Learning

Multiple-Input Multiple-Output (MIMO) technology was initially widely applied in the field of communications and later introduced into radar systems. By utilizing multiple transmit and receive antennas, MIMO technology enhances the degrees of freedom of radar systems, leading to its increasing attention in recent years. However, the presence of mutual coupling effects between antennas in practical applications negatively impacts the performance of radar systems. Mutual coupling errors, arising from these coupling effects, remain a limiting factor in MIMO radar performance. Existing research on mutual coupling error correction algorithms for MIMO radar mostly adopts active correction methods, which are challenging to implement in engineering practice. This paper proposes a novel array mutual coupling error correction algorithm based on deep learning, which employs self-correction to compensate for the mutual coupling errors. The algorithm leverages deep learning to obtain the compensation matrix for mutual coupling errors, enabling array error correction with certain universality. First, we analyze the signal model of MIMO radar and establish an error generation module. In this module, a data covariance matrix dataset is obtained, which simultaneously contains target information and mutual coupling error information. Subsequently, Convolutional Neural Networks (CNN) are employed to learn the relationship between input covariance matrix data and the compensation matrix for mutual coupling errors, yielding the output compensation matrix. Finally, the obtained compensation matrix is utilized to perform error correction on the radar system, reducing the computational complexity. Experimental results demonstrate that compared to other algorithms, this algorithm exhibits favorable error correction effects and offers a new approach for centralized MIMO radar error self-correction.