A Subspace Spectrum Estimation Method for MIMO Radar Models

In MIMO radar models, we propose a novel subspace spectrum estimation method that employs an over-estimation technique to address source number estimation challenges. Unlike traditional MUSIC algorithms, our approach directly transforms the data covariance matrix through mathematical operations to construct both signal subspace projection matrices and noise subspace projection matrices. The key implementation advantage is the elimination of eigenvalue decomposition, which bypasses the complex problem of distinguishing between small and large eigenvalues that MUSIC algorithms typically face in non-ideal conditions. This design significantly reduces algorithmic complexity while maintaining accuracy across any number of snapshots. The method's core functionality involves matrix transformations and projection operations rather than spectral decomposition, making it computationally efficient. Furthermore, in scenarios with coherent sources, our technique accurately estimates target incidence angles without producing false peaks, demonstrating robust performance in challenging signal processing environments for MIMO radar applications. The implementation typically involves calculating the covariance matrix from received data, applying linear transformations to derive projection matrices, and then performing spectrum estimation through projection operations. This approach provides a more stable and efficient solution for MIMO radar signal processing compared to conventional subspace methods.