MUSIC Algorithm Based on Higher-Order Cumulants

The MUSIC algorithm based on higher-order cumulants is an improved direction-of-arrival (DOA) estimation method designed for array signal analysis in signal processing applications. While the conventional MUSIC algorithm relies on second-order statistics (such as covariance matrices) for signal and noise subspace decomposition, its performance may degrade under non-Gaussian noise or correlated signal source environments.

Higher-order cumulants (HOC) can extract non-Gaussian characteristics of signals, suppress Gaussian noise interference, and preserve phase information. By replacing the traditional covariance matrix with a higher-order cumulant matrix, this enhanced MUSIC algorithm improves noise resistance and resolution. In MATLAB implementation, the algorithm typically involves these key steps:

Higher-order cumulant calculation: Compute fourth-order cumulants (or other higher-order statistics) from array received signals to construct the cumulant matrix. This can be implemented using MATLAB's signal processing functions to handle multidimensional array data. Subspace decomposition: Perform eigenvalue decomposition on the cumulant matrix to separate signal and noise subspaces. The MATLAB 'eig' or 'svd' functions are commonly used for this matrix decomposition operation. Spatial spectrum estimation: Construct the spatial spectrum function using the noise subspace and determine signal source directions through peak search algorithms. MATLAB's peak finding functions can be integrated for efficient DOA estimation.

This algorithm is particularly suitable for multi-source localization in complex electromagnetic environments, including radar systems, sonar applications, and smart antenna technologies in wireless communications. Compared to the traditional MUSIC algorithm, the higher-order cumulants version demonstrates superior performance in low signal-to-noise ratio scenarios and with coherent signals, though it requires higher computational complexity. MATLAB's matrix operations and Signal Processing Toolbox provide efficient implementation capabilities for this algorithm.