AIC Criterion for Source Number Estimation

The AIC (Akaike Information Criterion) is a widely adopted method for estimating the number of signal sources in multidimensional datasets. This MATLAB implementation provides a computational framework for source number detection, typically involving eigenvalue decomposition of covariance matrices and information-theoretic model selection. While AIC serves as a fundamental approach, alternative criteria like Bayesian Information Criterion (BIC) and Minimum Description Length (MDL) offer different trade-offs between model complexity and goodness-of-fit. The MATLAB code structure generally includes: 1) Data covariance matrix computation 2) Eigenvalue sorting and analysis 3) Information criterion calculation through iterative model comparison 4) Threshold-based source number determination The optimal criterion selection depends on application-specific factors including sample size, signal-to-noise ratio, and algorithmic requirements. Implementation considerations should address: - Eigenvalue spectrum processing techniques - Penalty term optimization in criteria formulas - Validation methods for estimation accuracy Recommendations include comparative analysis of multiple criteria and performance evaluation under varying noise conditions to ensure robust source estimation.