Noise separation in analog integrated circuits using blind source separation

Most of the existing noise analysis techniques in analog integrated circuits apply to only single noise source, and cannot take into account the evolving reality of multiple noise sources interacting with each other. Again the individual and relative impacts of various noise sources will determine what types of remedial steps can be taken. This paper proposes the concept of analyzing the impacts of multiple concurrent noise sources in a circuit network, and applies blind source separation technique (BSS) technique to analyze the characteristics of this compound noise effect in analog integrated circuits. The proposed algorithm extracts the time characteristics of individual noise sources from observed noise at a circuit node. The estimated noise sources can aid in timing and spectral analysis and yield better design techniques.