Cross-Spectral Density (CSD)
June 2, 2021
The cross-spectral density (CSD) function is the discrete-time Fourier transform of the cross-correlation function Cyx(m). It is defined as:
The CSD expresses how correlated a signal x is in relation to another signal y. The CSD is nearly the same as the auto spectral density (ASD), except the signal may be compared to another signal.
The ASD of x(n) is a special case of the CSD of x(n) and y(n), where y(n)=x(n). That is, Rxx(m)=Cxx(m).
- A pair of signals with CSD=0 indicates they are very uncorrelated.
- If the CSD is flat but non-zero, then y and x are uncorrelated but are either the same signal or almost the same signal.
- If the CSD has a main lobe, then the pair of signals has some degree of correlation. In general, the wider the main lobe, the more uncorrelated the pair. Likewise, the narrower the main lobe, the more correlated the pair.
- A CSD with a spike at some frequency (F) in Hertz indicates the pair of signals is periodically correlated every 1/F seconds. This may be due to a resonant frequency at F.
- If the CSD is nearly flat, then the samples are pretty much uncorrelated.
- A CSD with a single spike at 0Hz is likely caused by the presence of a large DC offset in both signals.
- A CSD with a medium width main lobe indicates that samples in one signal are correlated with other samples in the other located close in time, but that the correlation diminishes with separation in time.
CSD and Spectral Power
The CSD and power spectral density (PSD) are closely related. In fact, the PSD is a special case of CSD. If two signals x and y analyzed by the CSD are the same (x=y), then the CSD is the PSD. And, just as with PSD, the true CSD is unknowable but can be estimated.
- Bendat, Julius S. and Allan G. Piersol, Engineering Applications of Correlation and Spectral Analysis. John Wiley & Sons, 1993.
- Bendat, Julius S. and Allan G. Piersol, Random Data: Analysis and Measurement Procedures. John Wiley & Sons, 2011.
- Cadzow, James A., Foundations of Digital Signal Processing and Data Analysis. New York: Macmillan Publishing Company, 1987.
- Carter, G. Clifford. Estimation of the magnitude-squared coherence function (spectrum) (Technical Report AD0743945). Newport: Naval Underwater Systems Center, 1972.
- Clarkston, Peter M., Optimal and Adaptive Signal Processing, 1st ed. CRC Press, 1993.
- Koopmans, Lambert H., The Spectral Analysis of Time Series, 2nd ed. Academic Press, 1995.
- Papoulis, Anthanasios, Probability, Random Variables, and Stochastic Processes, 2nd ed. New York: McGraw-Hill, 1984.
- Papoulis, Anthanasios, Probability, Random Variables, and Stochastic Processes, 3rd ed. New York: McGraw-Hill, 1991.
- Stoica, Petre and Randolph Moses, Spectral Analysis of Signals. Upper Saddle River: Prentice Hall, 2005.
- Welch, P. D., “The use of fast Fourier transform for the estimation of power spectra: A method based on time averaging over short, modified periodograms.” IEEE Transactions on Audio and Electroacoustics 15, no. 2 (1967): 70-73. DOI: 10.1109/TAU.1967.1161901.
- Wiener, Norbert, “Generalized harmonic analysis.” Acta Matematica 55 (1930): 117-258. DOI: 10.1007/BF02546511.