Analysis of the Correlation Coefficient Between Component Noise Squared Norms for OFDM Systems

We present an exact analysis of the correlation coefficient, in the frequency domain, between the real and imaginary parts of the impulse noise squared norms for OFDM systems. We analyze a scenario where M Gaussian randomly positioned impulses interfere the OFDM signal, with and without the presence of a background thermal noise. Variability of M is taken into account through the Bernoulli-Gauss model. We demonstrate that rather high correlation values occur with not negligible probabilities under certain operation conditions. This has an impact on the design of impulse noise suppression techniques as well as on the error rate performance.