The system performance degradations of inband crosstalk produced by distributed Rayleigh scattering and a single discrete time delayed path have been measured in the same system. Rayleigh scatter crosstalk degrades performance more than equal amounts of discrete crosstalk. By parameterizing the system bit error rate (BER) and received power, we have separated the optical and electrical noise contributions to the BER based on their intensity and crosstalk scaling. We observed behavior consistent with earlier models, but found an unexpected increase in the total optical noise. This excess noise had a very regular linear scaling with crosstalk power. Because of its well defined crosstalk and intensity scaling, this easily measured noise term can be used as a correction to established models to more accurately estimate system performance at high optical powers and low crosstalk levels.
Michael J. Yadlowsky and Valeria L. da Silva, "Experimental Comparison of the Effect of Discrete and Distributed Path Inband Crosstalk on System Performance: Application to Predicting System Performance Penalties," J. Lightwave Technol. 16, 1813- (1998)