Comparison of the Performance of Various Volterra Filters Used for the Equalization of the Channel Dynamic Nonlinearity

In this work, the performance of Volterra filters of various orders (both standard and modified) in compensating for dynamic nonlinear distortions arising during digital transmission was investigated through simulation. A binary signal was employed, and the nonlinear dynamic element was modeled either as a Wiener–Hammerstein device or based on actual LED rate equations. The modulation error ratio (MER) at the equalizer output was used as the primary performance metric. It was demonstrated that the minimum effective sampling rate for the processed signal is two samples per bit. Increasing the sampling rate yielded only marginal improvements in certain cases, while significantly increasing the number of equalizer coefficients. In scenarios involving severe nonlinear and temporal distortions in the signal path, the standard third-order Volterra filter exhibited the best performance. Although modified Volterra filters resulted in lower MER values, they offer reduced complexity due to a smaller number of coefficients, which may be advantageous in some applications. In contrast, linear filters were found to be ineffective for compensating for nonlinear distortions.