Modeling nonlinear communication systems

Signal degradation may be due to noise or to distortion. Linear distortion may be caused by filters, dispersion or multipath effects. These effects are well understood and can often be minimized by linear equalizers. To make further improvements it is necessary to consider the correction of nonlinear distortion. Distortion correction can be performed by using pre or post distortion to compensate (Bellafemina and Benedetto 1985, Benedetto et al 1979, Falconer 1977, Saleh 1982, Saleh and Salz 1982, Schetzen 1980). It is desirable to know whether the structure chosen for a distortion compensator is capable of correcting a particular class of distortion. To this end it is necessary to have a valid model of the distortion to be corrected. The model must be able to simulate the output quite accurately. This apphcation is more demanding than some where a reasonable estimate of the magnitude of the distortion may be all that is required.