On nonlinear decision functions applied to adaptive equalization for fading channels

In a previous paper (see Proc. of IEEE Malaysia International Conference on Communication, p.280-4, 2001) we proposed a nonlinear decision function for adaptive equalization in the presence of Rayleigh fading. A conventional decision function in adaptive equalization makes use of a hard decision threshold device to generate a difference signal (sometimes also called error signal) necessary in the process of updating the tap weights of the equalizer. This has the disadvantage of not differentiating a very weak signal from a strong signal. Our proposed nonlinear decision function approach weighs less on weak signals and is shown to provide enhancement over a conventional decision function. In this paper, we attempt to justify the use of the proposed clipped parabolic nonlinear decision function and investigate various possible forms of nonlinear decision functions.