Abstract
As an emerging wireless application for indoor high-speed wireless access and positioning, visible light communication (VLC) has attracted people's attentions. However, the nonlinear components employed in VLC systems, especially the light-emitting diodes (LEDs), can cause nonlinear distortion to the transmitted signal, which could severely degrade the system performance. Aiming at the nonlinear characteristics of LED, the sparse kernel affine projection algorithm (SKAPA) is investigated in this paper to eliminate the nonlinear distortion for VLC systems. Based on that, to further remove the bad memory effect of LED, a memory effect depression (MED) scheme is proposed, which is carried out following SKAPA. Simulation results show that SKAPA can compensate for the nonlinear distortions, and the combined SKAPA-MED algorithm has a good performance in dealing with memory LEDs. Compared with Kernel Least-Mean-Square (KLMS), SKAPA exhibits better convergence and bit error rate (BER) performance. Additionally, it is also demonstrated that SKAPA has lower complexity than memory polynomial algorithm, and slightly better detection property can be obtained.