An MC-CDMA signal equalization and detection scheme based on support vector machines

Abstract

In this paper, we propose a novel nonlinear signal equalization and detection scheme applying the support vector machine (SVM) technique for downlink multicarrier code division multiple access (MC-CDMA) systems. The SVM is gaining popularity in the field of pattern classification due to its promising empirical performance, moderate computation complexity and its strong mathematical foundation. In the proposed scheme, the SVM-based equalizer is trained with the frequency domain received signal for the implicit frequency diversity effect. Simulations are conducted under time-varying multipath fading channel to verify the effectiveness of the proposed scheme in the presence of multiuser interference (MUI) and inter-channel interference (ICI), two major impairments that severely degrade the performance of MC-CDMA systems employing conventional linear equalizers. Simulation results confirmed that the proposed scheme effectively mitigates MUI and ICI and outperforms conventional linear equalizers with perfect channel estimation.