Practical training data for support vector machine receiver in a chaos-based CDMA

Abstract

In this paper, a support vector machine (SVM) receiver is used to recover the transmitted symbol on a multi-user chaos-based code division multiple access system. In comparison to all previous treatments, the SVM receiver here employs practical data for training. This means that the training dataset have been corrupted by both additive noise and multi-user interference in the channel. Simulation results show that the quality of the training data would have an influence on both the complexity and the detection performance of the receiver. All results indicate that if possible, the receiver should always collect training samples from a high signal-to-noise ratio (SNR).