Modulation Recognition for MIMO Relaying Broadcast Channels with Direct Link

In this letter, we investigate the performance of modulation identification based on pattern recognition approach using the decision tree (J48) classifier, for multiple-input multiple-output (MIMO) relaying broadcast channels with direct link (source-to-destination). The proposed system identifies the modulation type and order among different M-ary shift-keying linear modulations used by broadband technologies such as long term evolution-advanced (LTE-A) and worldwide interoperability for microwave access (WiMAX). The system under study employs features extraction based on higher order statistics (HOS) of the received signal. Based on receiver operating characteristic (ROC) curves, our study shows that J48 classifier is more efficient than the multilayer perceptron (MLP) classifier trained with resilient backpropagation training algorithm (RPROP) where it achieves close to perfect detection rate (over 99%) with reasonable training time in acceptable signal-to-noise ratio (SNR) range. We also show that the performance of the MIMO relaying broadcast network is remarkably better than the traditional MIMO one.