Machine Learning Based Blind Decoding for Space–Time Line Code (STLC) Systems

In this correspondence paper, a novel machine learning (clustering) based blind decoding method is proposed for the space-time line code (STLC) systems without the information of modulation size and channels. The number of clusters, which is equivalent to the modulation size, is estimated from the received signals by using a k-mean cluster validation metrics, such as silhouette score and Davies-Bouldin index, and the cluster indices are directly mapped to the transmitted binary information bits. To improve the clustering performance, received signal normalization and the initial centroids of the clusters are designed by exploiting the combined STLC signal structure. From the numerical results, it is verified that the proposed blind decoding method can achieve near-coherent decoding performance with either small-size modulation, low noise at the receiver, or a large number of transmit antennas.