Machine Learning-Based Massive Augmented Spatial Modulation (ASM) for IoT VLC Systems

Abstract

Massive Multiple-Input Multiple-Output (MIMO) technology aims to further the diversity/multiplexing gains of wireless communication systems. Spatial modulation (SM) is a renowned low-complexity MIMO scheme that jointly uses transmitting source indices along with the data stream to convey information. However, the reliability of the spatial stream in SM is significantly influenced by the correlation between channel coefficients. Hence, applying massive-MIMO in visible light communications (VLC) remains an under-investigated area of research, due to the ill-conditioned massive-MIMO VLC channel. Augmented SM (ASM) is an approach that overcomes the channel uniqueness requirement for SM-based VLC systems. This letter adopts massive-ASM for Internet-of-Things applications with a focus on investigating ASM’s complexity and introducing different machine learning based receiver designs, including; support vector machine (SVM), logistic regression (LR), and a neural network (NN). The computational time and transmitter identification accuracy are compared and the system’s bit-error-rate performance is evaluated.