A Novel MIMO IoT Antenna Selection Strategy Assisted by Machine Learning

In this paper, we suggest a transmission antenna selecting (AS) method assisted by a multi-label convolutional neural networks for end-to-end multiple-input multiple-output Internet of Things (IoT) connectivity networks under correlation channels circumstances. In the suggested convolutional neural networks-aided transmission AS multiple-input multiple-output IoT systems, we choose to use the notion of multi-label rather than the traditional single-label multi-class categorization ML schemes, which may significantly shorten the duration of train labeling in the case of multi-antenna choice. Using the multi-label idea may also greatly increase the trained convolutional neural networks model’s prediction performance in related large-scale multiple-input multiple-output channels situations with less train information. According to the relevant simulations outcomes, the suggested convolutional neural networks-aided AS system may be capable to achieve near-optimal capacities efficiency in real time, where the efficiency is somewhat unaffected by the consequences of faulty CSI.