Predicting Wi-Fi link quality through artificial neural networks

Artificial intelligence is one of the key technologies behind the Industry 4.0 (r)evolution. It can be profitably employed in a variety of different applications contexts and with different goals, most of which are characterized by the fact that reliable models for some parts of the involved systems either do not exist or are unavailable. In this paper we tried to exploit artificial neural networks to predict the quality of the transmission channel in Wi-Fi better than what techniques employed in conventional adaptive solutions permit. Applicability of such an approach is quite broad, but we believe that the main intended goal is to improve communication dependability and system resilience. Preliminary results highlighted that artificial neural networks show higher prediction accuracy, and there is still extensive room for improvements.