ML-based estimation of the number of devices in industrial networks using unlicensed bands

Abstract—Advanced automation is being adopted by manufacturing facilities and wireless technologies are set to be a key
component in driving the factories of the future. It is expected that private cellular networks and WLAN technologies would
be deployed for smart factory operations. Since both wireless technologies can operate on the same channel in unlicensed
bands, then efficient resource sharing becomes important. When multiple devices compete for the resource, the estimation of
number of devices contending for the channel resource can help the design of an efficient resource sharing scheme. This paper
aims to address the challenge of estimating the number of factory devices contending to transmit over the unlicensed channel. We
adopt three machine learning (ML) techniques and develop a novel device number estimation system by collating and analysing
the idle-time interval between transmission across the channel. By using NS-3 simulation, the performance of the proposed
estimation approach is evaluated. The results presented reveal the significance of the chosen features and performance of each
ML algorithm used.