Skip to main content Skip to main navigation


AI-driven Scheduling in an IIoT Scenario

Julian Ahrens; Lia Ahrens; Hans Dieter Schotten
In: 2022 Workshop on Next Generation Networks and Applications. Workshop on Next Generation Networks and Applications (NGNA-2022), November 24, Kaiserslautern, Germany, Technische Universität Kaiserslautern, 11/2022.


In this work, a deployment of two neighbouring machines, each endowed with several sensing components communicating over the air, is investigated in an industrial internet of things (IIoT) scenario. Employing the well-known hidden Markov model for characterising the transmission behaviour of the individual sensing components, the scheduling problem and occurrence of collision are defined by means of parameters in such a context. Based on this, a machine learning method for dynamic scheduling is proposed, which comprises a neural network predictor for estimating future interference and a neural network scheduler for optimising resource allocations according to the prediction results. The proposed neural networks are trained and evaluated on simulation data. Some first results are presented herein, which in particular show the high efficacy of the neural network predictor on the one hand and bring to attention some numerical issues encountered during the implementation of the neural network scheduler on the other. Finally, these technical challenges and possible future extensions of the considered scenario are discussed.