Machine Learning Approach to Delay Spread Estimation in Industrial Environments

Delay spread is a measure of the multipath characteristics of the wireless channel and predicting it is very important as the time dispersion of the wireless channel leads to intersymbol interference and thus signal distortion. Existing models dealing with delay spread estimation are lacking in flexibility, as they tackle specific deterministic cases. In this work, a flexible approach machine learning tool for the assessment of delay spread values in an industrial environment is presented. The model is tailored to the industrial environment characterized by few macro-parameters, where wireless technologies have been gaining increasing importance in the development of next-generation smart factories. Machine learning is leveraged to get flexibility and fast predictions, i.e. to evaluate the delay spread value based on the geometry of the environment. Results show good performance and confirm the overall physical soundness of the tool.