Fog-enabled Industrial IoT Network Slicing model based on ML-enabled Multi-objective Optimization

Abstract

Fog Computing, as a distributed middleware layer, is expected to bring Cloud capabilities closer to the IoT edge devices. Industrial IoT systems may benefit from the geographically distributed features of the fog nodes, to enhance several QoS requirements such as delay. However, heterogeneous IIoT data traffics require a specific process for each type of them. Thus we refer to a priority classification scheme to slice a fog-enabled IIoT network. For this purpose, we perform a multi-objective optimization algorithm enabled by a machine learning workbench to set a slicing strategy, taking into account the specific and relevant QoS metrics of each priority class. Our slicing model performs better results, in terms of data-rate, e2e delay and network usage, compared to a non-sliced reference model.