Using Federated Learning in the Context of Software-Defined Mobility Systems for Predictive Quality of Service

With the drastic increase in connected devices, such as vehicles or Internet of Things sensors, enormous amounts of data are being generated in a very short time. On the one hand, this big data provides a better opportunity to train machine learning models in Software-Defined Mobility Systems where vehicles are connected to other vehicles and infrastructure, imposing high Quality of Service requirements for data transfer. On the other hand, data dependency in these systems is high and requires integrity and privacy protection as sensitive information is shared by individuals or organizations over a communication link. Federated Learning is one of the promising solutions where algorithms are trained on decentralized devices while datasets are kept locally and it is often used in mission-critical environments, where privacy requirements are high. However, there is a lack of research on predicting Quality of Service parameters using Federated Learning in the context of Software-Defined Mobility Systems to avoid transmission of user-sensitive data over the network. Therefore, we propose a new approach. Our results show that when predicting quality of service in Software Defined Mobility systems, it is possible to preserve the privacy of local data and prepare the system to make proactive decisions before unexpected connectivity degradation occurs. © 2023 The Authors.