• 6G
• data privacy
• federated learning
• fuzzy decision trees
• xai

This thesis work regards the design and development of privacy-preserving learning algorithms for explainable artificial intelligence (XAI) models for classification tasks. We also discuss a case study in the context of future beyond-5g networks. An initial extensive study activity has been carried out to evaluate the current state of art on i) Federated Learning algorithms and frameworks and ii)XAI models . After this first stage, we have specifically selected a Python framework, namely Flower, and Fuzzy Decision Trees as XAI models to adopt in our analysis. We have designed a novel strategy for generating aggregated XAI models in a federated learning fashion. Finally, an extensive experimental campaign has been carried out both on benchmark datasets and on a Quality of Experience (QoE) scenario for validation purposes. The results have been thoroughly analysed in order to provide insights, strengths, weaknesses, and future directions.