Advancing Edge Intelligence: Federated and Reinforcement Learning for Smarter Embedded Systems
Doktorsavhandling, 2024

Context: The rapid growth of embedded devices and edge computing has brought new opportunities for creating intelligent systems. However, these systems face challenges such as limited computational power and the need to protect user privacy. As a result, there is a need for machine learning methods that can scale effectively, maintain privacy, and adapt to changing conditions in embedded applications.


Objective: This thesis focuses on improving the performance of machine learning models in embedded systems by using federated learning and reinforcement learning. The main goal is to develop methods that allow edge devices to work together without sharing raw data, which helps maintain privacy. Another goal is to make these systems more adaptable to dynamic environments, so they can perform better under changing conditions. Additionally, the research seeks to improve the efficiency of communication and computation across devices.

Method: The research uses a mix of case studies, simulations and real-world experiments. Federated learning is applied to allow edge devices to train models without centralizing the data, keeping sensitive information local. Reinforcement learning is used to help devices learn how to make better decisions by interacting with their environment. These two methods are tested in different scenarios to evaluate improvements in model accuracy, resource use, and adaptability.

Results: The results of this thesis highlight significant advancements in federated learning (FL) and reinforcement learning (RL) for embedded systems. A comprehensive literature review identified six key challenges and open research questions in FL, emphasizing the need for efficient communication, scalability, and privacy preservation. Case studies in telecommunications and automotive applications demonstrated that FL, particularly with asynchronous aggregation protocols, improves model performance, reduces communication overhead, and speeds up training in real-time, dynamic environments. Novel algorithms, such as AF-DNDF and deep RL approaches, further enhanced decision-making capabilities and adaptability in applications like autonomous driving and UAV base station deployment for disaster scenarios. The development of frameworks like EdgeFL provided practical solutions to overcome FL's implementation challenges, offering scalable, low-effort alternatives. Overall, the integration of FL and RL into embedded systems resulted in improved model accuracy, resource utilization, and adaptability, making these approaches highly suitable for real-world industrial use cases.

Conclusion: This research advances the field of edge intelligence by providing a practical approach to deploying machine learning models that are scalable, privacy-focused, and adaptive in embedded systems. The work demonstrates clear improvements in performance and offers a foundation for future research, which could explore more complex learning approaches and apply these techniques to a wider range of embedded systems.

Federated Learning

Software Engineering

Machine Learning

Reinforcement Learning

Embedded Systems

Mötesrum 473, Campus Lindholmen. Building Jupiter. Entrance from Hörselgången 5. Go to floor 4.
Opponent: Xavier Franch, Professor Informatics, UPC Universitat Politècnica de Catalunya, Spain

Författare

Hongyi Zhang

Software Engineering 1

Engineering Federated Learning Systems: A Literature Review

Lecture Notes in Business Information Processing,;Vol. 407(2021)p. 210-218

Paper i proceeding

Towards Federated Learning: A Case Study in the Telecommunication Domain

Lecture Notes in Business Information Processing,;Vol. 434 LNBIP(2021)p. 238-253

Paper i proceeding

Federated learning systems: Architecture alternatives

Proceedings - Asia-Pacific Software Engineering Conference, APSEC,;Vol. 2020-December(2020)p. 385-394

Paper i proceeding

Real-time end-to-end federated learning: An automotive case study

Proceedings - 2021 IEEE 45th Annual Computers, Software, and Applications Conference, COMPSAC 2021,;(2021)p. 459-468

Paper i proceeding

AF-DNDF: Asynchronous Federated Learning of Deep Neural Decision Forests

Proceedings - 2021 47th Euromicro Conference on Software Engineering and Advanced Applications, SEAA 2021,;(2021)p. 308-315

Paper i proceeding

Autonomous Navigation and Configuration of Integrated Access Backhauling for UAV Base Station Using Reinforcement Learning

Proceedings - 2022 IEEE Future Networks World Forum, FNWF 2022,;(2022)p. 184-189

Paper i proceeding

Deep Reinforcement Learning in a Dynamic Environment: A Case Study in the Telecommunication Industry

Proceedings - 48th Euromicro Conference on Software Engineering and Advanced Applications, SEAA 2022,;(2022)p. 68-75

Paper i proceeding

5G Network on Wings: A Deep Reinforcement Learning Approach to the UAV-Based Integrated Access and Backhaul

IEEE Transactions on Machine Learning in Communications and Networking ,;Vol. 2(2024)p. 1109-1126

Artikel i vetenskaplig tidskrift

Enabling efficient and low-effort decentralized federated learning with the EdgeFL framework

Information and Software Technology,;Vol. 178(2025)

Artikel i vetenskaplig tidskrift

In an era where technology is deeply integrated into our daily lives, ensuring privacy and efficiency in intelligent systems is a increasing challenge. This thesis focuses on two key machine learning techniques, Federated Learning and Reinforcement Learning, to develop smarter, more adaptive, and privacy-preserving embedded systems. Federated Learning enables devices to collaboratively train AI models locally, without transmitting sensitive data to centralized servers, while Reinforcement Learning enables systems to learn and adapt through trial and error in dynamic environments.

My research investigates how FL can be effectively engineered for resource-constrained devices, ensuring robust privacy and minimal communication costs. It also explores scalable architectures that optimize performance in large, distributed networks. By integrating FL and RL, this work demonstrates how intelligent systems such as autonomous vehicles and drones can not only collaborate to improve their capabilities but also make real-time decisions in complex scenarios, like navigating traffic or managing 5G network connectivity. This thesis provides a foundation for creating smarter embedded systems that prioritize privacy, adaptability, and efficiency. From safer, more autonomous vehicles to highly responsive network systems, the findings highlight the transformative potential of combining Federated and Reinforcement Learning in real-world applications.

Software Engineering for AI/ML/DL

Chalmers AI-forskningscentrum (CHAIR), 2019-11-01 -- 2022-11-01.

Infrastruktur

C3SE (Chalmers Centre for Computational Science and Engineering)

Ämneskategorier

Programvaruteknik

ISBN

978-91-8103-143-0

Doktorsavhandlingar vid Chalmers tekniska högskola. Ny serie: 5601

Utgivare

Chalmers

Mötesrum 473, Campus Lindholmen. Building Jupiter. Entrance from Hörselgången 5. Go to floor 4.

Online

Opponent: Xavier Franch, Professor Informatics, UPC Universitat Politècnica de Catalunya, Spain

Mer information

Senast uppdaterat

2024-11-22