Federated Learning meets 6G

Machine learning and data-driven approaches have recently received much attention as a key enabler for future networks. To date, most existing learning solutions for wireless networks have relied on conventional machine learning approaches that require centralizing the training data and inference processes on a single data center. However, in future intelligent wireless networks, due to privacy constraints and limited communication resources for data transmission, it is impractical for all wireless devices that are engaged in learning to transmit all of their collected data to a data center that can subsequently use a centralized learning algorithm for data analytics or network self-organization. To this end, distributed learning frameworks are needed, to enable the wireless devices to collaboratively build a shared learning model with training their collected data locally. Federated learning is one of the most popular distributed learning algorithms, which enables users to collaboratively learn a shared prediction model while remaining their collected data on their devices. For wireless communication, federated learning admits many use cases. For example, federated reinforcement learning algorithms can be used to solve complex convex and non-convex optimization problems that arise in various use cases such as network control, user clustering, resource management, and interference alignment. Moreover, federated supervised learning can be used for user behavior predictions, user identifications, and wireless environment analysis.

However, training federated learning also requires wireless devices and the data center exchange significant amount of information via wireless transmission. Therefore, wireless impairments such as noise, interference, and imperfect knowledge of wireless channel states will significantly affect the training process and performance of federated learning. For example, transmission delay can significantly impact the convergence time of federated learning algorithms. In consequence, it is necessary to optimize wireless network performance for the implementation of distributed learning algorithms.

This research topic, therefore, aims to gather contributions that focus on: a) Optimization of wireless network performance for the implementation of federated learning over wireless networks, and b) federated learning for solving communication problems and optimizing network performance.

High quality technical papers submissions reporting on original algorithmic, theoretical, numerical, and experimental results are welcome. Exceptional survey/tutorial-like papers may also be considered. We solicit original submissions in the following areas:

• Fundamental limits of federated learning
• Wireless network optimization for improving the performance of federated learning
• Radio resource management for federated learning
• Multiple access for federated learning
• Data compression for federated learning
• Adaptive transmission for federated learning
• Interference management in federated learning networks
• Emerging theories and techniques such as age of information and blockchain for federated learning
• Modeling and performance analysis of federated learning networks
• Ultra-low latency federated learning
• Data analytics driven wireless communication
• Federated reinforcement learning for intelligent network control and optimization
• Network architectures and communication protocols for edge machine learning
• Experimental testbeds and techniques of federated learning
• Privacy and security issues of federated learning
• Federated learning for intelligent signal processing, e.g., signal detection • Federated learning for mobile user behavior analysis and inference
• Federated learning for emerging applications, e.g., vehicle to everything (V2X), UAV-enabled communication, Internet of Things, intelligent reflecting surface (IRS), Massive MIMO, virtual reality (VR) and augmented reality (AR).