Abstract:The concept of federated learning (FL) was first proposed by Google in 2016. Thereafter, FL has been widely studied for the feasibility of application in various fields due to its potential to make full use of data without compromising the privacy. However, limited by the capacity of wireless data transmission, the employment of federated learning on mobile devices has been making slow progress in practical. The development and commercialization of the 5th generation (5G) mobile networks has shed some light on this. In this paper, we analyze the challenges of existing federated learning schemes for mobile devices and propose a novel cross-device federated learning framework, which utilizes the anonymous communication technology and ring signature to protect the privacy of participants while reducing the computation overhead of mobile devices participating in FL. In addition, our scheme implements a contribution-based incentive mechanism to encourage mobile users to participate in FL. We also give a case study of autonomous driving. Finally, we present the performance evaluation of the proposed scheme and discuss some open issues in federated learning.

What problem does this paper attempt to address?

The main problems that this paper attempts to solve are several key challenges faced when applying Federated Learning (FL) on mobile devices, especially in the 5G network environment. Specifically, these problems include: 1. **Privacy Leakage**: - During the federated learning process, participants need to upload updated gradients or model parameters to the server or other participants, which may lead to the leakage of private information of the training data. Attackers may use this information to conduct model extraction attacks, model inversion attacks or membership inference attacks, etc. - Formula representation: Let \( \mathbf{w}_i \) be the model parameter of the \( i \) -th participant, and \( \nabla L_i(\mathbf{w}_i) \) be the corresponding gradient. If an attacker can obtain this information, sensitive data may be inferred. 2. **Unreliable Mobile Devices and Limited Computing Power**: - Mobile devices usually rely on wireless network communication and are widely distributed, resulting in unstable communication. In addition, the computing power of mobile devices is limited, and it is difficult to support complex privacy - protection technologies (such as homomorphic encryption). - Formula representation: Let \( C_i \) be the computing power of the \( i \) -th device, and \( D_i \) be the size of the local data set. Since \( C_i \) is small and \( D_i \) is small, traditional privacy - protection methods may not be applicable. 3. **Incentive Mechanisms and Fairness**: - Mobile users may be unwilling to participate in federated learning because they are worried about privacy leakage or are unwilling to contribute computing resources. Most of the existing incentive mechanisms are based on monetary rewards, ignoring the value of the model itself. - Formula representation: Let \( R_i \) be the contribution score of the \( i \) -th participant, and \( M \) be the global model. A mechanism needs to be designed so that \( R_i \) can accurately reflect the contribution of \( i \) and allocate corresponding model access rights according to \( R_i \). To solve the above problems, the paper proposes a practical cross - device federated learning framework with the following characteristics: - **Anonymous Communication Technologies and Ring Signatures**: Protect the privacy of participants and reduce computing overhead. - **Contribution - Based Incentive Mechanisms**: Quantify the contributions of participants, provide model rewards instead of monetary rewards, to improve participation enthusiasm and ensure fairness. - **System Robustness**: By reducing the need for multi - round interactions, improve the tolerance for single - device failures or disconnections. Through these improvements, the paper aims to promote the practical application of federated learning on mobile devices, especially in fields such as self - driving in the 5G network environment.

A Practical Cross-Device Federated Learning Framework over 5G Networks

Hierarchical Federated Learning: the Interplay of User Mobility and Data Heterogeneity

Federated Learning with Unsourced Random Access.

A Secure Federated Learning Framework for 5G Networks

Towards Energy Efficient Federated Learning over 5G+ Mobile Devices

Toward Energy-Efficient Federated Learning Over 5G+ Mobile Devices

To Talk or to Work: Energy Efficient Federated Learning over Mobile Devices Via the Weight Quantization and 5G Transmission Co-Design

Federated Learning in Mobile Edge Networks: A Comprehensive Survey

Asynchronous Federated Learning over Wireless Communication Networks.

A Joint Communication and Federated Learning Framework for Internet of Things Networks

Privacy-Preserving Federated Learning for Data Heterogeneity in 6G Mobile Networks

Decentralized P2P Federated Learning for Privacy-Preserving and Resilient Mobile Robotic Systems

FedDCT: A Dynamic Cross-Tier Federated Learning Framework in Wireless Networks

Federated Learning for 6G: Applications, Challenges, and Opportunities

Towards Effective Device-Aware Federated Learning

Federated Learning for 6G: Paradigms, Taxonomy, Recent Advances and Insights

A Platform of Federated Learning Management for Enhanced Mobile Collaboration

Blockchain and Federated Learning for 5G Beyond

Wireless Communications for Collaborative Federated Learning

Blockchain-Based Federated Learning in Mobile Edge Networks with Application in Internet of Vehicles

AFLPC: An Asynchronous Federated Learning Privacy-Preserving Computing Model Applied to 5G-V2X