References & Further Reading
References
- A. R. Elkordy, G. Caire, and A. S. Avestimehr, Federated Learning with Information-Theoretic Privacy and Representation Sharing, 2023
**CommIT contribution.** The IT-secure federated representation learning result of §17.4. Combines AirComp with sum-zero masks for IT privacy of learned representations.
- H. B. McMahan, E. Moore, D. Ramage, S. Hampson, and B. A. y Arcas, Communication-Efficient Learning of Deep Networks from Decentralized Data, 2017
FedAvg paper. Canonical FL algorithm, basis of Chapter 17's wireless extension.
- M. M. Amiri and D. Gunduz, Machine Learning at the Wireless Edge: Distributed Stochastic Gradient Descent Over-the-Air, 2020
Wireless FL over AirComp: convergence analysis under aggregation MSE. Foundation for §17.2.
- K. Yang, T. Jiang, Y. Shi, and Z. Ding, Federated Learning via Over-the-Air Computation, 2020
Canonical reference for AirComp-based FL: scheduling, power control, convergence. Basis for §17.3.
- X. Cao, G. Zhu, J. Xu, and K. Huang, Optimized Power Control Design for Over-the-Air Federated Edge Learning, 2020
Threshold scheduling and Pareto- optimal power control for AirComp FL. Basis for Theorem 17.3.1.
- M. Chen, Z. Yang, W. Saad, C. Yin, H. V. Poor, and S. Cui, A Joint Learning and Communications Framework for Federated Learning over Wireless Networks, 2021
Joint optimization of learning and communication for wireless FL. Referenced in §17.3 for the energy water-filling and fairness- aware scheduling results.
- L. Bottou, F. E. Curtis, and J. Nocedal, Optimization Methods for Large-Scale Machine Learning, 2018
SGD convergence theory under noisy gradients. Underlies Theorems 17.2.1-17.2.3.
- K. Bonawitz et al., Practical Secure Aggregation for Privacy-Preserving Machine Learning, 2017
The pairwise-masking mask generator that underlies §17.4's sum-zero cryptographic masks.
- B. Nazer and M. Gastpar, Computation Over Multiple-Access Channels, 2007
Foundational computation-over-MAC paper. Referenced for AirComp primitive inheritance.
- J. Konecny, H. B. McMahan, F. X. Yu, P. Richtarik, A. T. Suresh, and D. Bacon, Federated Learning: Strategies for Improving Communication Efficiency, 2016
Early survey of FL communication- efficiency strategies. Motivates AirComp as a bandwidth-saving primitive for FL.
- P. Kairouz et al., Advances and Open Problems in Federated Learning, 2021
Comprehensive FL survey. Useful context for wireless FL in the broader privacy-utility landscape.
- G. Zhu, Y. Wang, and K. Huang, Broadband Analog Aggregation for Low-Latency Federated Edge Learning, 2020
Analog aggregation for low-latency FL. Early practical AirComp-FL system design.
- T. Sery and K. Cohen, On Analog Gradient Descent Learning Over Multiple Access Fading Channels, 2020
Rigorous analysis of analog gradient descent in fading wireless FL.
Further Reading
Resources for deeper study of wireless FL convergence, scheduling, and IT-secure representation learning.
Wireless FL — tutorial
Amiri & Gunduz, IEEE T-SP 2020; Yang et al., IEEE T-WC 2020
The two canonical tutorials on wireless FL, together covering the convergence analysis, scheduling, and power-control strategies of Chapters 17.1-17.3.
IT-secure federated representation learning
Elkordy, Caire, Avestimehr, IEEE JSAIT 2023
The CommIT contribution that closes §17.4. Read in full for the rate-privacy trade-off region and the mask-design strategy.
Joint learning-communication optimization
Chen et al., IEEE T-WC 2021
Full joint optimization of learning and communication; the decomposition framework of §17.3.
FL open problems survey
Kairouz et al., FTML 2021
Breadth context: where wireless FL fits in the global FL research landscape, and the currently-open problems.
AirComp primitives foundations
Nazer & Gastpar, IEEE T-IT 2007 (original MAC-computation); Goldenbaum et al., IEEE T-SP 2013
The physics and information theory underlying the AirComp aggregation primitive used throughout Chapter 17.