One-Bit Over-the-Air Aggregation for Communication-Efficient Federated Edge Learning: Design and Convergence Analysis
Open accessThe DOI matches its Crossref record; 35 of the 37 references it cites were found in a registry or an index.
Abstract
Federated edge learning (FEEL) is a popular framework for model training at an edge server using data distributed at edge devices (e.g., smart-phones and sensors) without compromising their privacy. In the FEEL framework, edge devices periodically transmit high-dimensional stochastic gradients to the edge server, where these gradients are aggregated and used to update a global model. When the edge devices share the same communication medium, the multiple access channel (MAC) from the devices to the edge server induces a communication bottleneck. To overcome this bottleneck, an efficient broadband analog transmission scheme has been recently proposed, featuring the aggregation of analog modulated gradients (or local models) via the waveform-superposition property of the wireless medium. However, the assumed linear analog modulation makes it difficult to deploy this technique in modern wireless systems that exclusively use digital modulation. To address this issue, we propose in this work a novel digital version of broadband over-the-air aggregation, called one-bit broadband digital aggregation (OBDA). The new scheme features one-bit gradient quantization followed by digital quadrature amplitude modulation (QAM) at edge devices and over-the-air majority-voting based decoding at edge server. We provide a comprehensive analysis of the effects of wireless channel hostilities (channel noise, fading, and channel estimation errors) on the convergence rate of the proposed FEEL scheme. The analysis shows that the hostilities slow down the convergence of the learning process by introducing a scaling factor and a bias term into the gradient norm. However, we show that all the negative effects vanish as the number of participating devices grows, but at a different rate for each type of channel hostility.
Overview
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The text we found for this paper looks like another document, so the self-check wasn’t run.
Citations
Cites
Deep Residual Learning for Image Recognition
He, K., Zhang, X., Ren, S., & Sun, J. (2016) · 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2016.90 · Cited by 235,474
Communication-Efficient Learning of Deep Networks from Decentralized Data
McMahan, H. B., Moore, E., Ramage, D., Hampson, S., & Arcas, B. A. Y. (2016) · arXiv · doi:10.48550/arxiv.1602.05629 · Cited by 14,215
A Survey on Mobile Edge Computing: The Communication Perspective
Mao, Y., You, C., Zhang, J., Huang, K., & Letaief, K. B. (2017) · IEEE Communications Surveys & Tutorials · doi:10.1109/comst.2017.2745201 · Cited by 5,607 · cited 1×
“(Aggregated quantized local gradient estimates) g̃ = [ g̃[1] , g̃[2] , .”
Self-citationFederated Learning: Strategies for Improving Communication Efficiency
Konečný, J., McMahan, H. B., Yu, F. X., Richtárik, P., Suresh, A. T., & Bacon, D. (2016) · arXiv · doi:10.48550/arxiv.1610.05492 · Cited by 3,766
A Joint Learning and Communications Framework for Federated Learning over Wireless Networks
Chen, M., Yang, Z., Saad, W., Yin, C., Poor, H. V., & Cui, S. (2019) · arXiv · doi:10.48550/arxiv.1909.07972 · Cited by 1,547 · cited 1×
“In [21], a novel bandwidth allocation strategy is proposed for minimizing the model training loss of FEEL via convergence analysis accounting for packet transmission errors.”
MethodA Joint Learning and Communications Framework for Federated Learning Over Wireless Networks
Chen, M., Yang, Z., Saad, W., Yin, C., Poor, H. V., & Cui, S. (2021) · IEEE Transactions on Wireless Communications · doi:10.1109/twc.2020.3024629 · Cited by 1,547 · cited 1×
“In [21], a novel bandwidth allocation strategy is proposed for minimizing the model training loss of FEEL via convergence analysis accounting for packet transmission errors.”
MethodDeep Gradient Compression: Reducing the Communication Bandwidth for Distributed Training
Lin, Y., Han, S., Mao, H., Wang, Y., & Dally, W. J. (2017) · arXiv · doi:10.48550/arxiv.1712.01887 · Cited by 1,134 · cited 1×
“The last approach focuses on update compression by exploiting the sparsity of gradient updates [12], [13] and low-resolution gradient/model-parameter quantization [14]–[16].”
BackgroundFederated Learning via Over-the-Air Computation
Yang, K., Jiang, T., Shi, Y., & Ding, Z. (2020) · IEEE Transactions on Wireless Communications · doi:10.1109/twc.2019.2961673 · Cited by 1,084 · cited 4×
“Existing works on over-the-air aggregation, as well as its application in the context of FEEL [6], [7], [17]–[19] consider analog modulation assuming that the transmitter can modulate the carrier waveform as desired, freely choosing the I/Q coefficients as arbitrary real number.”
BackgroundMethodInfluentialQSGD: Communication-Efficient SGD via Gradient Quantization and Encoding
Alistarh, D., Grubic, D., Li, J., Tomioka, R., & Vojnović, M. (2016) · London School of Economics and Political Science Research Online (London School of Economics and Political Science) · doi:10.48550/arxiv.1610.02132 · Cited by 859
Broadband Analog Aggregation for Low-Latency Federated Edge Learning
Zhu, G., Wang, Y., & Huang, K. (2020) · IEEE Transactions on Wireless Communications · doi:10.1109/twc.2019.2946245 · Cited by 761 · cited 8×
“A performance comparison between digital transmission using OFDMA, BAA developed in [6], and the proposed OBDA is presented in Fig.”
BackgroundMethodInfluentialSelf-citation
Research works printed but not linked: 2
- arunabha, 2010, Fundamentals of LTE
- 2010, Timing Advance (TA) in LTE
Cited by
Items 101–200 of 435
Over-The-Air Federated Learning Over Scalable Cell-Free Massive MIMO
Sifaou, H., & Li, G. Y. (2023) · IEEE Transactions on Wireless Communications · doi:10.1109/twc.2023.3315962 · Cited by 40
On the Differential Privacy in Federated Learning Based on Over-the-Air Computation
Park, S., & Choi, W. (2023) · IEEE Transactions on Wireless Communications · doi:10.1109/twc.2023.3316788 · Cited by 39
“These techniques include communication load reduction [12], [13], [14], analysis of the tradeoff between learning performance and aggregation error [15], power control [16], [17], [18], [19], beamforming [20], and security enhancement [21].”
MethodKou, W.-B., Lin, Q., Tang, M., Xu, S., Ye, R., Leng, Y., Wang, S., Li, G., Chen, Z., Zhu, G., et al. (2025) · IEEE Transactions on Intelligent Transportation Systems · doi:10.1109/tits.2025.3581939 · Cited by 39
“Federated Learning (FL) [10], [11] is an effective solution to make use of diverse data [12], [13] from locationally distributed vehicles for improving the AD model generalization while preserving data privacy.”
Self-citationTurning Channel Noise Into an Accelerator for Over-the-Air Principal Component Analysis
Zhang, Z., Zhu, G., Wang, R., Lau, V. K. N., & Huang, K. (2022) · IEEE Transactions on Wireless Communications · doi:10.1109/twc.2022.3162868 · Cited by 39
Self-citationOver-the-Air Computation with Imperfect Channel State Information
Chen, Y., Zhu, G., & Xu, J. (2022) · IEEE International Workshop on Signal Processing Advances in Wireless Communications (SPAWC) · doi:10.1109/spawc51304.2022.9834026 · Cited by 39
Self-citationLiu, P., Jiang, J., Zhu, G., Cheng, L., Jiang, W., Luo, W., Du, Y., & Wang, Z. (2022) · Frontiers of Information Technology & Electronic Engineering · doi:10.1631/fitee.2100538 · Cited by 38
“Two main lossy compression techniques, namely quantization and sparsification, as well as the combination of them, have been considered in the literature (Alistarh et al., 2017; Stich et al., 2018; Wangni et al., 2018; Amiri and Gündüz, 2020a, 2020b; Basu et al., 2020; Reisizadeh et al., 2020; Zhu et al., 2021).”
BackgroundMethodSelf-citationToward Tailored Models on Private AIoT Devices: Federated Direct Neural Architecture Search
Zhang, C., Yuan, X., Zhang, Q., Zhu, G., Cheng, L., & Zhang, N. (2022) · IEEE Internet of Things Journal · doi:10.1109/jiot.2022.3154605 · Cited by 39
“This motivates recent studies that aim at designing more communication-efficient strategies [33]– [36].”
Self-citationOver-the-Air Federated Learning with Joint Adaptive Computation and Power Control
Yang, H., Qiu, P., Jia, L., & Yener, A. (2022) · IEEE International Symposium on Information Theory (ISIT) · doi:10.1109/isit50566.2022.9834342 · Cited by 38
Federated Learning Over-the-Air by Retransmissions
Hellström, H., Fodor, V., & Fischione, C. (2023) · IEEE Transactions on Wireless Communications · doi:10.1109/twc.2023.3268742 · Cited by 38
“With analog communication, this appears to be inapplicable, but with recent ideas of one-bit digital Over-the-Air Computation, there might be possibilities to explore in this direction [44], [45].”
BackgroundDigital Twin-Assisted Federated Learning Service Provisioning Over Mobile Edge Networks
Zhang, R., Xie, Z., Yu, D., Liang, W., & Cheng, X. (2023) · IEEE Transactions on Computers · doi:10.1109/tc.2023.3337317 · Cited by 38
Li, L., Yang, L., Guo, X., Shi, Y., Wang, H., Chen, W., & Letaief, K. B. (2021) · IEEE Journal on Selected Areas in Communications · doi:10.1109/jsac.2021.3118431 · Cited by 37
“, analog gradient methods [11], quantization schemes [12], and gradient sparsification approaches [13].”
MethodWaveforms for Computing Over the Air: A groundbreaking approach that redefines data aggregation
Pérez-Neira, A. I., Martinez-Gost, M., Şahin, A., Razavikia, S., Fischione, C., & Huang, K. (2025) · IEEE Signal Processing Magazine · doi:10.1109/msp.2024.3500775 · Cited by 38
“…i.e., occurring only at trans-mitters, distortions, like timing errors due to the imperfect clock feeding the digital circuit, residual synchronization errors in the baseband, CFO, phase offset (PO), and phase noise in the radio-frequency front end, can occur at both the transmitter and receiver.”
Self-citationDistributed Learning over a Wireless Network with FSK-Based Majority Vote
Şahin, A., Everette, B., & Hoque, S. S. M. (2021) · International Conference on Advanced Communication Technologies and Networking (CommNet) · doi:10.1109/commnet52204.2021.9641940 · Cited by 36
Wang, B., Fang, J., Li, H., & Zeng, B. (2023) · IEEE Transactions on Signal Processing · doi:10.1109/tsp.2023.3316588 · Cited by 36
“Other works propose to select only a portion of users for uploading their gradients [2], [17], [18], [19], [20], [21] or compress/quantize the local gradients in order to reduce the communication overhead [19], [22], [23], [24], [25], [26], [27], [28], [29], [30], [31].”
BackgroundScalable Hierarchical Over-the-Air Federated Learning
Azimi‐Abarghouyi, S. M., & Fodor, V. (2024) · IEEE Transactions on Wireless Communications · doi:10.1109/twc.2024.3350923 · Cited by 36
SumComp: Coding for Digital Over-the-Air Computation via the Ring of Integers
Razavikia, S., da Silva, J. M. B., & Fischione, C. (2024) · IEEE Transactions on Communications · doi:10.1109/tcomm.2024.3450794 · Cited by 34
Temporal-Structure-Assisted Gradient Aggregation for Over-the-Air Federated Edge Learning
Fan, D., Yuan, X., & Zhang, Y. A. (2021) · IEEE Journal on Selected Areas in Communications · doi:10.1109/jsac.2021.3118348 · Cited by 34
“Various over-the-air FEEL approaches [19]–[23] have been developed to overcome the hostile effects of wireless links between devices and the PS.”
BackgroundPark, S., & Choi, W. (2023) · IEEE Journal on Selected Areas in Communications · doi:10.1109/jsac.2023.3242722 · Cited by 34
“As another approach to reducing communication loads, one-bit compression of each element of the LMUs was proposed in the framework of FL via over-the-air computation in [26].”
MethodData-Driven Participant Selection and Bandwidth Allocation for Heterogeneous Federated Edge Learning
Albaseer, A. M., Abdallah, M. M., Al‐Fuqaha, A., & Erbad, A. (2023) · IEEE Transactions on Systems Man and Cybernetics Systems · doi:10.1109/tsmc.2023.3276329 · Cited by 34
“For instance, to address the latency problem, the broadband analog aggregation (BAA) scheme was proposed in [20], [21], [22], and [23] to reduce the transmission time between edge devices and the orchestrator server by utilizing the superposition property of wireless channels via over-the-air computation (AirComp) [24…”
MethodImperfect CSI: A Key Factor of Uncertainty to Over-the-Air Federated Learning
Yao, J., Yang, Z., Xu, W., Niyato, D. T., & You, X. (2023) · IEEE Wireless Communications Letters · doi:10.1109/lwc.2023.3318555 · Cited by 34
“In [8], the authors considered a bounded CSI error and analyzed the impact of imperfect CSI on the convergence rate of FL.”
BackgroundMethodOver-the-Air Computation Based on Balanced Number Systems for Federated Edge Learning
Şahin, A. (2023) · IEEE Transactions on Wireless Communications · doi:10.1109/twc.2023.3320116 · Cited by 35
Digital Over-the-Air Computation: Achieving High Reliability via Bit-Slicing
Liu, J., Gong, Y., & Huang, K. (2025) · IEEE Transactions on Wireless Communications · doi:10.1109/twc.2025.3540455 · Cited by 33
“1) Reliable multi-symbol AirComp : The state-of-the-art digital AirComp only supports the mapping of each analog value to a single digital symbol (e.g. QAM symbol), lacking the flexibility of traditional digital communication [15], [22], [23], [24], [25], [26].”
Self-citationFedOComp: Two-Timescale Online Gradient Compression for Over-the-Air Federated Learning
Xue, Y., Su, L., & Lau, V. K. N. (2022) · IEEE Internet of Things Journal · doi:10.1109/jiot.2022.3165268 · Cited by 32
“Zhu et al. [11] proposed an FL scheme featuring 1-bit gradient quantization over the broadband wireless channel with the QAM modulation at the devices and over-the-air majority-vote-based gradient decoding at the server.”
BackgroundMethodCHARLES: Channel-Quality-Adaptive Over-the-Air Federated Learning over Wireless Networks
Mao, J., Yang, H., Qiu, P., Liu, J., & Yener, A. (2022) · IEEE International Workshop on Signal Processing Advances in Wireless Communications (SPAWC) · doi:10.1109/spawc51304.2022.9833989 · Cited by 32
Li, H., Wang, R., Zhang, W., & Wu, J. H. (2022) · IEEE Transactions on Wireless Communications · doi:10.1109/twc.2022.3198881 · Cited by 33
“Note that even for a moderate mini-batch size, we expect the central limit theorem to kick in rendering typical gradient noise distributions close to Gaussian, whicn is the same as [7].”
BackgroundMethodResultInfluentialCommunication-Efficient Federated Multitask Learning Over Wireless Networks
Ma, H., Guo, H., & Lau, V. K. N. (2022) · IEEE Internet of Things Journal · doi:10.1109/jiot.2022.3201310 · Cited by 32
“…L ( w ) is the summation of the local objectives L k ( w ) in all users [20], [21], [22], [23], [24], [25], [26], [27], [28], [29], [30], [31], [32], [33], [34], [35], [36], [37], [38], [39], [40], [41], [42], [43], [44] In contrast to conventional FL [20], [21], [22], [23], [24], [25], [26],…”
BackgroundInfluentialExploring Representativity in Device Scheduling for Wireless Federated Learning
Chen, Z., Yi, W., & Nallanathan, A. (2023) · IEEE Transactions on Wireless Communications · doi:10.1109/twc.2023.3281765 · Cited by 31
“Similar to many existing works, e.g., [17], [26], and [27], we set p k = K .”
ResultZhang, X., Tian, H., Ni, W., Yang, Z., & Sun, M. (2023) · IEEE Transactions on Green Communications and Networking · doi:10.1109/tgcn.2023.3307428 · Cited by 31
“…order to satisfy the low-latency and massive-connectivity requirements of FL in IoT networks with a lot of edge devices, a communication and computation co-design approach, called over-the-air FL (AirFL), has attracted considerable attention for achieving fast model aggregation [4], [5], [6], [7].”
BackgroundAsynchronous Wireless Federated Learning With Probabilistic Client Selection
Yang, J., Liu, Y., Chen, F., Chen, W., & Li, C. (2023) · IEEE Transactions on Wireless Communications · doi:10.1109/twc.2023.3337773 · Cited by 31
“Wireless resource allocation and client selection for synchronous FL has been extensively studied [26], [27], [29], [30], [31], [32], [33], [34], [35], [36], with the goal of achieving higher model accuracy or lower energy/latency cost with limited resources.”
Performance-Oriented Design for Intelligent Reflecting Surface-Assisted Federated Learning
Zhao, Y., Wu, Q., Chen, W., Wu, C., & Poor, H. V. (2023) · IEEE Transactions on Communications · doi:10.1109/tcomm.2023.3283799 · Cited by 30
“Specifically, there have been two main lines of research in existing works: one is system optimization from the communication perspective including device scheduling [19], transceiver design [20], and parameter/gradient compressing [21], the other is the inherent design of hyperparameters in the learning procedure, su…”
BackgroundCommunication-Efficient Federated Learning Over-the-Air With Sparse One-Bit Quantization
Oh, J., Lee, D., Won, D., Noh, W., & Cho, S. (2024) · IEEE Transactions on Wireless Communications · doi:10.1109/twc.2024.3432758 · Cited by 29
“That is, the MLP and 5NN are referred from [2], and the 4NN is referred from [27].”
BackgroundMethodInfluentialOver-the-Air Computation with DFT-spread OFDM for Federated Edge Learning
Şahin, A., Everette, B., & Hoque, S. S. M. (2022) · IEEE Wireless Communications and Networking Conference · doi:10.1109/wcnc51071.2022.9771881 · Cited by 29
Semi-Federated Learning: Convergence Analysis and Optimization of a Hybrid Learning Framework
Zheng, J., Ni, W., Tian, H., Gündüz, D., Quek, T. Q. S., & Han, Z. (2023) · IEEE Transactions on Wireless Communications · doi:10.1109/twc.2023.3270908 · Cited by 30
“In the first case, the BS uses all accumulated data samples to calculate the CL gradient, while a decreasing learning rate is adopted in the second case.”
MethodEnergy and Spectrum Efficient Federated Learning via High-Precision Over-the-Air Computation
Li, L., Huang, C., Shi, D., Wang, H., Zhou, X., Shu, M., & Pan, M. (2023) · IEEE Transactions on Wireless Communications · doi:10.1109/twc.2023.3287549 · Cited by 29
“For example, non-orthogonal multiple access and reconfigurable intelligent surface techniques can be integrated into an M-AirComp FL framework, which may improve the FL performance when facing massive connectivity and unfavorable propagation channel.”
MethodOver-the-Air Computing With Imperfect CSI: Design and Performance Optimization
Evgenidis, N. G., Papanikolaou, V. K., Diamantoulakis, P. D., & Karagiannidis, G. K. (2023) · IEEE Transactions on Wireless Communications · doi:10.1109/twc.2023.3330092 · Cited by 30
“Recently, AirComp has been discussed as an attractive technology to be implemented with federated learning in the network edge [20], [21], [22], [23], [24], [25], [26], [27], [28], [29], [30].”
BackgroundDecentralized Edge Learning via Unreliable Device-to-Device Communications
Jiang, Z., Yu, G., Cai, Y., & Jiang, Y. (2022) · IEEE Transactions on Wireless Communications · doi:10.1109/twc.2022.3172147 · Cited by 28
Hierarchical Over-the-Air Federated Edge Learning
Aygun, O., Kazemi, M., Gündüz, D., & Duman, T. M. (2022) · International Conference on Communications · doi:10.1109/icc45855.2022.9839230 · Cited by 27
Client Selection for Federated Bayesian Learning
Yang, J., Liu, Y., & Kassab, R. (2023) · IEEE Journal on Selected Areas in Communications · doi:10.1109/jsac.2023.3242720 · Cited by 27
“, through the joint design of learning and communication such as learning update compression [25], [26], radio resource management [27], [28], [29] and over-the-air computation [30], [31], [32].”
BackgroundCommunication-efficient Federated Learning Through 1-Bit Compressive Sensing and Analog Aggregation
Fan, X., Wang, Y., Huo, Y., & Tian, Z. (2021) · IEEE International Conference on Communications workshops · doi:10.1109/iccworkshops50388.2021.9473872 · Cited by 26
“Secondly, to facilitate power control, most existing works assume that the signals to be transmit from local workers, i.e., local gradients, can be normalized with zero mean and unit variance, e.g., [14]–[17].”
Federated Learning With Erroneous Communication Links
Shirvanimoghaddam, M., Salari, A., Gao, Y., & Guha, A. (2022) · IEEE Communications Letters · doi:10.1109/lcomm.2022.3167094 · Cited by 26
A New Fuzzy Robust Control for Linear Parameter-Varying Systems
Chen, F., Qiu, X., Alattas, K. A., Mohammadzadeh, A., & Ghaderpour, E. (2022) · Mathematics · doi:10.3390/math10183319 · Cited by 26
Wireless Federated Langevin Monte Carlo: Repurposing Channel Noise for Bayesian Sampling and Privacy
Liu, D., & Simeone, O. (2022) · IEEE Transactions on Wireless Communications · doi:10.1109/twc.2022.3215663 · Cited by 26
“It is commonly made in many related papers, such as [5], [6]and [7].”
BackgroundMethodInfluentialBayesian Federated Learning over Wireless NetworksPreprint
Lee, S., Park, C., Hong, S., Eldar, Y. C., & Lee, N. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2012.15486 · Cited by 26
“Over-the-air aggregation is a new paradigm to jointly design the communication and learning system by harnessing the superposition nature of the wireless medium [19]–[24].”
BackgroundCommunication-Efficient and Distributed Learning Over Wireless Networks: Principles and Applications
Park, J., Samarakoon, S., Elgabli, A., Kim, J., Bennis, M., Kim, S.-L., & Debbah, M. (2021) · Proceedings of the IEEE · doi:10.1109/jproc.2021.3055679 · Cited by 25
Denoising Noisy Neural Networks: A Bayesian Approach With Compensation
Shao, Y.-L., Liew, S., & Gündüz, D. (2021) · IEEE Transactions on Signal Processing · doi:10.1109/tsp.2023.3290327 · Cited by 18
“, how to estimate the uncontaminated weights from their noisy counterparts? All prior works [6], [12], [13], [14], [16], [17], [18], [19], [20], [21], [35], [36], [37], [38], [39] take the noisy observations/manifestations of DNNs directly as the estimated DNN weights.”
BackgroundComputing Functions Over-the-Air Using Digital Modulations
Razavikia, S., da Silva, J. M. B., & Fischione, C. (2023) · International Conference on Communications · doi:10.1109/icc45041.2023.10279784 · Cited by 25
“We can add a conventional orthogonal frequency division multiplexing module [12] for all nodes, right after encoder module in Fig.”
BackgroundMethodAccelerating Federated Edge Learning via Topology Optimization
Huang, S., Zhang, Z., Wang, S., Wang, R., & Huang, K. (2022) · IEEE Internet of Things Journal · doi:10.1109/jiot.2022.3164914 · Cited by 24
“hostilities is investigated in [14], where model convergence is”
Self-citationMultiple Parallel Federated Learning via Over-the-Air Computation
Shi, G., Guo, S., Ye, J., Saeed, N., & Dang, S. (2022) · IEEE Open Journal of the Communications Society · doi:10.1109/ojcoms.2022.3194821 · Cited by 24
“Then the author in [26] extended their work to one-bit over-the-air computation FL in [28] and [29], where a new scheme featuring one-bit quantization followed by modulation at edge devices and majority-voting based decoding at the edge server was proposed.”
BackgroundYin, T., Li, L., Lin, W., Ni, T., Liu, Y., Xu, H., & Han, Z. (2023) · IEEE Transactions on Vehicular Technology · doi:10.1109/tvt.2023.3333329 · Cited by 24
“A substantial amount of research seeks to modify the FL algorithm to lower the network’s communication overhead [16], [17], [18], [19], [20].”
BackgroundOFDMA-F²L: Federated Learning With Flexible Aggregation Over an OFDMA Air Interface
Hu, S., Yuan, X., Ni, W., Wang, X., Hossain, E., & Poor, H. V. (2024) · IEEE Transactions on Wireless Communications · doi:10.1109/twc.2023.3334691 · Cited by 24
“In a different yet relevant context, over-the-air (OTA) FL has been investigated to alleviate the communication bottleneck of WFL [28], [29], [30], [31].”
BackgroundReconfigurable Intelligent Surface-Assisted Wireless Federated Learning With Imperfect Aggregation
Sun, P., Liu, E., Ni, W., Wang, R., Xing, Z., Li, B., & Jamalipour, A. (2024) · IEEE Transactions on Communications · doi:10.1109/tcomm.2024.3450605 · Cited by 24
“Moreover, compared to existing one-bit schemes [31], [32], multi-bit quantization offers a higher fault tolerance rate.”
Energy Minimization for Federated Learning with IRS-Assisted Over-the-Air Computation
Hu, Y., Chen, M., Chen, M., Yang, Z., Shikh‐Bahaei, M., Poor, H. V., & Cui, S. R. (2021) · IEEE International Conference on Acoustics Speech and Signal Processing · doi:10.1109/icassp39728.2021.9414785 · Cited by 24
“There has been a number of existing works such as [4, 6], and [9] that investigate the use of AirComp method for FL.”
MethodInfluentialBroadband Digital Over-the-Air Computation for Asynchronous Federated Edge Learning
Zhao, X., You, L., Cao, R., Shao, Y., & Fu, L. (2022) · International Conference on Communications · doi:10.1109/icc45855.2022.9838947 · Cited by 23
Enabling Ubiquitous Non-Orthogonal Multiple Access and Pervasive Federated Learning via STAR-RIS
Ni, W., Liu, Y., Eldar, Y. C., Yang, Z., & Tian, H. (2021) · Globecom · doi:10.1109/globecom46510.2021.9685556 · Cited by 23
“Specifically, the convergence rate of FL was analyzed in [1], [9], whose results were extended to analog aggregation scenarios in [8], and a digital version was developed in [3].”
BackgroundMethodInfluentialAccelerating Distributed Optimization via Over-the-Air Computing
Mitsiou, N. A., Bouzinis, P. S., Diamantoulakis, P. D., Schober, R., & Karagiannidis, G. K. (2023) · IEEE Transactions on Communications · doi:10.1109/tcomm.2023.3286915 · Cited by 23
“We make the assumption that if there exists a bottleneck device with severe path-loss for all transmission rounds, which does not allow channel inversion within the given power budget, this device will be excluded from the optimization procedure [34].”
MethodOver-the-Air Computation in OFDM Systems With Imperfect Channel State Information
Chen, Y., Xing, H., Xu, J., Xu, L., & Cui, S. R. (2023) · IEEE Transactions on Communications · doi:10.1109/tcomm.2023.3347570 · Cited by 23
“…scenario with best-effort computation, the AP aims to compute a given number of function values over a specified time duration (e.g., for gradient/model aggregation in Air-FEEL systems [5], [6], [7]), and the batch of function values is treated as a whole (e.g., for the model update in Air-FEEL).”
BackgroundMethodResultInfluentialDynamic Clustering and Power Control for Two-Tier Wireless Federated Learning
Guo, W. M., Huang, C. Z., Qin, X., Yang, L. X., & Zhang, W. (2023) · IEEE Transactions on Wireless Communications · doi:10.1109/twc.2023.3288606 · Cited by 23
“Instead of introducing additional relays, the authors [29] proposed a cluster-based two-tier wireless FL scheme based on the method proposed in [22], where they select a set of devices as relays in each cluster to aggregate the gradients from the devices within the same cluster and the selected…”
BackgroundMethodInfluentialRandom Orthogonalization for Federated Learning in Massive MIMO Systems
Wei, X., Shen, C., Yang, J., & Poor, H. V. (2023) · IEEE Transactions on Wireless Communications · doi:10.1109/twc.2023.3302335 · Cited by 22
“The approaches to reduce the payload size or communication frequency include sparsification [16], [17] and quantization [18], [19], [20].”
MethodYan, Z., & Li, D. (2024) · IEEE Transactions on Communications · doi:10.1109/tcomm.2024.3362143 · Cited by 22
“…exponent factor. is the channel noise between the i - th device and its neighbors in the j -th communication round, and After the signal y i,j received by the devices, it will be recovered by the channel inversion method, which is widely adopted in FL with analog transmission (e.g., [50], [51]).”
MethodBlind Asynchronous Over-the-Air Federated Edge Learning
Razavikia, S., Peris, J. A., da Silva, J. M. B., & Fischione, C. (2022) · IEEE Globecom Workshops (GC Wkshps) · doi:10.1109/gcwkshps56602.2022.10008588 · Cited by 21
“Here we do not consider this loss in accuracy, but one could refer to [17] for more information about the effect of the quantization error on the inference problem.”
BackgroundMulti-Feature Traffic Prediction Based on Signaling Information for Cellular Network
Liu, J., Tang, X., Zhu, G., Cheng, X., Zhao, L., & Li, H. (2023) · IEEE Transactions on Vehicular Technology · doi:10.1109/tvt.2023.3319351 · Cited by 21
“Then, we introduce the model based on federated learning [27], [28] for application in a distributed environment.”
MethodSelf-citationOver-the-Air Federated Learning via Weighted Aggregation
Azimi‐Abarghouyi, S. M., & Tassiulas, L. (2024) · IEEE Transactions on Wireless Communications · doi:10.1109/twc.2024.3463754 · Cited by 21
Cluster-Based Cooperative Digital Over-the-Air Aggregation for Wireless Federated Edge Learning
Jiang, R., & Zhou, S. (2020) · IEEE/CIC International Conference on Communications in China (ICCC) · doi:10.1109/iccc49849.2020.9238982 · Cited by 21
“Recently, an AirComp scheme using digital modulation is proposed in [5] by integrating AirComp with a specific learning algorithm called signSGD with majority vote [6].”
BackgroundMethodInfluentialDecentralized Federated Learning via Non-Coherent Over-the-Air Consensus
Michelusi, N. (2023) · International Conference on Communications · doi:10.1109/icc45041.2023.10279097 · Cited by 20
Dong, Y., Wang, L., Wang, J., Hu, X., Zhang, H., Yu, F. R., & Leung, V. C. M. (2023) · IEEE Transactions on Wireless Communications · doi:10.1109/twc.2023.3329375 · Cited by 20
“Different from [13], [26], [27], [28], [29], [30], [31], and [32], we consider the non-i.i.d. local gradients and the non-i.i.d. locations of workers.”
BackgroundMethodInfluentialOver-the-Air Majority Vote Computation With Modulation on Conjugate-Reciprocal Zeros
Şahin, A. (2024) · IEEE Transactions on Wireless Communications · doi:10.1109/twc.2024.3456572 · Cited by 20
“Thus, the relative positions of the transmitters to the receiver do not change our analyses as in [20], [21], [36].”
BackgroundLiang, Y., Chen, Q., & Jiang, H. (2024) · IEEE Internet of Things Journal · doi:10.1109/jiot.2024.3486121 · Cited by 20
“To facilitate the convergence analysis, we adopt the assumptions commonly used in existing work [17], [19] regarding the loss functions.”
Over-the-Air Federated Learning with Energy Harvesting Devices
Aygun, O., Kazemi, M., Gündüz, D., & Duman, T. M. (2022) · Globecom · doi:10.1109/globecom48099.2022.10000620 · Cited by 20
Robust Design of Federated Learning for Edge-Intelligent Networks
Qi, Q., & Chen, X. (2022) · IEEE Transactions on Communications · doi:10.1109/tcomm.2022.3175921 · Cited by 19
A Demonstration of Over-the-Air Computation for Federated Edge Learning
Şahin, A. (2022) · IEEE Globecom Workshops (GC Wkshps) · doi:10.1109/gcwkshps56602.2022.10008778 · Cited by 19
“For the model, we consider a convolution neural network (CNN) that consists of two 2D convolutional layers with the kernel size [5, 5], stride [1, 1], and padding [2, 2], where the former layer has 1 input and 16 output channels and the latter one has 16 input and 32 output channels.”
BackgroundMethodFederated Edge Learning with Misaligned Over-The-Air Computation
Shao, Y., Gündüz, D., & Liew, S. C. (2021) · IEEE International Workshop on Signal Processing Advances in Wireless Communications (SPAWC) · doi:10.1109/spawc51858.2021.9593155 · Cited by 19
Over-the-Air Computation over Balanced Numerals
Şahin, A., & Yang, R. J. (2022) · IEEE Globecom Workshops (GC Wkshps) · doi:10.1109/gcwkshps56602.2022.10008587 · Cited by 19
Covert Federated Learning via Intelligent Reflecting Surfaces
Zheng, J., Zhang, H., Kang, J., Gao, L. C., Ren, J., & Niyato, D. T. (2023) · IEEE Transactions on Communications · doi:10.1109/tcomm.2023.3281880 · Cited by 18
Folino, F., Folino, G., Pisani, F. S., Pontieri, L., & Sabatino, P. (2024) · Journal Of Big Data · doi:10.1186/s40537-024-00933-6 · Cited by 18
“Other approaches to curbing overall carbon emissions exploited gradient quantization (see, e.g., [16] addressing the case of over-the-air FL systems), precision and bandwidth optimization [18], and methods for balancing energy consumption with model loss function optimization [17].”
MethodDigital Over-the-Air Federated Learning in Multi-Antenna Systems
Wang, S., Chen, M., Shen, C., Yin, C., & Brinton, C. G. (2024) · IEEE Transactions on Wireless Communications · doi:10.1109/twc.2024.3425732 · Cited by 18
“The authors in [34] designed one-bit quantization and modulation schemes for edge devices.”
BackgroundThreshold-Based Data Exclusion Approach for Energy-Efficient Federated Edge Learning
Albaseer, A. M., Abdallah, M. M., Al‐Fuqaha, A., & Erbad, A. (2021) · IEEE International Conference on Communications workshops · doi:10.1109/iccworkshops50388.2021.9473806 · Cited by 18
Over-the-Air Personalized Federated Learning
Sami, H. U., & Güler, B. (2022) · IEEE International Conference on Acoustics Speech and Signal Processing · doi:10.1109/icassp43922.2022.9746750 · Cited by 19
“To ensure proper synchronization among the users, a truncated channel inversion policy is proposed in [8, 9].”
BackgroundDistributed Learning Over Networks With Graph-Attention-Based Personalization
Tian, Z., Zhang, Z., Yang, Z., Jin, R., & Dai, H. (2023) · IEEE Transactions on Signal Processing · doi:10.1109/tsp.2023.3282071 · Cited by 17
“challenges, some recent progress has been made on the communication-efficient implementation of FL in wireless communication system [24], [25], [26], [27], through resource allocation [24], [25], reducing communication cost per iteration [26] and accelerating convergence [27].”
BackgroundOver-the-Air Federated Edge Learning With Hierarchical Clustering
Aygün, O., Kazemi, M., Gündüz, D., & Duman, T. M. (2024) · IEEE Transactions on Wireless Communications · doi:10.1109/twc.2024.3457591 · Cited by 19
“Other approaches on wireless FL include model compression for the local updates and location-based user scheduling [8], [21], [27], [28].”
BackgroundMethodMulti-cell Non-coherent Over-the-Air Computation for Federated Edge Learning
Adeli, M. H., & Şahin, A. (2022) · International Conference on Communications · doi:10.1109/icc45855.2022.9839037 · Cited by 16
“, OBDA [10]) is that EDs and ESs do not need to utilize the CSI.”
MethodInfluentialAirCon: Over-the-Air Consensus for Wireless Blockchain Networks
Xie, X., Hua, C., Hong, J., Gu, P., & Xu, W. (2023) · IEEE Transactions on Mobile Computing · doi:10.1109/tmc.2023.3292898 · Cited by 16
“AirComp also has been extended for model aggregation in the distributed edge learning system [35], [36], [37], which is shown to reduce the communication latency without significant loss of the learning accuracy [38].”
BackgroundAsynchronous Federated Learning via Over-the-Air Computation in LEO Satellite Networks
Huang, Y., Li, X., Zhao, M., Li, H., & Peng, M. (2024) · IEEE Transactions on Wireless Communications · doi:10.1109/twc.2024.3487986 · Cited by 16
“These studies, such as those by Mohammadi Amiri and Gündüz [25], [26], and by Zhu et al. [27], [28], have demonstrated the effectiveness of these technologies in enhancing communication efficiency and reducing latency in edge learning scenarios.”
Communication-Efficient Distributed SGD Using Random Access for Over-the-Air Computation
Choi, J. (2022) · IEEE Journal on Selected Areas in Information Theory · doi:10.1109/jsait.2022.3188585 · Cited by 16
“Note that if D = L as an extreme case, it is equivalent to 1-bit quantization that is used in [22], where the sign of each element is used for the quantization.”
BackgroundMethodInfluentialOptimized Power Control for Over-the-Air Federated Averaging With Data Privacy Guarantee
Jiang, J., Han, K., Du, Y., Zhu, G., Wang, Z., & Cui, S. R. (2022) · IEEE Transactions on Vehicular Technology · doi:10.1109/tvt.2022.3210335 · Cited by 16
“For instance, a source-coding algorithm exploiting gradient sparsification was proposed to compress updates in [7], while a digital solution to implement Air-FEEL was proposed in [8].”
MethodSelf-citationJoint Compression and Deadline Optimization for Wireless Federated Learning
Zhang, M., Li, Y., Liu, D., Jin, R., Zhu, G., Zhong, C., & Quek, T. Q. S. (2023) · IEEE Transactions on Mobile Computing · doi:10.1109/tmc.2023.3344108 · Cited by 16
“A lot of quantization schemeshadbeendevelopedforthedistributedlearningsettings, e.g.,SignSGD[4],[5],[6],[7],TernSGD[8],QSGD[9],etc. Bysparsification,theinsignificantelementswillbedropped, suchthatthecompressioncanbeachievedbyonlyencoding 1536-1233 © 2023 IEEE.”
BackgroundSelf-citationData Partition and Rate Control for Learning and Energy Efficient Edge Intelligence
Li, X., Wang, S., Zhu, G., Zhou, Z., Huang, K., & Gong, Y. (2022) · IEEE Transactions on Wireless Communications · doi:10.1109/twc.2022.3173262 · Cited by 16
Self-citationSecure and Private Over-the-Air Federated Learning: Biased and Unbiased Aggregation Design
Yan, N., Wang, K., Zhi, K., Pan, C., Chai, K. K., & Poor, H. V. (2025) · IEEE Transactions on Wireless Communications · doi:10.1109/twc.2025.3550159 · Cited by 15
“Our subsequent convergence analyses of SP-OTA-FL rely on several assumptions on the loss function and gradients as follows, which are widely adopted for distributed optimization [19], [42], [43], [44].”
Optimized Power Control for Over-the-Air Federated Edge Learning
Cao, X., Zhu, G., Xu, J., & Cui, S. R. (2021) · International Conference on Communications · doi:10.1109/icc42927.2021.9500664 · Cited by 15
“Furthermore, to allow Air-FEEL compatible with digital chips embedded in modern edge devices, Air-FEEL based on digital modulation was proposed in [14] featuring one-bit quantization and modulation at the edge devices and majority-vote based decoding at the edge server.”
MethodSelf-citationDistributed Joint Resource Optimization for Federated Learning Task Distribution
Zhang, C., Dong, M., & Ota, K. (2023) · IEEE Transactions on Green Communications and Networking · doi:10.1109/tgcn.2023.3262516 · Cited by 15
“Devices in a single slice provide corresponding computation, communication, and storage resources for FL tasks [35].”
BackgroundROAR-Fed: RIS-Assisted Over-the-Air Adaptive Resource Allocation for Federated Learning
Mao, J., & Yener, A. (2023) · International Conference on Communications · doi:10.1109/icc45041.2023.10278624 · Cited by 15
“Similar to [17], we apply the Taylor expansion: h i t ĥt = 1 1+ ∆t ht = 1− ∆ i t ht +O(( i t ht )(2)).”
MethodFederated Learning from Heterogeneous Data via Controlled Air Aggregation with Bayesian Estimation
Gafni, T., Cohen, K., & Eldar, Y. C. (2024) · IEEE Transactions on Signal Processing · doi:10.1109/tsp.2024.3351469 · Cited by 15
“In the extension here, we assume that the participating entities have channel state information (CSI), i.e., knowledge of the fading coeffi-cients, as commonly assumed in wireless communications and particularly OTA FL (see e.g., [37], [54], [55] and subsequent studies).”
BackgroundPerformance Optimization of Federated Learning over Mobile Wireless Networks
Chen, M., Poor, H. V., Saad, W., & Cui, S. R. (2020) · IEEE International Workshop on Signal Processing Advances in Wireless Communications (SPAWC) · doi:10.1109/spawc48557.2020.9154300 · Cited by 15
Harnessing wireless channels for scalable and privacy-preserving federated learning
Elgabli, A., Park, J., Issaid, C. B., & Bennis, M. (2021) · IEEE Transactions on Communications · doi:10.1109/tcomm.2021.3078783 · Cited by 14
Liu, Y., Chen, K., & Zhu, L. (2024) · Computer Networks · doi:10.1016/j.comnet.2024.110471 · Cited by 14
Fang, W., Han, D.-J., & Brinton, C. G. (2025) · IEEE Transactions on Networking · doi:10.1109/ton.2025.3538661 · Cited by 15
“The authors of [30], [31], and [32] proposed to utilize AirComp to accelerate the convergence of FedAvg in single-cell networks.”
Channel-Driven Monte Carlo Sampling for Bayesian Distributed Learning in Wireless Data Centers
Liu, D., & Simeone, O. (2021) · IEEE Journal on Selected Areas in Communications · doi:10.1109/jsac.2021.3118406 · Cited by 14
CFLIT: Coexisting Federated Learning and Information Transfer
Lin, Z., Liu, H., & Zhang, Y. A. (2023) · IEEE Transactions on Wireless Communications · doi:10.1109/twc.2023.3263148 · Cited by 14
“In the former, local models are quantized into finite-bit digits to reduce the amount of data in model uploading [21], [22].”
MethodJoint Design of Coding and Modulation for Digital Over-the-Air Computation
Xie, X., Hua, C., Hong, J., Wei, Y., & Gu, P. (2026) · IEEE Internet of Things Journal · doi:10.1109/jiot.2026.3655679 · Cited by 14
“That is, the destructive interference from di ff erent transmitters is turned into a constructive one by appropriately matching the structure of the wireless channel and the function of the signals to be computed, which greatly reduces the communication and computing costs of wireless networks.”
Decentralized Descent Optimization With Stochastic Gradient Signs for Device-to-Device Networks
Phuong, T. T., & Phong, L. T. (2021) · IEEE Wireless Communications Letters · doi:10.1109/lwc.2021.3087156 · Cited by 14
Over-the-Air Federated Averaging With Limited Power and Privacy Budgets
Yan, N., Wang, K., Pan, C., Chai, K. K., Shu, F., & Wang, J. (2023) · IEEE Transactions on Communications · doi:10.1109/tcomm.2023.3338738 · Cited by 13
“On the other hand, over-the-air FL (OTA-FL) [1], [11], [16], [17], [18], [19], [20], [21], [22], [23], [24] is promising to alleviate the above-mentioned communication bottleneck and reduce the communication and computation latency.”
Background
Metadata
- Authors
- Zhu, G., Du, Y., Gunduz, D., & Huang, K.
- Affiliations
- Shenzhen Research Institute of Big Data (CN)
- University of Hong Kong (HK)
- Venue
- IEEE Transactions on Wireless Communications
- Published
- 2021-03
- Volume
- 20
- Issue
- 3
- Pages
- 2120-2135
- ISSN
- 1536-1276 (ISSN-L)
- 1558-2248
- OpenAlex
- W3108590825
- Cited by
- 376
- References
- 37
- Citations by year
- 2026: 30 · 2025: 45 · 2024: 61 · 2023: 93 · 2022: 68 · 2021: 58
- FWCI
- 23.57
- Type
- journal article
- Language
- en
- Open access
- green
- arxiv.org
- Locations
- IEEE Transactions on Wireless Communications (publishedVersion)
- arXiv (Cornell University) (submittedVersion)
- IRIS UNIMORE (University of Modena and Reggio Emilia) (submittedVersion, cc-by)
- Spiral (Imperial College London) (submittedVersion)
- arXiv (Cornell University)
- Funders
- National Natural Science Foundation of China (62001310)
- National Key Research and Development Program of China (2018YFB1800800)
- Guangdong Province Key Area Research and Development Program (2018B030338001)
- Shenzhen Peacock Plan (KQTD2015033114415450)
- European Research Council (ERC) through the Starting Grant BEACON (677854)
- U.K. EPSRC (EP/T023600/1)
- Guangdong Basic and Applied Basic Research Foundation (2019B1515130003)
- Hong Kong Research Grants Council (17208319, 17209917)
- Innovation and Technology Fund (GHP/016/18GD)
- Topics
- Physical Sciences › Computer Science › Artificial Intelligence › Privacy-Preserving Technologies in Data
- Physical Sciences › Computer Science › Computer Networks and Communications › IoT and Edge/Fog Computing
- Physical Sciences › Computer Science › Computer Networks and Communications › Distributed Sensor Networks and Detection Algorithms
- Keywords
- federated edge learning, over-the-air aggregation, 1-bit quantization, convergence analysis, channel estimation error, fading, channel noise, communication bottleneck, edge devices
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.