Federated Learning via Over-the-Air Computation
Open accessThe DOI matches its Crossref record; 61 of the 62 references it cites were found in a registry or an index.
Abstract
The stringent requirements for low-latency and privacy of the emerging high-stake applications with intelligent devices such as drones and smart vehicles make the cloud computing inapplicable in these scenarios. Instead, edge machine learning becomes increasingly attractive for performing training and inference directly at network edges without sending data to a centralized data center. This stimulates a nascent field termed as federated learning for training a machine learning model on computation, storage, energy and bandwidth limited mobile devices in a distributed manner. To preserve data privacy and address the issues of unbalanced and non-IID data points across different devices, the federated averaging algorithm has been proposed for global model aggregation by computing the weighted average of locally updated model at each selected device. However, the limited communication bandwidth becomes the main bottleneck for aggregating the locally computed updates. We thus propose a novel over-the-air computation based approach for fast global model aggregation via exploring the superposition property of a wireless multiple-access channel. This is achieved by joint device selection and beamforming design, which is modeled as a sparse and low-rank optimization problem to support efficient algorithms design. To achieve this goal, we provide a difference-of-convex-functions (DC) representation for the sparse and low-rank function to enhance sparsity and accurately detect the fixe…
Overview
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
In the citing paper
“For instance, over-the-air federated learning (FL) provides a collaborative ML framework to train a global statistical model over wireless networks without accessing edge devices' private raw data [31].”
From Edge artificial intelligence for 6g: vision, enabling technologies, and applications (2022) · cites this paper as [31]
In the citing paper
“Specifically, the transmitted signals at edge devices are first multiplied by the fading channels and then superposed over-the-air with additive channel noise, resulting in a noisy weighted sum of transmitted signals [31].”
From Edge artificial intelligence for 6g: vision, enabling technologies, and applications (2022) · cites this paper as [31]
In the citing paper
“Specifically, for edge training systems, we need to jointly optimize the subcarrier and bandwidth allocation [88], [90], [241], transmit power and receive beamforming [31], [48], [178], passive beamforming at RIS [48], [178], device selection [31], [242] and activity detection [76], local updates computation [92], and global aggregation frequency control [130], thereby reducing the optimality gap”
From Edge artificial intelligence for 6g: vision, enabling technologies, and applications (2022) · cites this paper as [31]
In the citing paper
“For instance, the group sparsity can represent the combinatorial variables for edge devices selection in FL [31], edge devices activity detection [76], and inference tasks selection [53].”
From Edge artificial intelligence for 6g: vision, enabling technologies, and applications (2022) · cites this paper as [31]
In the citing paper
“Most of the resource allocation problems in edge AI need to solve a series of nonconvex optimization problems, e.g., nonconvex sparse optimization for device selection in wireless FL, nonconvex quadratic programming for transceiver design in over-the-air FL [31], lowrank matrix optimization for interference management in edge device distributed inference [52], and unit modulus constrained phase”
From Edge artificial intelligence for 6g: vision, enabling technologies, and applications (2022) · cites this paper as [31]
2 more claims about this paper
Create a free account to see every claim other papers make about this one.
Citations
Cites
Hao, X.-B., & Zhang, G.-G. (2017) · Encycl. Semantic Comput. Robotic Intell. · doi:10.1142/s2425038416300184 · Cited by 86,401 · cited 1×
“THE astounding growth in data volume promotes widespread artificial intelligent applications such as image recognition and natural language processing [1], thanks to”
BackgroundLeCun, Y., Bengio, Y., & Hinton, G. (2015) · Nature · doi:10.1038/nature14539 · Cited by 84,126
Boyd, S., & Vandenberghe, L. (2004) · doi:10.1017/cbo9780511804441 · Cited by 56,601
Vorontsova, E., Hildebrand, R., Gasnikov, A. V., & Stonyakin, F. (2010) · Computer Vision · doi:10.1007/978-1-4614-5257-7_6 · Cited by 39,270 · cited 2×
“The resulting problem is known as the sum-of-infeasibilities in the literature of optimization [39].”
Learning Multiple Layers of Features from Tiny Images
Krizhevsky, A. (2024) · doi:10.57702/zp44cu3g · Cited by 23,395
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 · cited 7×
“In particular, the federated averaging (FedAvg) algorithm [12] turns out to be a promising way to efficiently average the locally updated model at each device with unbalanced and non-IID data, thereby reducing the number of communication rounds between the center node and the end devices.”
InfluentialFederated Machine Learning: Concept and Applications
Yang, Q., Liu, Y., Chen, T., & Tong, Y. (2019) · ACM Transactions on Intelligent Systems and Technology · doi:10.1145/3298981 · Cited by 6,322
Han, S., Mao, H., & Dally, W. J. (2015) · arXiv · doi:10.48550/arxiv.1510.00149 · Cited by 4,662 · cited 1×
“To address this issue, there is a growing body of recent works to reduce the storage overhead, time and power consumption in the inference process using the model compression methods via hardware and software co-design [7], [8].”
MethodTowards Federated Learning at Scale: System Design
Bonawitz, K., Eichner, H., Grieskamp, W., Huba, D., Ingerman, A., Ivanov, V., Kiddon, C., Konečný, J., Mazzocchi, S., McMahan, H. B., et al. (2019) · arXiv · doi:10.48550/arxiv.1902.01046 · Cited by 2,334 · cited 2×
“It has promising applications [14], [15] such as smart retail, smart healthcare, financial services, mobile content predictions, etc.”
BackgroundFederated Learning with Non-IID Data
Zhao, Y., Li, M., Lai, L., Suda, N., Civin, D., & Chandra, V. (2018) · arXiv · doi:10.48550/arxiv.1806.00582 · Cited by 2,077 · cited 1×
“, the data is generated by distinct distributions across different devices), imposes significant statistical challenges to fit a mode from the non-IID data [13], [17].”
Background
Research works printed but not linked: 2
- rockafellar, 2015, Convex Analysis
- tao, 1997, Convex analysis approach to DC programming: Theory, algorithms and applications, Acta Math Vietn, 22, 289
Cited by
Items 401–500 of 999
Communication-Efficient Federated Learning Over MIMO Multiple Access Channels
Jeon, Y., Amiri, M. M., & Lee, N. (2022) · IEEE Transactions on Communications · doi:10.1109/tcomm.2022.3198433 · Cited by 23
“The most widely adopted approach for federated learning over the wireless MAC is over-the-air computation which exploits the superposition property of the wireless MAC [22].”
The Magic of Superposition: A Survey on Simultaneous Transmission Based Wireless Systems
Altun, U., Kurt, G. K., & Ozdemır, E. (2022) · IEEE Access · doi:10.1109/access.2022.3195056 · Cited by 22
“[177] Reduces the convergence rate of the learning algorithm by considering device selection and beamforming.”
BackgroundInfluentialCSIT-Free Model Aggregation for Federated Edge Learning via Reconfigurable Intelligent Surface
Liu, H., Yuan, X., & Zhang, Y. A. (2021) · IEEE Wireless Communications Letters · doi:10.1109/lwc.2021.3102601 · Cited by 20
“For simplicity of exposition, we assume that the channel coefficients remain invariant during the training process by following [2], [9], [10].”
BackgroundMethodWen, D., Jeon, K. J., Bennis, M., & Huang, K. (2021) · IEEE Transactions on Wireless Communications · doi:10.1109/twc.2021.3092075 · Cited by 19
“ess techniques, called over-the-air computation (AirComp), that supports fast aggregation of local updates over-the-air by exploiting the waveform superposition property of a multi-access channel [5]–[7]. An alternative approach is to design vector quantization techniques for compressing highdimensional stochastic gra…”
MethodPrivate Over-the-Air Federated Learning at Band-Limited Edge
Tao, Y., Chen, S., Zhang, C., Wang, D., Yu, D., Cheng, X., & Dressler, F. (2024) · IEEE Transactions on Mobile Computing · doi:10.1109/tmc.2024.3411295 · Cited by 20
“As a result, federated learning based on AirComp, i.e., over-the-air federated learning (OTA-FL), has recently gained considerable interest in edge intelligence research [13], [14], [15], [16], [17], [18].”
BackgroundCollaborative Edge AI Inference Over Cloud-RAN
Zhang, P., Wen, D., Zhu, G., Chen, Q., Han, K., & Shi, Y. (2024) · IEEE Transactions on Communications · doi:10.1109/tcomm.2024.3388488 · Cited by 22
“To collect all local feature vectors at the CP, the technique of AirComp is adopted to allow all devices transmitting their local feature vectors to the RRHs via a shared multiple access channel, which can significantly enhance the communication efficiency [42], [43], [44].”
MethodSelf-citationAlotaibi, B. K., Khan, F. A., & Mahmood, S. (2024) · Applied Sciences · doi:10.3390/app14072720 · Cited by 30
“…Over-the-air computation [179,197] is a technique used by exploiting the waveform superposition property, which directly obtains the aggregated model. computation [179,197] is a technique used by exploiting the waveform superposition property, which directly obtains the aggregated model.”
Communication-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 ) where 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],…”
BackgroundInfluentialWen, W., Jia, Y., & Xia, W. (2022) · China Communications · doi:10.23919/jcc.2022.01.010 · Cited by 16
“Specifically, based on the principle of over-the-air computation, the authors in [7] proposed a joint client scheduling and beamforming problem to maximize the number of scheduled clients at each iteration of FL.”
BackgroundSoft Delivery: Survey on a New Paradigm for Wireless and Mobile Multimedia Streaming
Fujihashi, T., Koike–Akino, T., & Watanabe, T. (2023) · ACM Computing Surveys · doi:10.1145/3607139 · Cited by 15
“…signal 𝑥 ⟨ d ⟩ 𝑖 and output vector signal of the soft delivery scheme 𝑥 ⟨ a ⟩ 𝑖 as follows: The BPSK-modulated symbol and the analog-modulated symbol are scaled by 𝑃 d and 𝑔 𝑖 , respectively. where 𝑏 𝑖 ∈ X = {± 1 } is the BPSK-modulated symbol and 𝚥 = √− 1 denotes the imaginary unit.”
Over-the-Air Statistical Estimation
Lee, C.-Z., Barnes, L., & Özgür, A. (2020) · Global Communications Conference · doi:10.1109/globecom42002.2020.9322345 · Cited by 16
“Such analog methods have seen wider development efforts [19], [20], including with privacy considerations in mind [21]–[24].”
BackgroundAI-Based Fluid Antenna Design for Client Selection in Over-the-Air Federated Learning
Ahmadzadeh, M., Pakravan, S., Hodtani, G. A., Zeng, M., Ye, Q. J., Chouinard, J., & Rusch, L. A. (2025) · IEEE Internet of Things Journal · doi:10.1109/jiot.2025.3594141 · Cited by 14
“Performance is compared between the FA system [15], [23], [29], [37] and the FPA system [23], [38] using a uniform location vector x = [ ( X / [ N + 1] ), . . . , ( NX / [ N + 1] ) ] T .”
Over-the-Air Clustered Federated Learning
Sami, H. U., & Güler, B. (2024) · IEEE Transactions on Wireless Communications · doi:10.1109/twc.2023.3345730 · Cited by 15
“Digital Object Identifier 10.1109/TWC.2023.3345730 of the multi-access channel [4], [5], [6].”
BackgroundMethodInfluentialA Survey on Federated Analytics: Taxonomy, Enabling Techniques, Applications and Open Issues
Wang, Z., Ji, H., Zhu, Y., Wang, D., & Han, Z. (2025) · IEEE Communications Surveys & Tutorials · doi:10.1109/comst.2025.3558755 · Cited by 24
“In addition, the over-the-air [112] technique brings efficiency as well as new challenges to FA and FL, such as signal interference and the requirement for precise synchronization.”
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
“The aggregation error, convergence rate and covertness lie with the number of participant devices and wireless resource allocation [7], [13].”
BackgroundInfluentialRiemannian Low-Rank Model Compression for Federated Learning With Over-the-Air Aggregation
Xue, Y., & Lau, V. K. N. (2023) · IEEE Transactions on Signal Processing · doi:10.1109/tsp.2023.3284381 · Cited by 18
“For example, over-the-air (OTA) aggregation [19–21] has been proposed to reduce the radio resource required for FL training with a large number of IoT devices [22].”
BackgroundRobust Federated Learning Over Noisy Fading Channels
Shah, S. M., Su, L., & Lau, V. K. N. (2022) · IEEE Internet of Things Journal · doi:10.1109/jiot.2022.3230452 · Cited by 20
“The over-the-air computation model and its suitability for FL has been previously considered in many works, including [24], [25], [27].”
BackgroundMethodInfluentialAhn, J.-H., Bennis, M., & Kang, J. (2022) · IEEE Transactions on Green Communications and Networking · doi:10.1109/tgcn.2022.3186538 · Cited by 19
Budget Distributed Support Vector Machine for Non-ID Federated Learning Scenarios
Navia-Vázquez, A., Díaz-Morales, R., & Fernández-Díaz, M. (2022) · ACM Transactions on Intelligent Systems and Technology · doi:10.1145/3539734 · Cited by 20
“Other methods are designed for linear models only and also require the exchange of local models/solutions, such as the Alternating Direction Method of Multipliers [15, 16], Distributed Dual Ascend [20], Distributed Block Minimization [36], the resilient approach presented in [56, 57, 59] or the approaches described in…”
MethodRobust Design of Federated Learning for Edge-Intelligent Networks
Qi, Q., & Chen, X.-M. (2022) · IEEE Transactions on Communications · doi:10.48550/arxiv.2205.06955 · Cited by 15
“1) Sparsity Inducing: For the untractable and non-convex sparse objective function in the form of l0-norm , we utilize a commonly used l1-relaxation in the form of l1-norm [28], which is based on minimizing the sum of the infeasibilities [36].”
BackgroundMethodInfluentialSparse Optimization for Green Edge AI Inference
Yang, X., Hua, S., Shi, Y., Wang, H., Zhang, J., & Letaief, K. B. (2020) · Journal of Communications and Information Networks · doi:10.23919/jcin.2020.9055106 · Cited by 19
“As for the edge training phase, the training of AI models can be performed on cloud, edge or end devices[13], however, this is beyond the scope of this work.”
Self-citationFederated Learning over Noisy Channels
Wei, X., & Shen, C. (2021) · ICC 2021 - IEEE International Conference on Communications · doi:10.1109/icc42927.2021.9500833 · Cited by 14
“Joint communication and computation is investigated [13], [14], [22], [32].”
BackgroundOptimal Power Control and CSI Acquisition for Over-the-Air Computation in OFDM System
Xie, X., Hua, C., Hong, J., & Xu, W. (2023) · IEEE Transactions on Wireless Communications · doi:10.1109/twc.2023.3333326 · Cited by 18
“They also considered the application of AirComp in other scenarios, e.g., the federated learning (FL) scenario [30], [31] and split inference scenario [32].”
BackgroundLi, F., Ye, Q., Fapi, E. T., Sun, W., & Jiang, Y. (2023) · IEEE Transactions on Vehicular Technology · doi:10.1109/tvt.2023.3298565 · Cited by 14
“The spatial diversity and multiplexing effects in multi-antenna systems bring further enhancement to the computation accuracy of AirComp [18], [19], [20], [21], [22].”
BackgroundMethodInfluentialCommunication and Energy Efficient Wireless Federated Learning With Intrinsic Privacy
Zhang, Z., Guo, Y., Fang, Y. M., & Gong, Y. (2023) · IEEE Transactions on Dependable and Secure Computing · doi:10.1109/tdsc.2023.3341788 · Cited by 11
“Over-the-aircomputation(AirComp)providesapromising solutiontoaddressingbothoftheaforementionedspectrumand privacychallengesinanintegratedmannerbyachievingscalable andbandwidth-efficientmodelupdateaggregationinwireless FL[9].”
BackgroundMethodInfluentialPersonalizing Federated Learning with Over-The-Air Computations
Chen, Z., Li, Z., Yang, H. H., & Quek, T. Q. S. (2023) · IEEE International Conference on Acoustics Speech and Signal Processing · doi:10.1109/icassp49357.2023.10095533 · Cited by 13
“As a result, for a digital communication based-FEEL system run over the resource-constrained network, the limited communication bandwidth would inevitably constrain its scalability, since every selected client in each round requires an orthogonal sub-channel to upload the updated information [5,6].”
Dara, S., Kanapala, A., Babu, A. R., Dhamercherala, S., Vidyarthi, A., & Agarwal, R. (2022) · Computers & Electrical Engineering · doi:10.1016/j.compeleceng.2022.108266 · 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
“To relieve the communication burden of FL model uploading and aggregation, over-the-air computation [5] has been introduced into the FL model aggregation process [6].”
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
“In [13] and [14], it is assumed that the channel state information (CSI) for each edge device (ED) is available at the edge server (ES).”
BackgroundNothing Wasted: Full Contribution Enforcement in Federated Edge Learning
Hu, Q., Wang, S., Xiong, Z., & Cheng, X. (2021) · IEEE Transactions on Mobile Computing · doi:10.1109/tmc.2021.3123195 · Cited by 19
“Leveraging on the over-the-air computation, several studies [9], [10], [11], [12] achieved more efficient FL aggregation by taking advantage of the superposition of signals in the wireless multiple access channel.”
MethodAI on the Edge: Architectural Alternatives
John, M. M., Olsson, H. H., & Bosch, J. (2020) · Euromicro Conference on Software Engineering and Advanced Applications (SEAA) · doi:10.1109/seaa51224.2020.00015 · Cited by 42
“Cloud computing facilitates the training and inference process for typical ML applications with the support of centralized data centers with high capacity for large scale data storage and processing [17].”
BackgroundNetworked Exponential Families for Big Data Over Networks
Jung, A. (2020) · IEEE Access · doi:10.1109/access.2020.3033817 · Cited by 15
Scheduling for Cellular Federated Edge Learning With Importance and Channel Awareness
Ren, J., He, Y., Wen, D., Yu, G., Huang, K., & Guo, D. (2020) · IEEE Transactions on Wireless Communications · doi:10.1109/twc.2020.3015671 · Cited by 243
“Based on this, the authors of [27] considered joint device selection and beamforming design in FEEL.”
BackgroundIncentivized Federated Learning: A Survey
Nair, A. K., Çöleri, S., Sahoo, J., Cenkeramaddi, L. R., & Raj, E. D. (2025) · IEEE Transactions on Emerging Topics in Computational Intelligence · doi:10.1109/tetci.2025.3547609 · Cited by 16
“The optimization of aggregation approaches [9], [10] and clustering [11], [12], [13] have been proposed as a remedy for non-independent and identically distributed (non-IID) data problems in FL systems.”
Trust management for IoT devices based on federated learning and blockchain
Wang, L., Li, J., & Zuo, L. (2024) · The Journal of Supercomputing · doi:10.1007/s11227-024-06715-4 · Cited by 22
“Yang et al. [26] used the signal superposition property of wireless multiple-access channels and the principle of over-the-air computation to design a novel rapid model aggregation method for federated learning.”
Federated Dynamical Low-Rank Training with Global Loss Convergence Guarantees
Schotthöfer, S., & Laiu, M. P. (2024) · arXiv.org · doi:10.48550/arxiv.2406.17887 · Cited by 13
“The latter could be addressed by sparse training [29, 41] and transfer learning [5].”
A Comprehensive Review on Machine Learning-based Approaches for Next Generation Wireless Network
Paul, S. (2024) · SN Computer Science · doi:10.1007/s42979-024-02831-3 · Cited by 17
“…for a good number of devices, it is a challenging task to implement by using conventional channel access mechanisms as they suffer due to significant delay. combat such challenges, a computation approach of over-the-air can be applied in a superposition way in wireless transmission [164, 165].”
Adaptive Federated Learning Over the Air
Wang, C., Chen, Z., Παππάς, Ν., Yang, H. H., Quek, T. Q. S., & Poor, H. V. (2025) · IEEE Transactions on Signal Processing · doi:10.1109/tsp.2025.3585002 · Cited by 10
“In response, a few works [30], [31], [32], [33] have suggested improving the signal quality in A-OTA FL systems through beamforming designs.”
Federated Learning over LEO Satellite
Wang, Y., Zou, C., Wen, D., & Shi, Y. (2022) · IEEE Globecom Workshops (GC Wkshps) · doi:10.1109/gcwkshps56602.2022.10008719 · Cited by 17
“We assume that xk is normalized with unit variance [7].”
BackgroundMethodSelf-citationOver-the-Air Federated Learning Exploiting Channel Perturbation
Hamidi, S. M., Mehrabi, M., Khandani, A. K., & Gündüz, D. (2022) · IEEE International Workshop on Signal Processing Advances in Wireless Communications (SPAWC) · doi:10.1109/spawc51304.2022.9833972 · Cited by 17
“To quantify the performance of the proposed method of data aggregation, similarly to [4], we use mean square error (MSE) that measures the distortion between As discussed in [14], a smaller MSE yields a larger learning convergence rate.”
BackgroundMethodDelay minimization for intelligent reflecting surface assisted federated learning
Huang, N., Wang, T.-S., Wu, Y., Bi, S., Qian, L., & Lin, B. (2022) · China Communications · doi:10.23919/jcc.2022.04.016 · Cited by 13
“(2) (Efficient model aggregation): [18] proposes a novel fast global model aggregation approach by combining FL with over-the-air computation, which explores the superposition property of a wireless multiple-access channel.”
BackgroundRobust federated learning for edge-intelligent networks
Gao, Z., Chen, X., & Shao, X. (2022) · Science China Information Sciences · doi:10.1007/s11432-020-3251-9 · Cited by 13
Son, H. M., Kim, M. H., & Chung, T. (2021) · arXiv (Cornell University) · doi:10.3390/app12199943 · Cited by 5
“2020), IoT devices (Yang et al. 2020), and even as a privacy-preserving method for medical records (Brisimi et al.”
BackgroundMethodData 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
Spectrum and Computing Resource Management for Federated Learning in Distributed Industrial IoT
Zhang, W., Yang, D., Wu, W. R., Peng, H., Zhang, H., & Shen, X. (2021) · IEEE International Conference on Communications workshops · doi:10.1109/iccworkshops50388.2021.9473515 · Cited by 15
“[5] aimed at maximizing the number of participating devices in each FL epoch, while more participating devices may result in higher communication cost due to the excessive parameter transmission.”
BackgroundRobust Design for Reconfigurable Intelligent Surface Assisted Over-the-Air Computation
An, Q., Zhou, Y., & Shi, Y. (2021) · IEEE Wireless Communications and Networking Conference · doi:10.1109/wcnc49053.2021.9417576 · Cited by 13
“Moreover, AirComp was also applied to support the global model aggregation in the federated learning framework [8].”
BackgroundMethodSelf-citationUAV-Assisted Over-the-Air Computation
Fu, M., Zhou, Y., Shi, Y., Wang, T., & Chen, W. (2021) · International Conference on Communications · doi:10.1109/icc42927.2021.9500918 · Cited by 16
“A common design metric that has been widely adopted in AirComp is the mean-squared error (MSE) between the estimated function value and the target function value [4]–[6], [10].”
BackgroundMethodSelf-citationOptimized 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
“In parallel, a joint design of device scheduling and beamforming in a multi-antenna system was presented in [12] to accelerate Air-FEEL.”
BackgroundMethodOver-the-Air Computing for 6G - Turning Air into a ComputerPreprint
Zhu, G., Xu, J., & Huang, K.-B. (2020) · arXiv.org · doi:10.48550/arxiv.2009.02181 · Cited by 13
Federated Over-Air Subspace Tracking From Incomplete and Corrupted Data
Narayanamurthy, P. K., Vaswani, N., & Ramamoorthy, A. (2022) · IEEE Transactions on Signal Processing · doi:10.1109/tsp.2022.3186540 · Cited by 16
“In this alternate paradigm, the summation can be performed “over-air” using the superposition property of the wireless channel and the summed aggregate (or its processed version) can then be broadcasted to all the nodes [9, 19, 20].”
A Hybrid Federated Learning Framework for Task-Oriented Semantic Communication
Sun, H., Ni, W., Tian, H., Zheng, J., Nie, G., & Niyato, D. T. (2025) · IEEE Internet of Things Journal · doi:10.1109/jiot.2025.3553504 · Cited by 13
“By harnessing the superposition property of wireless multiple access channel, Yang et al. [30] proposed an over-the-air computation (AirComp) method to enable efficient gradient aggregation among UEs over the same time-frequency resources.”
Over-the-Air ODE-Inspired Neural Network for Dual Task-Oriented Semantic Communications
Liu, M., An, J., Huang, C., & Yuen, C. (2025) · IEEE Transactions on Cognitive Communications and Networking · doi:10.1109/tccn.2025.3569604 · Cited by 11
“…for fast global model aggregation via exploring the superposition property of the wireless MACs, including optimizing device selection and beamforming for enhanced participation [8], accelerating the convergence rate [9], [10], and improving both training accuracy and convergence speed [11].”
Asynchronous 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
“…utilizes the signal superposition to avoid the interchannel interference, the signal is still affected by other factors, resulting in distortion [24], which can hardly be detected and corrected by the receiver for it just attains the overlap of signals from multiple transmitters and the…”
Novel Over-the-Air Federated Learning via Reconfigurable Intelligent Surface and SWIPT
Zheng, G., Fang, Y., Wen, M., & Ding, Z. (2024) · IEEE Internet of Things Journal · doi:10.1109/jiot.2023.3349176 · Cited by 11
“After completing the local computation, all selected device synchronously uploads their local gradient results to the AP for aggregation on the same time–frequency via AirComp [7], [18].”
BackgroundMethodInfluentialSelf-citationZhou, S., Feng, L., Mei, M., & Yao, M. (2024) · IEEE Internet of Things Journal · doi:10.1109/jiot.2024.3414472 · Cited by 15
“Selecting more clients will speed up convergence and improve the model accuracy [11], [12], [13], [14].”
BackgroundMethodDong, 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
“The signal encoding algorithms were designed for the analog WFL system [29], [36], [37].”
MethodEnergy Efficient and Differentially Private Federated Learning via a Piggyback Approach
Chen, R., Huang, C., Qin, X., Ma, N., Pan, M., & Shen, X. (2023) · IEEE Transactions on Mobile Computing · doi:10.1109/tmc.2023.3268323 · Cited by 13
“Energy Efficiency of FL: Recognizing that training large-scale FL models over mobile devices can be energy intensive, several research efforts have been made to reduce these costs through device scheduling [13], network optimization [17] and resource utilization optimization [14], [20], [22].”
BackgroundFederated Learning via Lattice Joint Source-Channel Coding
Azimi‐Abarghouyi, S. M., & Varshney, L. R. (2024) · IEEE International Symposium on Information Theory (ISIT) · doi:10.1109/isit57864.2024.10619502 · Cited by 11
“Independence is also a factor in other over-the-air FL schemes [11], [13].”
BackgroundWu, H., Wang, P., & Narayana, C. V. A. (2025) · IEEE Transactions on Network Science and Engineering · doi:10.1109/tnse.2025.3577910 · Cited by 17
“To reduce the communication cost, methods such as quantization [4], compression/sparsification [5], and over-the-air transmission [6] have been proposed.”
Joint 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
“For some applications where the source data are generated in the analog domain, such as wireless sensor networks [12], [13] and federated learning [3], [4], the quantization processing has to be carried out before utilizing digital AirComp technologies.”
Yi, P., Cao, Y., Kang, X., & Liang, Y. (2024) · IEEE Transactions on Communications · doi:10.1109/tcomm.2024.3468215 · Cited by 15
“More-over, AirComp has been widely used in federated learning to aggregate model parameters, which can preserve privacy and save communication overhead [38].”
MethodJoint Beamforming and Learning Rate Optimization for Over-the-Air Federated Learning
Kim, M., & Park, D. (2023) · IEEE Transactions on Vehicular Technology · doi:10.1109/tvt.2023.3276786 · Cited by 15
“We assume that perfect channel state information (CSI) is availableattheBSandtheoverheadforchannelestimationisnegligible, asin[5],[7].”
BackgroundMethodInfluentialChen, R., Wan, Q., Prakash, P., Zhang, L., Xu, Y., Gong, Y., Fu, X., & Pan, M. (2023) · IEEE/CVF International Conference on Computer Vision (ICCV) · doi:10.1109/iccv51070.2023.01559 · Cited by 10
“Recognizing that training large-scale FL models over edge devices is a time consuming task, several research efforts have been made on decreasing these costs via device scheduling [15], network optimization [35, 25] and resource utilization optimization [30, 28, 29, 24].”
BackgroundEnergy Harvesting Aware Client Selection for Over-the-Air Federated Learning
Chen, C., Chiang, Y.-H., Lin, H. F., Lui, J. C. S., & Ji, Y. (2022) · Globecom · doi:10.1109/globecom48099.2022.10001136 · Cited by 16
“Recently, works [5], [6] based on overthe-air (OTA) computation have been successfully used to reduce the communication cost of FL.”
BackgroundAirCon: 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].”
BackgroundDeep Reinforcement Learning for Over-the-Air Federated Learning in SWIPT-Enabled IoT Networks
Zhang, X., Tian, H., Ni, W., & Sun, M. (2022) · IEEE Vehicular Technology Conference · doi:10.1109/vtc2022-fall57202.2022.10012702 · Cited by 13
“In [5], the authors proposed an over-the-air FL (AirFL) scheme to complete model aggregation in a spectrum-efficient manner.”
Interference Management of Analog Function Computation in Multicluster Networks
Zhang, H., Chen, L., Zhao, N., Chen, Y., & Yu, F. R. (2022) · IEEE Transactions on Communications · doi:10.1109/tcomm.2022.3172996 · Cited by 10
“CoMAC can be applied in this area as a communication-efficient technique, e.g., transceivers were proposed to enable each node to concurrently send its gradient to the server via an MAC, based on which fast global model aggregation was realized at the server [24], [25].”
BackgroundCommunication-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
“To mitigate this problem, in [10], [11], [12], the notion of over-the-air computation [13], [14] is adopted for users so that they upload local gradient vectors simultaneously in distributed SGD.”
BackgroundMethodResultInfluentialOnline Model Updating with Analog Aggregation in Wireless Edge Learning
Wang, J., Dong, M., Liang, B., Boudreau, G., & Abou-Zeid, H. (2022) · IEEE Conference on Computer Communications (INFOCOM) · doi:10.1109/infocom48880.2022.9796860 · Cited by 13
“The design of a common λt among the N mobile devices at each iteration t was studied in [17], [18], [20], [21], [23]-[26], and is outside the scope of this paper.”
BackgroundMethodFederated Learning with Downlink Device Selection
Amiri, M. M., Kulkarni, S. R., & Poor, H. V. (2021) · IEEE International Workshop on Signal Processing Advances in Wireless Communications (SPAWC) · doi:10.1109/spawc51858.2021.9593132 · Cited by 15
“Recently a new line of research, namely federated edge learning (FEEL), has been created considering the physical layer characteristics of the underlying wireless network in the FL framework [4]–[7].”
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
Byzantine Resilient Distributed Multi-Task LearningPreprint
Li, J., Abbas, W., & Koutsoukos, X. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2010.13032 · Cited by 16
“This approach is naturally suited to model distributed learning in multi-agent systems such as mobile phones, autonomous vehicles, and smart cities [2, 3, 4].”
BackgroundOn-Device Federated Learning via Second-Order Optimization with Over-the-Air Computation
Hua, S., Yang, K., & Shi, Y. (2019) · IEEE Vehicular Technology Conference · doi:10.1109/vtcfall.2019.8891310 · Cited by 15
“The difference-of-convex-functions (DC) [11] programming approach has recently been widely used in low-rank optimization [12], [13].”
MethodSelf-citationLiu, J., Li, X., Xu, Y., Lyu, C., Wang, Y., & Liu, X. (2025) · IEEE Internet of Things Journal · doi:10.1109/jiot.2025.3543917 · Cited by 13
“With the increasingly stringent regulations related to data privacy, the adoption of FL can facilitate collaborative learning for the development of effective AI models, without the exchange of potentially sensitive raw data [25], [26], [27].”
Experimental Demonstration of Over the Air Federated Learning for Cellular Networks
Pradhan, S., Koç, A., Alemdar, K., Arfaoui, M., Pietraski, P., Periard, F., Zhang, G.-D., Hudon, M., & Chowdhury, K. R. (2025) · 2025 IEEE International Conference on Machine Learning for Communication and Networking (ICMLCN) · doi:10.1109/icmlcn64995.2025.11140494 · Cited by 11
“In [11], Yang et al. focuses on solving joint device selection and beamforming to perform FL under the assumption of perfect CSI at the clients.”
Robust Resource Allocation for Over-The-Air Computation Networks with Fluid Antenna Array
Pakravan, S., Ahmadzadeh, M., Zeng, M., Hodtani, G. A., Chouinard, J., & Rusch, L. A. (2024) · IEEE Globecom Workshops (GC Wkshps) · doi:10.1109/gcwkshp64532.2024.11100909 · Cited by 13
“Furthermore, recent investigations have explored the integration of machine learning and artificial intelligence techniques to enhance intelligent decision making and data analytics within AirComp systems [5], [6].”
FedLoRa: IoT Spectrum Sensing Through Fast and Energy-Efficient Federated Learning in LoRa Networks
Busacca, F., Mangione, S., Neglia, G., Tinnirello, I., Palazzo, S., & Restuccia, F. (2024) · IEEE International Conference on Mobile Ad-Hoc and Sensor Systems (MASS) · doi:10.1109/mass62177.2024.00047 · Cited by 10
“Existing work on FML has considered either wired networks and/or high-bandwidth wireless networks [9]–[11].”
BackgroundToward Privacy-Preserving Waste Classification in the Internet of Things
Quan, M. K., Nguyen, D. C., Nguyen, V.-D., Wijayasundara, M., Setunge, S., & Pathirana, P. (2024) · IEEE Internet of Things Journal · doi:10.1109/jiot.2024.3386727 · Cited by 11
“On the contrary, federated transfer learning (FTL), as advocated by Yang et al. [10], provided a reliable and secure approach to implement FL by leveraging cross-domain knowledge transfer.”
BackgroundDecentralized Over-the-Air Federated Learning by Second-Order Optimization Method
Yang, P., Jiang, Y., Wen, D., Wang, T., Jones, C. N., & Shi, Y. (2023) · IEEE Transactions on Wireless Communications · doi:10.1109/twc.2023.3327610 · Cited by 14
“In order to further mitigate communication overhead and improve spectral efficiency in this step, we adopt the over-the-air computation (AirComp) technique [48], [49] that is widely used in wireless FL studies [39], [40], [41], [42], [43], [44].”
BackgroundMethodSelf-citationExploring Age-of-Information Weighting in Federated Learning Under Data Heterogeneity
Wang, K., Ding, Z., So, D. K. C., & Ding, Z. J. (2025) · IEEE Transactions on Wireless Communications · doi:10.1109/twc.2025.3545456 · Cited by 13
“Some existing works focused on addressing system heterogeneity by selecting devices based on hardware specifications and communication environment [10], [11], [12], [13].”
Self-citationAge-of-Information-Aware Federated Learning
Xu, Y., Xiao, M., Wu, C., Wu, J., Zhou, J., & Sun, H. (2024) · Journal of Computer Science and Technology · doi:10.1007/s11390-024-3914-x · Cited by 6
Liu, Z., Lan, Q., & Huang, K. (2025) · IEEE Transactions on Wireless Communications · doi:10.1109/twc.2025.3527331 · Cited by 13
“Channel inversion precoding is adopted for magnitude alignment between received signals [9], [42].”
Latency Minimization for Wireless Federated Learning With Heterogeneous Local Model Updates
Zhu, J., Shi, Y., Fu, M., Zhou, Y., Wu, Y., & Fu, L. (2023) · IEEE Internet of Things Journal · doi:10.1109/jiot.2023.3285937 · Cited by 12
“…FL, Zhang and Tao [25] and Sery et al. [26] adopted an analog model aggregation scheme in wireless FL, i.e., over-the-air computation (AirComp) [27], and considered the statistical heterogeneity of the gradients transmitted by different edge devices in different communication rounds and data…”
MethodSelf-citationHong, J., Park, S., & Choi, W. (2023) · IEEE Transactions on Wireless Communications · doi:10.1109/twc.2023.3249252 · Cited by 13
“For this reason, the power control with (3) makes it possible to achieve lower peak-to-average power ratio (PAPR) than the existing power control methods with the average power constraint [16], [17], [22], [23], [24].”
BackgroundMethodInfluentialFedDD: Toward Communication-Efficient Federated Learning With Differential Parameter Dropout
Feng, Z., Chen, X., Wu, Q., Wu, W. R., Zhang, X., & Huang, Q. (2023) · IEEE Transactions on Mobile Computing · doi:10.1109/tmc.2023.3311188 · Cited by 20
“The existing works [37], [38] show the convergence of training can be accelerated with more involved clients.”
Joint Downlink-Uplink Beamforming for Wireless Multi-Antenna Federated Learning
Zhang, C., Dong, M., Liang, B., Afana, A., & Ahmed, Y. (2023) · International Symposium on Modeling and Optimization in Mobile, Ad-Hoc and Wireless Networks · doi:10.23919/wiopt58741.2023.10349883 · Cited by 11
“While beamforming was considered in [9] for FL downlink analog transmission and in [4] for uplink over-the-air aggregation, there is no existing work on joint downlink-uplink beamforming design.”
BackgroundMethodWang, J., Mao, Y., Wang, T., & Shi, Y. (2023) · IEEE Transactions on Wireless Communications · doi:10.1109/twc.2023.3317129 · Cited by 13
“Different from many existing works on FL that adopts the conventional 32-bit floating-point format [39], [40], [41], in this work, we adopt a QNN structure to save the computation energy.”
MethodSelf-citationCost-Aware Hierarchical Federated Learning via Over-the-Air Computing
Xue, D., Luo, J., Jiang, C., & Gao, L. X. (2022) · Globecom · doi:10.1109/globecom48099.2022.10000988 · Cited by 12
“, [10]- [14]), we consider a quasi-static FL 2022 IEEE Global Communications Conference: Signal Processing for Communications”
BackgroundAl-Abiad, M. S., Obeed, M., Hossain, M. J., & Chaaban, A. (2022) · arXiv (Cornell University) · doi:10.48550/arxiv.2206.02981 · Cited by 6
“vector machine with li(w) = max{0, 1 − yixi w}, yi ∈ {−1, 1} [11], [37].”
Incentive and Knowledge Distillation Based Federated Learning for Cross-Silo Applications
Li, B., Shi, Y., Guo, Y., Kong, Q., & Jiang, Y. (2022) · IEEE Conference on Computer Communications Workshops (INFOCOM WKSHPS) · doi:10.1109/infocomwkshps54753.2022.9798320 · Cited by 14
“On the one hand, federated learning aims to train a global model for all clients and new participants [4].”
BackgroundSelf-citationAdaptive Participant Selection in Heterogeneous Federated Learning
Albelaihi, R., Sun, X., Craft, W. D., Yu, L., & Wang, C. (2021) · Globecom · doi:10.1109/globecom46510.2021.9685077 · Cited by 14
“Similarly, Yang et al. [18] proposed a joint participant selection and beamforming approach to maximize the number of selected participants in each global iteration while satisfying the mean-squared-error requirement.”
BackgroundFundamental limits of over-the-air optimization: Are analog schemes optimal?
Jha, S. K., Mayekar, P., & Tyagi, H. (2021) · Globecom · doi:10.1109/globecom46510.2021.9685770 · Cited by 3
“Similar scaling approaches are also presented in [19], [21]–[23].”
BackgroundResultToward Optimal Rate-Delay Tradeoff for Computation Over Multiple Access Channel
Chen, L., Zhao, N., Chen, Y., Yu, F. R., & Wei, G. (2021) · IEEE Transactions on Communications · doi:10.1109/tcomm.2021.3068493 · Cited by 11
“Recently, CoMAC scheme has been applied into the communication-efficient federated learning over wireless networks [13]–[15].”
MethodIntelligent Reflecting Surfaces Enhanced Federated Learning
Ni, W., Liu, Y., & Tian, H. (2020) · IEEE Globecom Workshops (GC Wkshps) · doi:10.1109/gcwkshps50303.2020.9367578 · Cited by 9
“Proposition 1: For the purpose of minimizing the MSE in (5a), i.e., (cid:5) Kk =1 | a ¯ h k p k / √ η − 1 | 2 = 0 , the optimal transmit scalar at the k -th device can be designed as Proof: Please see Appendix A in [3].”
BackgroundMethodInfluentialA Sequential Gradient-Based Multiple Access for Distributed Learning over Fading Channels
Sery, T., & Cohen, K. (2019) · Allerton Conference on Communication, Control, and Computing · doi:10.1109/allerton.2019.8919883 · Cited by 12
“In [43], multiple antennas were used at the receiver, where beamforming was used to maximize the number of devices scheduled for transmission.”
MethodRahmati, M., & Rahmati, N. (2026) · Journal of Electrical Systems and Information Technology · doi:10.1186/s43067-025-00285-y · Cited by 15
Over-the-Air Hierarchical Personalized Federated Learning
Zhou, F., Wang, Z., Shan, H., Wu, L., Tian, X., Shi, Y., & Zhou, Y. (2024) · IEEE Transactions on Vehicular Technology · doi:10.1109/tvt.2024.3499349 · Cited by 12
“The effectiveness of AirComp-assisted FL was verified in [18], where the authors maximize the number of clients participating in model training under the model aggregation error constraint.”
Self-citationOver-the-Air Fair Federated Learning via Multi-Objective Optimization
Hamidi, S. M., Bereyhi, A., Asaad, S., & Poor, H. V. (2025) · IEEE Communications Letters · doi:10.1109/lcomm.2025.3567387 · Cited by 12
“The impact of OTA computation can be handled using standard OTA-FL techniques [4] that model channel noise and signal superposition.”
Zhang, S., Yang, Z., Chen, G., Dong, Z., & Han, Z. (2024) · IEEE Transactions on Wireless Communications · doi:10.1109/twc.2024.3453398 · Cited by 11
“Note that the aggregation error ε ( t ) mainly relies on the influence of the channel condition and Gaussian noise which will significantly affect the global aggregation and convergence [33], [34], [35].”
BackgroundCompute-Update Federated Learning: A Lattice Coding Approach
Azimi‐Abarghouyi, S. M., & Varshney, L. R. (2024) · IEEE Transactions on Signal Processing · doi:10.1109/tsp.2024.3491993 · Cited by 11
“This conventional approach, recommended for ideal, error-free communication conditions as outlined in the seminal FL paper [2], has been consistently used in over-the-air FL research [8], [9], [10], [11], [12], [13], [14], [15], [16], [17], [18], [21], [22], [23], [24], [25], [26], [27].”
Influential
Metadata
- Authors
- Yang, K., Jiang, T., Shi, Y., & Ding, Z.
- Affiliations
- Chinese Academy of Sciences (CN)
- ShanghaiTech University (CN)
- University of California, Davis (US)
- Venue
- IEEE Transactions on Wireless Communications
- Published
- 2020-03
- Volume
- 19
- Issue
- 3
- Pages
- 2022-2035
- ISSN
- 1536-1276 (ISSN-L)
- 1558-2248
- OpenAlex
- W2999074226
- Cited by
- 1,084
- References
- 62
- Citations by year
- 2026: 86 · 2025: 154 · 2024: 176 · 2023: 190 · 2022: 189 · 2021: 190 · 2020: 80
- FWCI
- 63.79
- Type
- journal article
- Language
- en
- Open access
- green
- arxiv.org
- Licence
- Locations
- IEEE Transactions on Wireless Communications (publishedVersion)
- arXiv (Cornell University) (submittedVersion)
- arXiv (Cornell University)
- Funders
- National Natural Science Foundation of China (61601290)
- National Science Foundation (CNS-1702752, ECCS-1711823)
- 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 › Engineering › Electrical and Electronic Engineering › Advanced Wireless Communication Technologies
- Keywords
- federated learning, over-the-air computation, edge machine learning, model aggregation, device selection, beamforming design, difference of convex functions, DC algorithm, non-IID data, data privacy
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.