Transfer Learning and Meta Learning-Based Fast Downlink Beamforming Adaptation
Open accessThe DOI matches its Crossref record; 47 of the 47 references it cites were found in a registry or an index.
Abstract
This article studies fast adaptive beamforming optimization for the signal-to-interference-plus-noise ratio balancing problem in a multiuser multiple-input single-output downlink system. Existing deep learning based approaches to predict beamforming rely on the assumption that the training and testing channels follow the same distribution which may not hold in practice. As a result, a trained model may lead to performance deterioration when the testing network environment changes. To deal with this task mismatch issue, we propose two offline adaptive algorithms based on deep transfer learning and meta-learning, which are able to achieve fast adaptation with the limited new labelled data when the testing wireless environment changes. Furthermore, we propose an online algorithm to enhance the adaptation capability of the offline meta algorithm in realistic non-stationary environments. Simulation results demonstrate that the proposed adaptive algorithms achieve much better performance than the direct deep learning algorithm without adaptation in new environments. The meta-learning algorithm outperforms the deep transfer learning algorithm and achieves near optimal performance. In addition, compared to the offline meta-learning algorithm, the proposed online meta-learning algorithm shows superior adaption performance in changing environments.
Overview
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Integrity signals
Registry
Cites
Statistics
Figures
![Fig. 1 The DL-based BNN for uplink power prediction and beamforming recovery [25].](https://figures.cochat.ai/paper-figures/89769fd7a0c64aeb96f4b47aaf7e3c91/transfer-learning-and-meta-learning-based-fast-downlink-beamforming-adaptation-figure-1.png)





![Fig. 7 Manhattan road grid 750 m × 1299 m [41].](https://figures.cochat.ai/paper-figures/38ed1523d101487b9ce336f6a155e5b9/transfer-learning-and-meta-learning-based-fast-downlink-beamforming-adaptation-figure-7.png)









Verification
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
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
“Transfer learning [103] and continual learning [277] have recently been adopted to address such task mismatch issue in the”
From Edge artificial intelligence for 6g: vision, enabling technologies, and applications (2022) · cites this paper as [103]
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
Citations
Cites
Adam: A Method for Stochastic Optimization
Kingma, D. P., & Ba, J. (2014) · arXiv · doi:10.48550/arxiv.1412.6980 · Cited by 98,946 · cited 2×
“learning rate, ∇θLossD Tr (θ) is the gradient of the loss function over θ. Alternatively, the network parameter θ also can be updated by using the existing adaptive moment estimation (ADAM) algorithm [39]. Next, we move to the fine-tuning stage when the pre-training stage is finished. Fine-tuning aims to refine all or pa…”
MethodInfluentialPan, S. J., & Yang, Q. (2010) · IEEE Transactions on Knowledge and Data Engineering · doi:10.1109/tkde.2009.191 · Cited by 23,387 · cited 2×
“and corresponding applications have been introduced in [27].”
BackgroundFundamentals of Wireless Communication
Tse, D., & Viswanath, P. (2005) · doi:10.1017/cbo9780511807213 · Cited by 13,162 · cited 1×
“arge-scale case. Besides, the antenna gains of the BS and each vehicle user are set to be 8 dBi and 3 dBi, respectively. The decorrelation distance is set to 50 m. We use Clarke’s model introduced in [42] to generate the small-scale fading of the moving vehicles. The effects of the transmit power and the number of veh…”
MethodModel-Agnostic Meta-Learning for Fast Adaptation of Deep Networks
Finn, C., Abbeel, P., & Levine, S. (2017) · arXiv · doi:10.48550/arxiv.1703.03400 · Cited by 8,512 · cited 4×
“In order to eliminate the model architecture restriction on the applications of meta-learning, the authors in [31] proposed the model-agnostic meta-learning (MAML) algorithm.”
BackgroundMethodInfluentialHow transferable are features in deep neural networks?
Yosinski, J., Clune, J., Bengio, Y., & Lipson, H. (2014) · arXiv · doi:10.48550/arxiv.1411.1792 · Cited by 6,061
Five disruptive technology directions for 5G
Boccardi, F., Heath, R. W., Lozano, A., Marzetta, T. L., & Popovski, P. (2014) · IEEE Communications Magazine · doi:10.1109/mcom.2014.6736746 · Cited by 3,709
Shi, Q., Razaviyayn, M., Luo, Z.-Q., & He, C. (2011) · IEEE Transactions on Signal Processing · doi:10.1109/tsp.2011.2147784 · Cited by 2,129 · cited 1×
“In the last two decades, beamforming optimization has been well studied for some specifical problems, such as signal-to-interferenceplus-noise ratio (SINR) balancing problem [3], [4], power minimization problem [5], [6] and sum rate maximization problem [4], [7]–[9].”
BackgroundSelf-citationPower of Deep Learning for Channel Estimation and Signal Detection in OFDM Systems
Ye, H., Li, G. Y., & Juang, B.-H. (2018) · IEEE Wireless Communications Letters · doi:10.1109/lwc.2017.2757490 · Cited by 2,000
Deep Learning in Mobile and Wireless Networking: A Survey
Zhang, C., Patras, P., & Haddadi, H. (2019) · IEEE Communications Surveys & Tutorials · doi:10.1109/comst.2019.2904897 · Cited by 1,768
Online Learning and Online Convex Optimization
Shalev-Shwartz, S. (2012) · Foundations and Trends® in Machine Learning · doi:10.1561/2200000018 · Cited by 1,246
Research works printed but not linked: 5
- J. Hannan, "Approximation to bayes risk in repeated play," Contributions to the Theory of Games, vol. 3, pp. 97-139, 1957.
- P. Kyosti, "IST-4-027756 WINNER II D1.1.2 v.1.2: WINNER II channel models," Inf. Soc. Technol., 2007.
- 3GPP: Technical Specification Group Radio Access Network: Study LTE-Based V2X Services: (Release 14), Standard 3GPP TR 36.885 V2.0.0, Jun. 2016.
- Gan Zheng (S'05-M'09-SM'12) received the BEng and the MEng from Tianjin University, Tianjin, China, in 2002 and 2004, respectively, both in Electronic and Information Engineering, and the PhD degree in Electrical and Electronic Engineering from The University of Hong Kong in 2008. He is currently Reader of Signal Processing for Wireless Communications in the Wolfson School of Me- chanical, Electrical and Manufacturing Engineering, Loughborough University, UK. His research inter- ests include machine learning for communications, UAV communications, mobile edge caching, full-duplex radio and wireless power transfer. He is the first recipient for the 2013 IEEE Signal Processing Letters Best Paper Award, and he also received 2015 GLOBECOM Best Paper Award, and 2018 IEEE Technical Committee on Green Communications & Computing Best Paper Award. He was listed as a Highly Cited Researcher by Thomson Reuters/Clarivate Analytics in 2019. He currently serves as an Associate Editor for IEEE Communications Letters and IEEE Wireless Communications Letters.
- Kai-Kit Wong (M'01-SM'08-F'16) received the BEng, the MPhil, and the PhD degrees, all in Electri- cal and Electronic Engineering, from the Hong Kong University of Science and Technology, Hong Kong, in 1996, 1998, and 2001, respectively. After gradua- tion, he took up academic and research positions at the University of Hong Kong, Lucent Technologies, Bell-Labs, Holmdel, the Smart Antennas Research Group of Stanford University, and the University of Hull, UK. He is Chair in Wireless Communications at the Department of Electronic and Electrical Engi- neering, University College London, UK. His current research centers around 5G and beyond mobile communications. He is a co-recipient of the 2013 IEEE Signal Processing Letters Best Paper Award and the 2000 IEEE VTS Japan Chapter Award at the IEEE Vehicular Technology Conference in Japan in 2000, and a few other international best paper awards. He is Fellow of IEEE and IET and is also on the editorial board of several international journals.
Cited by
Items 101–136 of 136
Attention-transfer-based path loss prediction in asymmetric massive MIMO IoT systems
Zhang, Y., Chen, M., Yuan, M., Zhang, W., & Lago, L. A. (2024) · Computer Communications · doi:10.1016/j.comcom.2024.07.006 · Cited by 2
Lightweight and Generalizable AoA Estimation for IoT: A Novel Few-Shot Learning Approach
Mashaal, O., Mohammed, E., Digby, A., Leone, P., Swersky, L., Eshaghbeigi, A., & Abou-Zeid, H. (2025) · International Conference on Communications · doi:10.1109/icc52391.2025.11160710 · Cited by 1
“MAML-based methods have shown improved adaptation over transfer learning, particularly for power allocation prediction [8] and classification of wave angles in dynamic air-to-air networks using received signal strength (RSS) [9].”
Gradient-Driven Graph Neural Networks for Learning Digital and Hybrid Precoder
Li, Z., Han, S., & Yang, C. (2025) · IEEE Transactions on Communications · doi:10.1109/tcomm.2025.3628767 · Cited by 4
Machine Learning-Based Adaptive Codebook Design and Beamforming for Near-Field Communications
Zhang, M., Cai, Y., Yu, G., Xu, J., & Swindlehurst, A. L. (2025) · IEEE Transactions on Communications · doi:10.1109/tcomm.2025.3637068 · Cited by 1
Putri, D. A., Zakia, I., Hendrawan, H., & Annur, R. (2025) · IEEE Open Journal of the Communications Society · doi:10.1109/ojcoms.2025.3642009 · Cited by 1
“TL attempts to reuse a policy learned in one environment for a new related environment [14], while meta-learning seeks to learn initialization parameters that generalize well across a variety of tasks [22].”
ATLAS: AI-Native Receiver Test-and-Measurement by Leveraging AI-Guided Search
Belgiovine, M., Pradhan, S., Lange, J., Loehning, M., & Chowdhury, K. R. (2025) · IEEE International Symposium on Personal, Indoor and Mobile Radio Communications (PIMRC) · doi:10.1109/pimrc62392.2025.11274653 · Cited by 1
“[17] improves adaptability using transfer and meta-learning for beamforming but do not identify failure scenarios.”
Memristor-Based Lightweight Meta Learning for Beam Prediction in Non-Stationary Environments
Cao, Y., Ohtsuki, T. O., Maghsudi, S., & Quek, T. Q. S. (2026) · IEEE Transactions on Cognitive Communications and Networking · doi:10.1109/tccn.2026.3706535 · Cited by 2
“3 More details are stated as follows.”
Zengeler, N., Glasmachers, T., & Handmann, U. (2022) · Proceedings - International Conference on Pattern Recognition/Proceedings/International Conference on Pattern Recognition · doi:10.1109/icpr56361.2022.9956622 · Cited by 1
“In an application concerning adaptive beamforming optimization for a signal processing problem, the combination of Transfer and Meta Learning helps to solve the problem of performance deterioration when the testing environment changes [41].”
MethodLearning Adaptive Beamforming Policy for Different Optimization Problems
Ma, Y., Yang, C., Han, S., & Zhao, B. (2024) · IEEE Wireless Communications and Networking Conference · doi:10.1109/wcnc57260.2024.10570569 · Cited by 1
“TL and MAML have been employed for optimizing beam-forming [12] and power allocation [13].”
BackgroundMethodInfluentialFGNN-Based Improved Resource Distribution Framework for V2X Wireless Networks
Waqas, S. M., Tang, Y., Abbas, F., Hussain, M., & Bangash, Y. A. (2024) · IEEE Vehicular Technology Conference · doi:10.1109/vtc2024-spring62846.2024.10683058 · Cited by 1
“Noteworthy efforts have been made, employing methods such as deep learning, GAN, and deep graph neural network process [8], [9], [10], [11], [12], [13], [9], [14] to tackle challenges related to power allocation, sum rate, and latency.”
MethodChannel Characteristics-Based Domain Adaptation for Multiuser MISO Beamforming Design
Ge, X., & Dai, X. (2024) · International Conference on Wireless Communications and Signal Processing (WCSP) · doi:10.1109/wcsp62071.2024.10827055 · Cited by 1
“It fine tunes the last layer of the neural network described in [3] by using some labelled data from target domain.”
Distributed Learning in Non-Convex World: Theory, Algorithms, and Applications
Sun, H. (2021) · University of Minnesota Digital Conservancy (University of Minnesota)
2021 Index IEEE Transactions on Wireless Communications Vol. 20
(2021) · IEEE Transactions on Wireless Communications · doi:10.1109/twc.2021.3135649
An Improved Data-Driven Decision Feedback Receiver via Deep Unfolding
Liu, S., Zhao, L., & Song, T. (2022) · Mathematical Problems in Engineering · doi:10.1155/2022/5656880
Fast Adaptation of Radar Detection via Online Meta-learning
Khan, Z., Jiang, W., Haimovich, A. M., Govoni, M. A., Garner, T. J., & Simeone, O. (2022) · Asilomar Conference on Signals, Systems and Computers (ACSSC) · doi:10.1109/ieeeconf56349.2022.10051849
“Meta-learning has been applied to solve problems in communications, such as resource allocation in wireless networks [16], downlink beamforming optimization [17], demodulation over fading channels [18], and decoding for convolutional and turbo codes [19].”
MethodResearch On Adaptation and Generalization Issues in Image Processing
Liu, Y. (2024) · Highlights in Science Engineering and Technology · doi:10.54097/x85q2942
“Approaches [10-11] to domain adaptation that are based on features function by acquiring knowledge from both the target and source domains exhibit comparable representations.”
BackgroundGe, X., & Dai, X. (2025) · IEEE Transactions on Green Communications and Networking · doi:10.1109/tgcn.2025.3546932
“As a result, when the communication environment changes, DL-based beamforming solutions will face channel mismatch, which means that the training and testing CSI no longer follow the same distribution [7], [8], [9].”
InfluentialYe, N., Cao, X., Pan, J., Liu, W., & Hou, X. (2025) · IEEE Signal Processing Letters · doi:10.1109/lsp.2025.3576666
A Spectrum-Efficient Solution with Data Rate Guarantees in 5G/Next-G Networks
Acharya, S., Li, S., Wu, Y., Jiang, N., Lou, W., & Hou, Y. T. (2025) · IEEE Internet of Things Journal · doi:10.1109/jiot.2025.3595817
“Minimizing spectrum usage is typically not an objective in these works either [7], [8], [9], [10], [14], [15], [16], [17], [18], [19], [20], [21], [23], [24], [25], [26].”
InfluentialE2E Learning Massive MIMO for Multimodal Semantic Non-Orthogonal Transmission and Fusion
Wu, M., & Gao, Z. (2025) · IEEE Journal on Selected Areas in Communications · doi:10.1109/jsac.2025.3643817
“In precoding design, the authors of [38], [39], [40] proposed the use of supervised CNNs to approximate traditional iterative beamforming, which has been further extended to hybrid beamforming in massive MIMO systems in [41].”
Dictionary Learning for Phase-Less Beam Alignment Codebook Design in Multipath Channels
Domae, B. W., & Čabrić, D. (2026) · IEEE Transactions on Machine Learning in Communications and Networking · doi:10.1109/tmlcn.2026.3653010 · Cited by 1
“While we also use TL to accelerate codebook design implicitly through faster BA instances, [25] and [26] use TL to directly design BS downlink beams.”
Edge Sensing for Efficient UAV Beam Selection: A Semantic-Driven Distributed Learning Framework
Zhou, Q., Yang, X., Zhao, J., Cai, K., & Zhu, Y. (2025) · International Conference on Ubiquitous Communication (Ucom) · doi:10.1109/ucom67224.2025.11337204
“The high mobility of UAVs necessitates robust, low-latency millimeter-wave (mmWave) links, critically dependent on efficient beamforming [5], [6].”
Robust Machine Learning Approaches to Wireless Communication Networks
Zecchin, M. (2022) · theses.fr (ABES) · doi:10.70675/9a17dfe5z7c45z4353z8d5dzfac2445840ad
Secure Near-Field Location Division Multiple Access via Quantum-Classical Learning Workflow
Nguyen, Q. M., Narottama, B., Nguyen, M.-H. T., Sharma, V., Le, Q. N., & Duong, T. Q. (2026) · IEEE Internet of Things Journal · doi:10.1109/jiot.2026.3690067
Machine Learning (ML) in NOMA: Toward NGMA Networks
Matin, M. A., & Mahmood, M. R. (2026) · Synthesis lectures on communications · doi:10.1007/978-3-032-19420-6_2
Mudda, S., Loganathan, P., Abirami, M., Kumar, R. D., L.Sowmya, & Saveeithaa, S. V. (2026) · International Conference on Intelligent Communication Technologies and Virtual Mobile Networks (ICICV) · doi:10.1109/icicv68925.2026.11554624
“Meta-learning (or learning to learn) has been advocated as a promising approach for fast adaptation in changing environments[5].”
InfluentialMeng, F. Y., Liu, S., Wang, Z., Zhang, C. T., & Huang, Y. (2026) · Chinese Journal of Electronics · doi:10.23919/cje.2025.00.097
Liu, C., Long, H., Yuan, Y., & Ma, L. (2026) · International Conference on Signal Processing, Communication and Control Systems (SPCCS) · doi:10.1109/spccs69331.2026.11588569
“The primary evaluation metrics are the Output Signal-to-Interference-plus-Noise Ratio (SINR) for assessing beamforming efficacy and the Root Mean Square Error (RMSE) for quantifying DOA estimation accuracy [14,15].”
Self-citationTransfer Learning-Based Beamforming for MIMO Antenna Arrays Using Pre-Trained Neural Network Model
Wattamwar, A. M., Arora, S., & Kumar, V. N. (2026) · International Conference on Next Generation Electronics (NELEX) · doi:10.1109/nelex69209.2026.11590431
“Transfer learning that takes advantage of pre-trained models on big datasets from similar domains has become one of the most effective ways of dealing with the lack of data [19] [20] [21].”
Joint Compression and Resource Allocation for Semantic Communication Based Image Transmission
Li, Z., Jiang, W., Wang, Q., Qian, L., Wei, F., & Sun, Y. (2026) · International Conference on Communications · doi:10.1109/icc59461.2026.11588278
“[3], [13].”
Multistatic ISAC for Improved Sensing Under Jamming Attacks: A Gradient-Based Meta-Learning Approach
Bokhary, A. E., Amhaz, A., Sharafeddine, S., & Assi, C. M. (2026) · IEEE Wireless Communications Letters · doi:10.1109/lwc.2026.3716181
A Decomposed Learning Framework for Hybrid Precoding
Zhang, N., & Yang, C. (2026) · IEEE Open Journal of the Communications Society · doi:10.1109/ojcoms.2026.3719783
“Another model-driven method exploits the explicit structure of optimal solutions, such as the well-known structure of optimal fully-digital precoders in certain multi-user multi-input-single-output (MU-MISO) problems [15], [16], [17], [18], [19], [20].”
RSSI Forecasting With Off-the-Shelf Smartphones For Proactive Mobile Wi-Fi Link Adaptation
Mahalal, E. (2026) · International Conference on Intelligent Multimedia, Networking, and Security (IMNS) · doi:10.1109/imns67862.2026.11655266
“Early studies applied autoregressive and Kalman filters on complex channel coefficients [4], and more recent papers train LSTM and Trans-former models on CSI sequences to predict amplitude and phase over horizons of a few milliseconds [5], [6].”
Robust Unsupervised Learning for Wireless Resource Allocation Under Channel Distribution Shift
Liu, S., & Yang, C. (2026) · IEEE Transactions on Machine Learning in Communications and Networking · doi:10.1109/tmlcn.2026.3736371
“Although DNNs can be fine-tuned online to adapt to new environments using transfer learning [12], meta-learning [13], etc. , collecting and training with new datasets are time-consuming and expensive.”
MTSBP: Meta-Trajectory-Shifting Learning for Generalizable mmWave Beam Prediction
Mu, Y., Cui, Q., Wei, N., Li, Q., & Tao, X. (2026) · IEEE Transactions on Cognitive Communications and Networking · doi:10.1109/tccn.2026.3745642
Consistency-Based Test Time Adaptation for Deep Learning-Based Parametric Channel Estimation
Shin, W., Kim, H., & Shim, B. (2026) · IEEE VTS Asia Pacific Wireless Communications Symposium (APWCS) · doi:10.1109/apwcs71511.2026.11742821
“In [6], a meta-learning approach has been proposed to enable fast beamforming adaptation to new wireless environments from only a limited amount of labeled target data.”
Metadata
- Authors
- Yuan, Y., Zheng, G., Wong, K.-K., Ottersten, B., & Luo, Z.-Q.
- Affiliations
- Loughborough University (GB)
- University College London (GB)
- Shenzhen Research Institute of Big Data (CN)
- Venue
- IEEE Transactions on Wireless Communications
- Published
- 2021-03
- Volume
- 20
- Issue
- 3
- Pages
- 1742-1755
- ISSN
- 1536-1276 (ISSN-L)
- 1558-2248
- OpenAlex
- W3100120649
- Cited by
- 133
- References
- 45
- Citations by year
- 2026: 18 · 2025: 26 · 2024: 32 · 2023: 24 · 2022: 16 · 2021: 17
- FWCI
- 5.60
- Type
- journal article
- Language
- en
- Open access
- green
- arxiv.org
- Locations
- IEEE Transactions on Wireless Communications (publishedVersion)
- arXiv (Cornell University) (submittedVersion)
- Open Repository and Bibliography (University of Luxembourg) (submittedVersion, cc-by-nc-sa)
- UCL Discovery (University College London) (submittedVersion)
- Figshare (submittedVersion, cc-by)
- arXiv (Cornell University)
- Funders
- U.K. Engineering and Physical Sciences Research Council (EP/N007840/1)
- Leverhulme Trust Research Project (RPG-2017-129)
- EPSRC (EP/N008219/1)
- H2020-ERC-AdG-AGNOSTIC (742648)
- Leading Talents of Guangdong Province Program (00201501)
- National Natural Science Foundation of China (61731018)
- Development and Reform Commission of Shenzhen Municipality
- Shenzhen Fundamental Research and Discipline Layout project (KQTD201503311441545)
- Topics
- Physical Sciences › Computer Science › Artificial Intelligence › Domain Adaptation and Few-Shot Learning
- Physical Sciences › Engineering › Electrical and Electronic Engineering › Advanced MIMO Systems Optimization
- Physical Sciences › Engineering › Electrical and Electronic Engineering › Millimeter-Wave Propagation and Modeling
- Keywords
- downlink beamforming, transfer learning, meta-learning, online meta-learning, SINR balancing, multi-user MIMO, distribution mismatch, rapid adaptation, non-stationary environments, limited labeled data
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.