Optimal Wireless Resource Allocation With Random Edge Graph Neural Networks
Open accessThe DOI matches its Crossref record; 40 of the 41 references it cites were found in a registry or an index.
Abstract
We consider the problem of optimally allocating resources across a set of transmitters and receivers in a wireless network. The resulting optimization problem takes the form of constrained statistical learning, in which solutions can be found in a model-free manner by parameterizing the resource allocation policy. Convolutional neural networks architectures are an attractive option for parameterization, as their dimensionality is small and does not scale with network size. We introduce the random edge graph neural network (REGNN), which performs convolutions over random graphs formed by the fading interference patterns in the wireless network. The REGNN-based allocation policies are shown to retain an important permutation equivariance property that makes them amenable to transference to different networks. We further present an unsupervised model-free primal-dual learning algorithm to train the weights of the REGNN. Through numerical simulations, we demonstrate the strong performance REGNNs obtain relative to heuristic benchmarks and their transference capabilities.
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
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
“Graph neural network (GNN) has recently been shown to be able to harness the benefits of generalizability, interpretability, robustness, scalability, superior performance, real-time and distributed implementation for learning to optimize nonconvex problems, including power control [270], beamforming [23], and phase shift design [25].”
From Edge artificial intelligence for 6g: vision, enabling technologies, and applications (2022) · cites this paper as [270]
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
ImageNet classification with deep convolutional neural networks
Krizhevsky, A., Sutskever, I., & Hinton, G. E. (2017) · Communications of the ACM · doi:10.1145/3065386 · Cited by 112,155
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×
“In fast fading scenarios the instantaneous value of the reward r(t) in (1) is not directly experienced by end users.”
BackgroundBertsekas, D. P. (1997) · Journal of the Operational Research Society · doi:10.1038/sj.jors.2600425 · Cited by 11,726
(1999) · Cited by 5,147
Wide & Deep Learning for Recommender Systems
Cheng, H.-T., Koc, L., Harmsen, J., Shaked, T., Chandra, T., Aradhye, H., Anderson, G., Corrado, G., Chai, W., Ispir, M., et al. (2016) · Proceedings of the 1st Workshop on Deep Learning for Recommender Systems · doi:10.1145/2988450.2988454 · Cited by 3,535
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
Dual methods for nonconvex spectrum optimization of multicarrier systems
Yu, W., & Lui, R. (2006) · IEEE Transactions on Communications · doi:10.1109/tcomm.2006.877962 · Cited by 1,572 · cited 1×
“Some simplification is attained by working in the Lagrangian dual domain [1], [2] and subsequently using dual descent methods [3]–[5], or, alternatively, with heuristic optimization and scheduling methods [6]–[9].”
MethodDeep Convolutional Networks on Graph-Structured Data
Henaff, M., Bruna, J., & LeCun, Y. (2015) · arXiv · doi:10.48550/arxiv.1506.05163 · Cited by 1,255
Learning to Optimize: Training Deep Neural Networks for Interference Management
Sun, H., Chen, X., Shi, Q., Hong, M., Fu, X., & Sidiropoulos, N. D. (2018) · IEEE Transactions on Signal Processing · doi:10.1109/tsp.2018.2866382 · Cited by 874
Shi, Q., Razaviyayn, M., Luo, Z.-Q., & He, C. (2011) · 2011 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP) · doi:10.1109/icassp.2011.5946304 · Cited by 873
Cited by
Items 301–344 of 344
GNN for Optimizing RIS-Assisted Federated Edge Learning
Zhou, Y., Fang, W., Shi, Y., & Letaief, K. B. (2025) · Wireless networks · doi:10.1007/978-3-031-96649-1_5
Decentralized GNN-based Power Allocation with Varying Network Density
Amorosa, L. M., Gao, Z., Chahoud, T., Verdone, R., & Gündüz, D. (2025) · IEEE International Conference on Machine Learning for Communication and Networking (ICMLCN) · doi:10.1109/icmlcn64995.2025.11139371 · Cited by 1
“For instance, in [5], GNNs are employed to capture the dynamic structure of fading channel states, facilitating the learning of optimal resource allocation policies in wireless networks.”
InfluentialGraph Attention Network for Optimal User Association in Wireless Networks
Mirzaei, J., Mitra, J., & Poitau, G. (2025) · IEEE International Conference on Machine Learning for Communication and Networking (ICMLCN) · doi:10.1109/icmlcn64995.2025.11140312 · Cited by 1
Hybrid Interference Graphs Learning for Spatio-Temporal Channel Allocation in WLANs
Zhu, S., Jiang, J., & Liu, L. (2025) · International Conference on Data-driven Optimization of Complex Systems (DOCS) · doi:10.1109/docs67533.2025.11200692
“Eisen et al. [18] introduced the random edge graph neural network (REGNN), which performs convolutions over random graphs formed by fading interference patterns.”
Redrugina, N. M., & Tarabanov, I. (2025) · Telecom IT · doi:10.31854/2307-1303-2025-13-2-43-51
Optimization of Handover through Knowledge Graph Enabled Coordination between LLM and O-RAN
Rao, A., Gowtham, V., Shivaprasad, S., Schreiner, F., Corici, M., & Magedanz, T. (2025) · IEEE Future Networks World Forum (FNWF) · doi:10.1109/fnwf66845.2025.11317210
Gautam, V., & Gupta, R. (2025) · International Conference on Communication and Smart Devices (ICCoSD) · doi:10.1109/iccosd66074.2025.11348444
Edge-Aware Graphormer for Downlink Covert Communication Optimization in Aerial Cell-Free Networks
Chen, Y., Dai, Y., & Lyu, L. (2025) · International Conference on Wireless Communications and Signal Processing (WCSP) · doi:10.1109/wcsp68525.2025.1010539
GNN-ATIVE: An AI-native, Graph-based Orchestrator for Next-Generation Wireless Networks
Gowtham, V., Başaran, O. T., Dandekar, A., Kukkalli, H., Schreiner, F., Corici, M., Schulz-Zander, J., Dressler, F., Bauschert, T., Stańczak, S., et al. (2025) · Globecom · doi:10.1109/globecom59602.2025.11431808 · Cited by 2
Precoder Learning for Weighted Sum Rate Maximization
Deng, M., & Han, S. (2025) · Globecom · doi:10.1109/globecom59602.2025.11432429
“In particular, the effectiveness of graph neural networks (GNNs) for precoder learning has gained considerable attention [5–7].”
Ajayi, O. T., Wang, S., & Cheng, Y. (2026) · IEEE Transactions on Mobile Computing · doi:10.1109/tmc.2026.3681945
“Some works like [29], [54], [55] have proposed GNN techniques for link scheduling in wireless networks.”
Qian, S., & Cheng, M. (2026) · International Symposium on Computer Applications and Information Systems (ISCAIS) · doi:10.1109/iscait69154.2026.11477366
Graph Neural PPO for Joint User Association and Resource Allocation in Open RAN
Do, K.-H., Nguyen, T.-H., Luu, Q.-T., Nguyen, M.-T., Tran, D.-M., & Nguyen, V.-D. (2026) · International Conference on Information Networking · doi:10.1109/icoin68469.2026.11480477
“…policy gradient (DDPG), and proximal policy optimization (PPO) have shown strong potential for resource orchestration in dynamic wireless systems [16]– [18], while graph neural networks (GNNs) effectively capture spatial correlations and interference relationships in large-scale networks [19]–[21].”
Opportunistic routing in communications networks via learnable state-augmented policies
Das, S., Das, S., & Panda, K. G. (2026) · Frontiers in Communications and Networks · doi:10.3389/frcmn.2026.1779380
Marwani, M. (2026) · Espace ÉTS (ETS)
GNN–DQN based Joint Optimization of Pilot Allocation and Power Control in MIMO
Vijayarani, R., Thejeshwaane, J. S., Kalaiselvan, S., & Maamathi, K. S. (2026) · International Conference on Pervasive Computing and Social Networking (ICPCSN) · doi:10.1109/icpcsn68523.2026.11543867
Xu, M., Qin, F., Xu, S., Xing, F., Cao, D., Li, C., & Yang, L. (2026) · IEEE Wireless Communications Letters · doi:10.1109/lwc.2026.3703873
“…perceptrons (MLPs) or Euclidean-based convolutional neural networks (CNNs), the proposed GCN-PANet exploits the inherent permutation invariance and message-passing capabilities of graph neural networks [12], [13] to capture the coupled interference patterns in the irregular CFmMIMO topology.”
Restricted access window for 802.11ah dense wireless networks: a comprehensive survey
Guedes, B. T., Albuquerque, C., Passos, D. G., & Passos, F. G. O. (2026) · Wireless Networks · doi:10.1007/s11276-026-04175-7
“…method, traffic-aware grouping algorithm [ 39], Message Passing Graph Neural Networks (MPGNNs) [21], Random Edge Graph Neural Network (REGNN) [22], Orthogonal Frequency-Division Multiple Access (OFDMA) [24], Grouping-based channel access with TWT Scheduling for OBSS (GTSO) [25], and graph…”
Milovančević, M. (2026) · Journal of AI-Driven Communication Engineering · doi:10.31875/2979-1081.2026.02.03
“The permutation-equivariant property provides a strong inductive bias that reduces sample complexity compared to fully connected architectures [24], making GNNs particularly valuable for large-scale deployment where re-training on every new topology is impractical.”
Decentralized Multi-Channel MANET Power Optimization Using Graph Neural Networks
Alter, T., Shlezinger, N., & Segal, M. (2026) · International Conference on Communications · doi:10.1109/icc59461.2026.11587961 · Cited by 1
“We commence with presenting the communication system in Subsection II-A, and formulate the decentralized power allocation optimization problem in Subsection II-B, which constitutes the starting point for MANET-GNN.”
Risk-Aware Scheduling and Placement for Bus-Stop Wireless Powered Communication Networks
Kim, D. J. H., & Joe, I. (2026) · IEEE Transactions on Green Communications and Networking · doi:10.1109/tgcn.2026.3713675
“Related neural architectures, including graph neural networks and permutation-equivariant models, have also been studied for wireless resource management because user indexing is arbitrary [37], [38], [39].”
Ma, Z., Gong, H., Duan, H., & Wang, X. (2026) · IEEE Internet of Things Journal · doi:10.1109/jiot.2026.3714884
“Eisen and Ribeiro [25] and Patel et al. [26] have used GNNs for wireless / communication resource allocation, optimizing energy in network-heavy scenarios, but not for full task assignment.”
Pei, E., Chen, Y., Gong, W., Li, W., & Sun, Y. (2026) · IEEE Transactions on Cognitive Communications and Networking · doi:10.1109/tccn.2026.3717900
“More recent graph neural network models, such as GCN [30], GAT [31], GraphSAGE [32], and R-GCN [33], have become representative graph learning tools, and graph learning has also been applied to wireless resource allocation [34], [35].”
Kumar, K., Srivastava, M. K., & Shukla, M. M. (2026) · Journal of Optical Communications · doi:10.1515/joc-2026-0230
“Meanwhile, graph neural networks (GNNs) are well suited to wireless resource management because communication networks have an inherent graph structure formed by access points, users and propagation links [17, 18].”
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
“Such ideas have been applied to power control [7], [8], power allocation [9], link scheduling [10], and fully-digital precoding [11].”
Wu, Y., & Zhu, H. (2026) · Entropy · doi:10.3390/e28080884
FedGAT: Federated Graph Attention for User Association and Interference Mitigation in C-RAN
Taleb, H. (2026) · Electronics · doi:10.3390/electronics15163492
A Coverage-Preserving Dual-Graph Neural Sleep Scheduler for Wireless Sensor Networks
Djediai, A., Balbal, S., & Khaldi, A. (2026) · Statistics Optimization & Information Computing · doi:10.19139/soic-2310-5070-4245
TANGCO: Learning Topology-Aware Capacity Allocation for Overload-driven Cascading FailuresPreprint
İrsoy, O., Akoglu, L., & Yağan, O. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2608.13212
Xu, Z., Liu, L., Wu, C., & Chan, S. (2026) · IEEE Transactions on Cognitive Communications and Networking · doi:10.1109/tccn.2026.3725076
“Unlike traditional deep learning models, GNNs are uniquely designed to capture the graph-based relationships between APs, enabling them to effectively model both local interference conditions and the global network structure [20].”
Graph Neural Networks as Surrogate Models for Aerodynamic Prediction: A Review
Prat, A. L., Schütz, T., & Gau, H. (2026) · SAE technical papers on CD-ROM/SAE technical paper series · doi:10.4271/2026-01-5068
Semantic Video Transmission Over LEO Satellites With Graph Neural Networks
Karaa, M., Ghazzai, H., Setti, G., & Sboui, L. (2026) · IEEE Transactions on Machine Learning in Communications and Networking · doi:10.1109/tmlcn.2026.3725994
“At the same time, aggregation-based and random-edge GNNs have been introduced to address decentralized and interference-limited environments with strong transferability across network sizes [43], [44].”
Zhen, L., Zhang, Q., Meng, Q., Liu, L., Su, J., Jiang, J., Jingling, L., & Xianyu, W. (2026) · IEEE Open Journal of the Communications Society · doi:10.1109/ojcoms.2026.3726171
Guo, J., Ouyang, C., Liu, Y., Nallanathan, A., & Ding, Z. (2026) · IEEE Transactions on Wireless Communications · doi:10.1109/twc.2026.3724376
“2) GNNs for Learning Wireless Policies: GNNs have been proposed to learn wireless policies, including precoding [15] and power allocation [16], [22].”
Kazim, M. Z., Abbas, S., & Fayaz, M. (2026) · SSRN Electronic Journal · doi:10.2139/ssrn.7443053
Sensing-Prior Dual-Domain Graph Attention for User Scheduling in Wi-Fi 7 MU-OFDMA
Liu, Y., Meng, J., Song, P., Zeng, C., Wu, W., & Liu, Y. (2026) · IEEE/CIC International Conference on Communications in China (ICCC) · doi:10.1109/iccc70295.2026.11680095
“Existing methods either refine the receiver, e.g., successive interference cancellation (SIC)/MMSE-SIC and data-driven variants such as DeepSIC and GNNSIC [5], or optimize scheduling/resource allocation with GNNs such as IGCNet [6] and REGNN [7].”
Graph Neural Modelling of Geo-Spatial Dependencies in V2V Communication Networks
Kurup, R. P., & K, M. K. T. (2026) · International Conference on Machine Intelligence for GeoAnalytics and Remote Sensing (MIGARS) · doi:10.1109/migars72065.2026.11688384
Graph Neural Networks for Dynamic Resource Allocation in Cloud-Edge Continuum
Natarajan, D., Lakshmanan, A., & Lakshmanan, P. (2021) · Journal of Intelligent Systems and Future Computing · doi:10.5281/zenodo.22843839
Gu, F., Zhao, X., & Chen, X. (2026) · Buildings · doi:10.3390/buildings16183720
Hierarchical Transformer for Power Allocation in Multi-Cell Massive MIMO Systems
Wu, J., Jin, J., Lou, M., Zhu, Y., Li, Y., & Zhao, H. (2026) · IEEE International Symposium on Personal, Indoor and Mobile Radio Communications (PIMRC) · doi:10.1109/pimrc69400.2026.11711673
Ban, T.-W. (2026) · Sensors · doi:10.3390/s26206603
“Third, both schedulers are trained at N = 10 and evaluated only for N ≤ 10, whereas several GNN-based studies report transfer to networks larger than those seen in training [8,10,12].”
InfluentialNijalingappa, Y., Ali, T. M., Vasudevan, A., MARC, H. R. C., & LEE, J. (2026) · Global and Stochastic Analysis · doi:10.64837/gsa.13.5.3
Abdullah, S. A. (2026) · Journal of Al-Farabi For Engineering Sciences · doi:10.59746/zgs95h36
A survey on AI methods for energy-efficient cell-free MIMO networks
Lee, M., Oh, S. M., Kim, J., & Park, S. (2026) · ICT Express · doi:10.1016/j.icte.2026.10.012
Metadata
- Authors
- Eisen, M., & Ribeiro, A.
- Affiliations
- Intel (United States) (US)
- University of Pennsylvania (US)
- Venue
- IEEE Transactions on Signal Processing
- Published
- 2020
- Volume
- 68
- Pages
- 2977-2991
- ISSN
- 1053-587X (ISSN-L)
- 1941-0476
- OpenAlex
- W2971598376
- Cited by
- 327
- References
- 41
- Citations by year
- 2026: 31 · 2025: 53 · 2024: 59 · 2023: 60 · 2022: 56 · 2021: 44 · 2020: 23
- FWCI
- 15.72
- Type
- journal article
- Language
- en
- Open access
- green
- arxiv.org
- Licence
- Locations
- IEEE Transactions on Signal Processing (publishedVersion)
- arXiv (Cornell University) (submittedVersion)
- arXiv (Cornell University)
- Funders
- ARL DCIST CRA (W911NF-17-2-0181)
- Intel Science and Technology Center for Wireless Autonomous Systems
- Topics
- Physical Sciences › Engineering › Electrical and Electronic Engineering › Advanced Wireless Network Optimization
- Physical Sciences › Computer Science › Computer Networks and Communications › Wireless Communication Networks Research
- Physical Sciences › Computer Science › Artificial Intelligence › Advanced Graph Neural Networks
- Keywords
- wireless resource allocation, graph neural networks, permutation equivariance, model-free reinforcement learning, random graphs, convolutional neural networks
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.