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 101–200 of 344
How Neural Architectures Affect Deep Learning for Communication Networks?
Shen, Y., Zhang, J., & Letaief, K. B. (2022) · International Conference on Communications · doi:10.1109/icc45855.2022.9839205 · Cited by 19
“GNNs achieve good scalability, generalization, and interpretability by exploiting the permutation invariance property in communication networks [11], [13], [17].”
Improving Learning Efficiency for Wireless Resource Allocation with Symmetric Prior
Sun, C., Wu, J., & Yang, C. (2022) · IEEE Wireless Communications · doi:10.1109/mwc.003.21003437 · Cited by 24
“These structures have been demonstrated capable of reducing the numbers of training samples and trainable parameters of DNNs [13] and generalizing well to the problem scales (say the number of users) [14].”
InfluentialPower Allocation for Wireless Federated Learning Using Graph Neural Networks
Li, B., Swami, A., & Segarra, S. (2022) · IEEE International Conference on Acoustics Speech and Signal Processing · doi:10.1109/icassp43922.2022.9747764 · Cited by 23
GNN-Based Meta-Learning Approach for Adaptive Power Control in Dynamic D2D Communications
Luu, V.-C., & Hong, J. (2023) · IEEE Transactions on Vehicular Technology · doi:10.1109/tvt.2023.3332809 · Cited by 18
Rao, C. K., Sahoo, S. K., & Yanine, F. F. (2024) · Energy Harvesting and Systems · doi:10.1515/ehs-2023-0124 · Cited by 19
Recursive GNNs for Learning Precoding Policies With Size-Generalizability
Guo, J., & Yang, C. (2024) · IEEE Transactions on Machine Learning in Communications and Networking · doi:10.1109/tmlcn.2024.3480044 · Cited by 25
“Their generalizability to unseen graph sizes [11], [12], [13], [14], scalability to large scale systems [11], [12], [14], and higher sample efficiency [11], [12], [13], [16], [17] have been demonstrated by extensive simulations.”
BackgroundMethodInfluentialModel-Free Learning of Optimal Ergodic Policies in Wireless Systems
Kalogerias, D. S., Eisen, M., Pappas, G. J., & Ribeiro, A. (2020) · IEEE Transactions on Signal Processing · doi:10.1109/tsp.2020.3030073 · Cited by 20
“Further, by using more efficient policy parameterizations which exploit (stochastic) network structure, such as Random Edge Graph Neural Networks (REGNNs) [24], it is possible to not only reduce the number of the involved free parameters (Nφ in particular), but also train REGNN-based resource allocation policies on sm…”
BackgroundMethodInfluentialSelf-citationResource Allocation via Graph Neural Networks in Free Space Optical Fronthaul Networks
Gao, Z., Eisen, M., & Ribeiro, A. (2020) · Globecom · doi:10.1109/globecom42002.2020.9322426 · Cited by 22
Self-citationLearning-Based Resource Allocation in Heterogeneous Ultradense Network
Zhang, X., Zhang, Z. J., & Yang, L. (2022) · IEEE Internet of Things Journal · doi:10.1109/jiot.2022.3173210 · Cited by 18
“Eisen and Ribeiro [30], Cui et al. [31], Shen et al. [32] proposed different graph neural networks to approach this problem.”
BackgroundGraph Representation Learning for Wireless Communications
Mohsenivatani, M., Ali, S., Ranasinghe, V., Rajatheva, N., & Latva‐aho, M. (2023) · IEEE Communications Magazine · doi:10.1109/mcom.001.2200810 · Cited by 16
Li, Y., & Liu, Y. (2024) · IEEE Transactions on Communications · doi:10.1109/tcomm.2024.3385919 · Cited by 22
“As it is generally hard to get the labels for radio resource management, a prevalent solution is to apply unsupervised learning [28], [29], [30], [31], [32], [33] or reinforcement learning [34], where the loss function or reward is constructed directly from the optimization objective.”
BackgroundInfluentialLiu, Y., She, C., Zhong, Y. F., Hardjawana, W., Zheng, F., & Vucetic, B. S. (2023) · IEEE Transactions on Vehicular Technology · doi:10.1109/tvt.2023.3325464 · Cited by 18
“To tackle this issue, different kinds of Graph Neural Networks (GNN) were applied in [28], [29] and [30], respectively.”
MethodQuantization Analysis and Robust Design for Distributed Graph Filters
Saad, L. B., Beferull‐Lozano, B., & Isufi, E. (2021) · IEEE Transactions on Signal Processing · doi:10.1109/tsp.2021.3139208 · Cited by 18
On the size generalizibility of graph neural networks for learning resource allocation
Wu, J.-J., Sun, C., & Yang, C. (2024) · Science China Information Sciences · doi:10.1007/s11432-023-3880-y · Cited by 18
“Depending on the considered problem, a resource allocation policy may exhibit one-dimensional (1D)-PE, two-dimensional-PE (e.g., precoding [12]), joint-PE (e.g., power control or link scheduling [2,9]) properties, their combinations [11].”
BackgroundMethodInfluentialUnsupervised Optimal Power Flow Using Graph Neural NetworksPreprint
Owerko, D., Gama, F., & Ribeiro, A. (2022) · arXiv (Cornell University) · doi:10.48550/arxiv.2210.09277 · Cited by 15
“Meanwhile, we use the square function as the penalty for the equality constraints:”
MethodSelf-citationGraph Neural Network-Based Node Deployment for Throughput Enhancement
Yang, Y., Zou, D., & He, X. (2023) · IEEE Transactions on Neural Networks and Learning Systems · doi:10.1109/tnnls.2023.3281643 · Cited by 14
“For instance, the edge features used in [22], [23], and [24] are information such as channel response, channel state, and equivalent channel.”
BackgroundMethodMeta-Gating Framework for Fast and Continuous Resource Optimization in Dynamic Wireless Environments
Hou, Q., Lee, M., Yu, G., & Cai, Y. (2023) · IEEE Transactions on Communications · doi:10.1109/tcomm.2023.3292257 · Cited by 17
“In [7] and [8], GNNs were designed to optimally allocate resources across a set of transceiver pairs in a wireless network.”
BackgroundDeep Learning for MANET Routing
Danilchenko, K., Azoulay, R., Reches, S., & Haddad, Y. (2023) · IEEE Transactions on Machine Learning in Communications and Networking · doi:10.1109/tmlcn.2023.3324280 · Cited by 16
“Clearly, the per-link scheduling problem is a particular case of the per-route scheduling problem, and both are NP-hard problems [12].”
BackgroundGraph Representation Learning for Contention and Interference Management in Wireless Networks
Gu, Z., Vucetic, B. S., Chikkam, K., Aliberti, P., & Hardjawana, W. (2024) · IEEE/ACM Transactions on Networking · doi:10.1109/tnet.2024.3355935 · Cited by 21
“Then, GNNs obtain the control decisions by repeatedly aggregating features of neighboring vertices (or edges) along with trainable parameters [17], [18].”
MethodInfluentialMulti-Agent Reinforcement Learning for Power Control in Wireless Networks via Adaptive Graphs
Amorosa, L. M., Skocaj, M., Verdone, R., & Gündüz, D. (2024) · International Conference on Communications · doi:10.1109/icc51166.2024.10622170 · Cited by 14
“Notably, these properties find practical application in works such as [5], where GNNs are harnessed to capture the dynamic structure of fading channel states for the purpose of learning optimal resource allocation policies in wireless networks.”
BackgroundWireless control: Retrospective and open vistas
Pezzutto, M., Dey, S., Garone, E., Gatsis, K., Johansson, K. H., & Schenato, L. (2024) · Annual Reviews in Control · doi:10.1016/j.arcontrol.2024.100972 · Cited by 16
LaMar, S., Gosselin, J. J., Caceres, I., Kapple, S., & Jayasumana, A. P. (2021) · IEEE Military Communications Conference (MILCOM) · doi:10.1109/milcom52596.2021.9652977 · Cited by 14
“…same logic can be applied to surmise the algorithms usefulness for networking problems, which is precisely what other authors have done to learn routing protocols [13], predict jitter and delay [14], optimize resource allocation in wireless networks [15], and distributed transmission scheduling.”
BackgroundAbode, D., Adeogun, R., & Berardinelli, G. (2024) · IEEE Open Journal of the Communications Society · doi:10.1109/ojcoms.2024.3396749 · Cited by 18
“The already trained parameters are then used in the execution phase to determine the transmit power [19], [20], [21], [22], [23], [24], [25], [26], [27], [28], [29], [30], [31], [32].”
BackgroundMethodLearning End-to-End Hybrid Precoding for Multi-User mmWave Mobile System With GNNs
Wang, R., Yang, C., Han, S., Wu, J., Han, S., & Wang, X. (2024) · IEEE Transactions on Machine Learning in Communications and Networking · doi:10.1109/tmlcn.2024.3420269 · Cited by 17
“By harnessing the permutation properties of wireless policies [33], GNNs can achieve superior performance with fewer training samples and trainable parameters than fully-connected neural networks (FNNs) and convolutional neural networks (CNNs) [15], [34].”
BackgroundLe, H. A., Chien, T. V., & Choi, W. (2025) · IEEE Transactions on Communications · doi:10.1109/tcomm.2025.3558504 · Cited by 17
“With a promising potential, GNNs have been extended to solve many scalable wireless communication systems, such as power allocation, beamforming design, and so on [15]–[21].”
Tackling Resource Allocation for Decentralized Federated Learning: A GNN-Based Approach
Meng, C., Tang, M., Setayesh, M., & Wong, V. W. S. (2025) · IEEE Transactions on Mobile Computing · doi:10.1109/tmc.2025.3562834 · Cited by 16
“In recent years, graph neural networks (GNNs) have been developed as effective tools to solve resource allocation problems in wireless networks [31], [32], [33], [34].”
ICGNN: Graph Neural Network Enabled Scalable Beamforming for MISO Interference Channels
He, C., Lu, Y., Ai, B., Dobre, O. A., Ding, Z., & Niyato, D. T. (2025) · IEEE Transactions on Mobile Computing · doi:10.1109/tmc.2025.3570648 · Cited by 15
“In[13],theauthorsextendedtheREGNNtosolveasetofsystem utilitymaximizationproblemsforwirelessad-hocorcellular networks.”
Channel Estimation for mmWave Massive MIMO Systems Using Graph Neural Networks
Ye, M., Liang, X., Pan, C., Xu, Y., Jiang, M., & Li, C. (2023) · IEEE/CIC International Conference on Communications in China (ICCC) · doi:10.1109/iccc57788.2023.10233621 · Cited by 14
Zero-Shot Learning for Beam Management in LEO Satellite Communications
Geng, Z., She, C., Wang, R., Li, Y., & Vucetic, B. S. (2024) · IEEE Transactions on Wireless Communications · doi:10.1109/twc.2024.3392644 · Cited by 12
“8 with the GNN in [28], and develop a GRNN algorithm.”
MethodPower allocation using spatio-temporal graph neural networks and reinforcement learning
Jamshidiha, S., Pourahmadi, V., Mohammadi, A., & Bennis, M. (2024) · Wireless Networks · doi:10.1007/s11276-024-03814-1 · Cited by 12
“Optimal power allocation using GNNs in a large-scale wireless network is studied in [24, 25], and a scalable GNN architecture is introduced based on the GCN architecture that can outperform existing heuristic methods.”
Scalable Spatial and Geometric Learning Approach for Joint Power Control and Channel Allocation
Marwani, M., & Kaddoum, G. (2024) · IEEE Transactions on Wireless Communications · doi:10.1109/twc.2024.3449036 · Cited by 14
Stability of Graph Neural Networks to Relative Perturbations
Gama, F., Ribeiro, A., & Bruna, J. (2020) · IEEE International Conference on Acoustics Speech and Signal Processing · doi:10.1109/icassp40776.2020.9054341 · Cited by 12
“GNNs offer a nonlinear transformation of the input data that has achieved remarkable performance in wireless networks [12], decentralized control of robot swarms [13] and recommendation systems [14], among others [15,16].”
BackgroundSelf-citationLearning Beamforming for RIS-aided Systems with Permutation Equivariant Graph Neural Networks
Zhao, B., & Yang, C. (2023) · IEEE Vehicular Technology Conference · doi:10.1109/vtc2023-spring57618.2023.10200544 · Cited by 12
“The following proposition gives the PE property of the channel estimation policy, which can be proved with similar technique as in [11] and hence the proof is omitted.”
MethodGraph Neural Networks Over the Air for Decentralized Tasks in Wireless Networks
Gao, Z., & Gündüz, D. (2025) · IEEE Transactions on Signal Processing · doi:10.1109/tsp.2025.3534685 · Cited by 18
“The works in [28], [29] utilize GNNs to control transmission powers in interference networks based on instantaneous CSI, while [6], [7] impose real-world constraints and allocate powers with GNNs in a model-free manner.”
Transmission Power Control using Deep Neural Networks in TDMA-based Ad-hoc Network Clusters
Azoulay, R., Danilchenko, K., Haddad, Y., & Reches, S. (2021) · International Wireless Communications and Mobile Computing Conference (IWCMC) · doi:10.1109/iwcmc51323.2021.9498729 · Cited by 12
Active Sensing for Multiuser Beam Tracking With Reconfigurable Intelligent Surface
Han, H., Jiang, T., & Yu, W. (2024) · IEEE Transactions on Wireless Communications · doi:10.1109/twc.2024.3496393 · Cited by 18
“Toward this end, GNN is an indispensable tool for learning these spatial relations [20], [21], [22], [23].”
Adaptive Wireless Power Allocation with Graph Neural Networks
Naderializadeh, N., Eisen, M., & Ribeiro, A. (2022) · IEEE International Conference on Acoustics Speech and Signal Processing · doi:10.1109/icassp43922.2022.9746240 · Cited by 12
Self-citationRandall, M., Belzarena, P., Larroca, F., & Casas, P. (2022) · IEEE Latin-American Conference on Communications · doi:10.1109/latincom56090.2022.10000511 · Cited by 14
Intelligent Beamforming for UAV-Assisted IIoT Based on Hypergraph Inspired Explainable Deep Learning
Zhao, H., Liu, K., Liu, M., Garg, S., & Alrashoud, M. R. (2023) · IEEE Transactions on Consumer Electronics · doi:10.1109/tce.2023.3325128 · Cited by 11
“In details, the authors of [45] proposed a GNN named random edge graph neural network to extend the given power control scheme to the networks with different sizes with still satisfying performances.”
BackgroundResource allocation in heterogeneous network with node and edge enhanced graph attention network
Sun, Q., He, Y., & Petrosian, O. (2024) · Applied Intelligence · doi:10.1007/s10489-024-05391-4 · Cited by 11
A Unified Framework for Combinatorial Optimization Based on Graph Neural Networks
Jin, Y., Yan, X., Liu, S., & Wang, X. (2026) · Engineering · doi:10.1016/j.eng.2025.06.047 · Cited by 15
Discriminability of Single-Layer Graph Neural Networks
Pfrommer, S., Ribeiro, A., & Gama, F. (2021) · IEEE International Conference on Acoustics Speech and Signal Processing · doi:10.1109/icassp39728.2021.9414583 · Cited by 11
“Data generated by networks are increasingly common in power grids [1], robotics [2, 3], and wireless sensor networks [4, 5] among others [6,7].”
BackgroundSelf-citationDeep Neural Networks With Data Rate Model: Learning Power Allocation Efficiently
Guo, J., & Yang, C. (2023) · IEEE Transactions on Communications · doi:10.1109/tcomm.2023.3237241 · Cited by 13
“…many wireless policies exhibit permutation equivariance (PE) properties [21], [22], [23], which can be harnessed by designing graph neural networks (GNNs) [22], [23], [24] or by introducing parameter sharing into fully-connected DNNs (FNNs) to obtain permutation equivariance neural networks [25].”
BackgroundNaderializadeh, N. (2021) · arXiv (Cornell University) · doi:10.48550/arxiv.2110.01722 · Cited by 14
“…learning, and meta-learning, which do not necessarily rely on (extensive) labeling of the data for training the underlying neural networks [3–5 , In this paper, we consider graph representation learning algorithms for the problem of link scheduling in wireless networks with multiple…”
Prasanna, B. T., Ramya, D., Shelke, N. M., Fernandes, J. B., Galety, M. G., & Ashok, M. (2023) · Wireless Networks · doi:10.1007/s11276-023-03540-0 · Cited by 12
“The random edge graph neural network (REGNN) introduced by Eisen et al. [12] is used to conduct convo-lutions on random graphs produced by fading interference patterns in a wireless network.”
Zhong, Y. (2025) · Discover Internet of Things · doi:10.1007/s43926-025-00096-7 · Cited by 10
Zhao, Y., Zhou, Y., Wang, Z., Shi, Y., Cheng, N., & Zhou, H. (2025) · IEEE Transactions on Wireless Communications · doi:10.1109/twc.2025.3550315 · Cited by 11
“Additionally, GNN possesses unique features of permutation invariance and permutation equivalence [24], [25], [26], which set it apart from FCNNs and enable excellent algorithmic scalability.”
Deep Learning for Beamforming in Multi-User Continuous Aperture Array Systems
Guo, J., Liu, Y., Shin, H., & Nallanathan, A. (2025) · IEEE Transactions on Wireless Communications · doi:10.1109/twc.2025.3633569 · Cited by 12
“…the environmental parameters such as channel coefficients to decisions such as beamforming) from problems in SPD systems, including fully connected neural networks (FNNs) [16], [17], convolutional neural networks (CNNs) [18] and graph neural networks (GNNs) [19], [20], [21], [22], [23], [24], [25].”
Wide and Deep Graph Neural Networks with Distributed Online Learning
Gao, Z., Ribeiro, A., & Gama, F. (2021) · IEEE International Conference on Acoustics Speech and Signal Processing · doi:10.1109/icassp39728.2021.9415046 · Cited by 10
Self-citationDeep learning method for delay minimization in MANET
Danilchenko, K., Azoulay, R., Reches, S., & Haddad, Y. (2022) · ICT Express · doi:10.1016/j.icte.2022.01.004 · Cited by 10
A random graph-based neural network approach to assess glioblastoma progression from perfusion MRI
Ayaz, H., Khosravi, H., McLoughlin, I., Tormey, D., Özsunar, Y., & Unnikrishnan, S. (2023) · Biomedical Signal Processing and Control · doi:10.1016/j.bspc.2023.105286 · Cited by 10
Cognitive radio resource scheduling using an adaptive multiobjective evolutionary algorithm
Wang, H., Wang, Y., Zeng, F., & Wang, J. (2024) · Applied Intelligence · doi:10.1007/s10489-024-05398-x · Cited by 10
“Eisen et al. proposed a variant of a graph neural network, the random edge graph neural network (REGNN), which takes the channel and node states as inputs and combines them through a nonlinear graph convolution filter to obtain the resource allocation function [8].”
Biased Backpressure Routing Using Link Features and Graph Neural Networks
Zhao, Z., Radojičić, B., Verma, G., Swami, A., & Segarra, S. (2024) · IEEE Transactions on Machine Learning in Communications and Networking · doi:10.1109/tmlcn.2024.3461711 · Cited by 15
“To address this challenge, schemes of unsupervised learning [34], [35], [36], [37], primal-dual learning [38], [39], [40], [41], and reinforcement learning (RL) [29], [42], [43], [44], [46], [47] have been developed.”
BackgroundAn Optimal Resource Allocation Method for IIoT Network
Goswami, P., Mukherjee, A., Chatterjee, P., & Yang, L. (2021) · International Conference on Distributed Computing and Networking (ICDCN) · doi:10.1145/3427477.3429988 · Cited by 10
“In some cases, Lagrangian dual domain [24], dual descent [8] and other heuristic approaches are used to achieve those purposes.”
MethodAn endogenous intelligent architecture for wireless communication networks
He, S. (2023) · Wireless Networks · doi:10.1007/s11276-023-03545-9 · Cited by 11
“…competitive results in comparison to conventional model-based algorithms in areas such as channel estimation [3, 4], signal detection [5], channel decoding [6, 7], spectrum sensing [8, 9], power allocation [10–12], etc. References [13–15] systematically reviewed the related work in this direction.”
Jing, W., Wang, J., Ren, J., Lu, Z., Shao, H., & Wen, X. (2024) · IEEE Transactions on Vehicular Technology · doi:10.1109/tvt.2024.3497603 · Cited by 9
“GNNs have shown promise in solving power control problems [25], [26], [27], [28], [29], [30], beamforming optimization [31], [32], [33], [34], link or user scheduling policy optimization [35], [36], and joint channel and power allocation policy optimization [37].”
Liu, Z., Zhang, J., Zhu, Y., Shi, E., & Ai, B. (2025) · IEEE Transactions on Signal Processing · doi:10.1109/tsp.2025.3574005 · Cited by 9
“This has triggered a rapidly growing research field, namely multidimensional graph neural networks (MDGNNs) [24], [29], [30], which belong to the “ edge ” framework and learn effective wireless policies by updating hidden representations of hyper-edges to reduce information loss [31].”
A Primer on Temporal Graph Learning
Rahman, A. U., Elhag, A. A., & Coon, J. P. (2025) · ACM Computing Surveys · doi:10.1145/3771693 · Cited by 12
“As the node features evolve, we get a temporal graph that can be analyzed to perform dynamic resource allocation [Eisen and Ribeiro 2020] to optimise the data rate of all the users collectively.”
Hybrid GNN-Centric Architectures for AI-Native 6G Wireless Networks: A Comprehensive Survey
Ghourtani, M. R., Norouzi, S., Chen, J. Y., Ahmadi, H., Braun, T. I., Chowdhury, K. R., & Burr, A. G. (2026) · IEEE Communications Surveys & Tutorials · doi:10.1109/comst.2026.3681198 · Cited by 13
Learning for Intelligent Hybrid Resource Allocation in MEC-Assisted RAN Slicing Network
Zheng, C., Huang, Y., Zhang, C., & Quek, T. Q. S. (2024) · IEEE Transactions on Vehicular Technology · doi:10.1109/tvt.2024.3388164 · Cited by 9
“By further combining the GNN with the convolutional neural network whose dimensionality is small and does not scale with network size, authors in [20] propose a random edge graph convolutional network (GCN) to learn the optimal RA policy in a multi-node RAN.”
MethodCharacterizing the Communication Requirements of GNN Accelerators: A Model-Based Approach
Guirado, R., Jain, A., Abadal, S., & Alarcón, E. (2021) · IEEE International Symposium on Circuits and Systems proceedings · doi:10.1109/iscas51556.2021.9401612 · Cited by 9
Stability of Neural Networks on Manifolds to Relative Perturbations
Wang, Z., Ruiz, L., & Ribeiro, A. (2022) · IEEE International Conference on Acoustics Speech and Signal Processing · doi:10.1109/icassp43922.2022.9747865 · Cited by 11
Self-citationLearning Stochastic Graph Neural Networks With Constrained Variance
Gao, Z., & Isufi, E. (2023) · IEEE Transactions on Signal Processing · doi:10.1109/tsp.2023.3244101 · Cited by 9
“Graph neural networks (GNNs) exploit this structural information to model task-relevant representations from graph signals [3], [4], [5], [6], which have found applications in recommender systems [7], [8], multi-agent coordination [9], [10], and wireless communications [11], [12], [13].”
BackgroundLearning Autonomy in Management of Wireless Random Networks
Lee, H., Lee, S. H., & Quek, T. Q. S. (2021) · IEEE Transactions on Wireless Communications · doi:10.1109/twc.2021.3089701 · Cited by 9
“tions are presented [30], [31], handcrafted interaction policies”
BackgroundMethodA Topology-Aware GNN Learning Approach for Energy Optimization in Multihop LoRa Networks
Yang, H., Wu, X. S., Ji, H., Huang, Z., & Fang, J. (2025) · IEEE Internet of Things Journal · doi:10.1109/jiot.2025.3606698 · Cited by 10
“In [20], a random edge GNN (REGNN) was proposed to optimize resource allocation in the wireless network.”
Contrastive Self-Supervised Learning for Wireless Power Control
Naderializadeh, N. (2021) · IEEE International Conference on Acoustics Speech and Signal Processing · doi:10.1109/icassp39728.2021.9413621 · Cited by 13
“and unsupervised learning, which do not necessarily rely on manual labeling of the data for training the underlying neural networks [5, 6, 7, 8, 9, 10].”
BackgroundLink Activation Using Variational Graph Autoencoders
Jamshidiha, S., Pourahmadi, V., Mohammadi, A., & Bennis, M. (2021) · IEEE Communications Letters · doi:10.1109/lcomm.2021.3076190 · Cited by 8
“This problem is considered as a model-free statistical learning problem in [2, 3], where a graph neural network with lower computational complexity compared to a fully connected neural network has been utilized.”
BackgroundEdge-Wise Gated Graph Neural Network for User Association in Massive URLLC
Liu, X., She, C., Li, Y., & Vucetic, B. S. (2021) · IEEE Globecom Workshops (GC Wkshps) · doi:10.1109/gcwkshps52748.2021.9682005 · Cited by 9
“To address the scalability issue, the authors of [9] and [10] applied different types of graph neural networks (GNNs) to represent wireless networks [9, 10].”
BackgroundLearning to Transmit With Provable Guarantees in Wireless Federated Learning
Li, B., Perazzone, J. B., Swami, A., & Segarra, S. (2023) · IEEE Transactions on Wireless Communications · doi:10.1109/twc.2023.3341400 · Cited by 10
“Turning to another set of methods, machine learning – in particular, graph neural networks (GNNs) – has been recently applied to solve power allocation problems [34], [35], [36], [37], [38], [39].”
MethodResource Allocation in Large-Scale Wireless Control Systems with Graph Neural Networks
Lima, V., Eisen, M., Gatsis, K., & Ribeiro, A. (2020) · IFAC-PapersOnLine · doi:10.1016/j.ifacol.2020.12.378 · Cited by 8
“, 2019); in particular, we substitute graph neural networks (GNNs) instead of traditional neural networks (Eisen and Ribeiro, 2019; Shen et al., 2019).”
BackgroundMethodSelf-citationHu, H., Xie, Z., Shi, H., Liu, B., Zhao, H., & Gui, G. (2024) · IEEE Transactions on Vehicular Technology · doi:10.1109/tvt.2024.3430353 · Cited by 8
“(cid:2) Random Edge Graph Neural Network (REGNN) [21]: un-supervised learning method in which better performance of binary power allocation for Rayleigh channel is achieved, especially in high-noise wireless environment.”
BackgroundLearning Cooperative Beamforming with Edge-Update Empowered Graph Neural Networks
Wang, Y., Li, Y., Shi, Q., & Wu, Y. (2023) · International Conference on Communications · doi:10.1109/icc45041.2023.10278864 · Cited by 7
A Size-Generalizable GNN for Learning Precoding
Guo, J., & Yang, C. (2023) · IEEE Vehicular Technology Conference · doi:10.1109/vtc2023-fall60731.2023.10333362 · Cited by 8
“…it is important to design size-generalizable DNNs. Recently, graph neural networks (GNNs) have been introduced to learn power control/allocation [4]–[6], link scheduling [7], and precoding policies [8], [9], motivated by their generalizability to unseen graph topology [4], [5], [8], scalability…”
BackgroundInfluentialA Dynamic Power Allocation Approach for Downlink Cell-Free Massive MIMO With Graph Neural Network
Zhao, Y., Zhang, F., Zhou, Z., & Hu, G. (2024) · IEEE Transactions on Vehicular Technology · doi:10.1109/tvt.2024.3515056 · Cited by 7
“Therearesomepioneerworksusing GNNforcomplicatedproblemsinwirelessnetworks,e.g.,[23], [24],[25],[26],[27].”
Chen, S., Tan, C. W., Zhai, X., & Poor, H. V. (2025) · IEEE Transactions on Green Communications and Networking · doi:10.1109/tgcn.2025.3541338 · Cited by 7
“For machine learning baselines, we refer to the deep neural network (DNN) model in [15], graph neural networks (GNN) in [17], and the deep belief network (DBN) in [11].”
Zhou, M. F., Li, S., Yuan, P., Zhang, J., Zhang, J., Yang, L., & Zhu, H. (2025) · IEEE Internet of Things Journal · doi:10.1109/jiot.2025.3569075 · Cited by 7
“…Feature Aggregation: Multiscale convolution aggregates neighbor information from different ranges to enhance the model’s ability to capture complex spatial correlations [44], which solves the problem of limited receptive field in traditional single-layer graph convolutional networks (GCN).”
Graph Neural Networks based Resource Allocation in Heterogeneous Wireless Networks
Cheng, P., Chen, G., & Han, Z. (2022) · doi:10.1145/3570236.3570293 · Cited by 8
“Based on [5], the random edge graph neural network (REGNN) was proposed in [7] to implement a power control (PC) strategy in different networks.”
MethodApproaching Globally Optimal Energy Efficiency in Interference Networks via Machine Learning
Peng, B., Besser, K.-L., Raghunath, R., & Jorswieck, E. A. (2023) · IEEE Transactions on Wireless Communications · doi:10.1109/twc.2023.3269770 · Cited by 7
“The distributed nature of the GNN and its ability to perform message passing and information aggregation make it a competitive candidate to optimize cellular networks [39], [40], [41].”
BackgroundMethodTraining Graph Neural Networks on Growing Stochastic Graphs
Cerviño, J., Ruiz, L., & Ribeiro, A. (2023) · IEEE International Conference on Acoustics Speech and Signal Processing · doi:10.1109/icassp49357.2023.10094894 · Cited by 7
“In practice, GNNs have shown state-of-the-art performance in several learning tasks [3, 4], and have found applications in biology [5–7], recommendation systems [8–10], robotics [11, 12], and wireless communications [13,14].”
BackgroundSelf-citationYi Bing, S. O. N. G., Yang, T., Khalilsarai, M. B., & Caire, G. (2023) · International Conference on Communications · doi:10.1109/icc45041.2023.10279459 · Cited by 8
Liu, Z., Luo, C., Xie, J., & Luo, Y. (2024) · IEEE Wireless Communications Letters · doi:10.1109/lwc.2024.3362050 · Cited by 8
“The original model architectures are kept according to the previous works [9], [12].”
BackgroundMethodA Traffic-Aware Graph Neural Network for User Association in Cellular Networks
Jamshidiha, S., Pourahmadi, V., & Mohammadi, A. (2025) · IEEE Transactions on Mobile Computing · doi:10.1109/tmc.2025.3545464 · Cited by 9
“While this framework is usually referred to as “unsupervised learning” in the literature (e.g. [37], [38]), it is more appropriate to refer to this approach as policy gradient RL.”
Resource Allocation for Platoon Oriented Vehicular Communications: A Neural Network Approach
Fan, C., Li, C., Zhang, Y., Yue, W., & Zheng, J. (2021) · International Conference on Communications · doi:10.1109/icc42927.2021.9500878 · Cited by 7
“Although the optimization problems can be formulated [6], if the algorithms with high computation complexity are acceptable in realistic scenarios needs to be assessed further.”
BackgroundDecentralized Wireless Resource Allocation with Graph Neural Networks
Wang, Z., Eisen, M., & Ribeiro, A. (2020) · Asilomar Conference on Signals, Systems and Computers (ACSSC) · doi:10.1109/ieeeconf51394.2020.9443326 · Cited by 9
“We also compare with the existing centralized Selection Graph Neural Network method [6].”
BackgroundResultSelf-citationXie, Y., Wang, L., Chang, Z., Xu, L., Bi, S., & Han, Z. (2025) · IEEE Transactions on Wireless Communications · doi:10.1109/twc.2025.3625295 · Cited by 6
Zhang, J., Miao, C., Huang, J., & Zhao, R. (2024) · Ad Hoc Networks · doi:10.1016/j.adhoc.2024.103509 · Cited by 8
Efficient MU-MIMO Beamforming Based on Majorization-Minimization and Deep Unfolding
Xu, Q., Sun, J., & Xu, Z. (2025) · IEEE Transactions on Wireless Communications · doi:10.1109/twc.2025.3541144 · Cited by 7
“Recently, deep learning methods have been widely considered in a series of wireless communication applications [15], including resource management [16], [17], signal detection [18], [19], channel estimation [20], [21], [22] and beamforming [17], [23], [24], [25], [26].”
State-Augmented Opportunistic Routing in Wireless Communication Systems with Graph Neural Networks
Das, S., Naderializadeh, N., & Ribeiro, A. (2025) · IEEE International Conference on Acoustics Speech and Signal Processing · doi:10.1109/icassp49660.2025.10888417 · Cited by 7
“Extensive research has been carried out to focus on stochastic network utility maximization (NUM) [3], radio resource management (RRM) [4, 5, 6], routing [7, 8] and scheduling [9].”
InfluentialSelf-citationA GNN-Based Learning Approach for Energy Optimization in Relay-Assisted IoT Networks
Yang, H., Ji, H., Huang, Z., & Wu, X. S. (2025) · IEEE Wireless Communications and Networking Conference · doi:10.1109/wcnc61545.2025.10978775 · Cited by 6
“Network topology illustration of multi-hop LoRa. allocation [10], [11], and joint user association and power allocation [12].”
Architectural Design for Digital Twin Networks
Wieme, J., Baert, M., & Hoebeke, J. (2025) · Network · doi:10.3390/network5030024 · Cited by 8
Designing Heterogeneous GNNs With Desired Permutation Properties for Wireless Resource Allocation
Zhao, J., Yang, C., Liu, T., Han, S., & Wang, X. (2025) · IEEE Open Journal of the Communications Society · doi:10.1109/ojcoms.2025.3612442 · Cited by 7
“One motivation of adopting GNNs is to leverage the permutation prior for reducing training costs [10], [13], [14], [16], [22], enabling size generalizability [8], [10], [11], [14], [20], [21], [22], [23], and improving learning performance [14], [24].”
InfluentialStable and Transferable Wireless Resource Allocation Policies Via Manifold Neural Networks
Wang, Z., Ruiz, L., Eisen, M., & Ribeiro, A. (2022) · IEEE International Conference on Acoustics Speech and Signal Processing · doi:10.1109/icassp43922.2022.9746313 · Cited by 8
“Graph neural networks (GNNs) [6] have been recently considered in many wireless resource allocation problems due to their low dimensionality and invariance to network Supported by NSF CCF 1717120, Theorinet Simons. topological structure [4,5,7,8].”
Self-citationHow to Improve Learning Efficiency of GNN for Precoding?
Guo, J., & Yang, C. (2023) · IEEE Vehicular Technology Conference · doi:10.1109/vtc2023-spring57618.2023.10200078 · Cited by 6
“For example, graph neural networks (GNNs) were proposed to learn power allocation policies in [5], [6] and precoding policies in [2], [7], where the permutation equivariance (PE) properties of the policies are harnessed.”
BackgroundBai, J., Zhao, X., Wang, Z., Wang, D., Zhang, L., Zhang, C., & Wu, Q. (2024) · IEEE Access · doi:10.1109/access.2024.3381513 · Cited by 6
“…the EP-WPN is equipped with network autonomy, capable of autonomously accomplishing intelligent real-time scheduling of communication resources [45], autonomous planning and intelligent decision-making, service level agreement guarantee, self-configuration and self-optimization dynamic…”
BackgroundAutomatic AI Model Selection for Wireless Systems: Online Learning via Digital Twinning
Hou, Q., Zecchin, M., Park, S., Cai, Y., Yu, G., Chowdhury, K. R., & Simeone, O. (2025) · IEEE Transactions on Wireless Communications · doi:10.1109/twc.2025.3527741 · Cited by 7
“2, in which the distribution p ( c ) reflects macroscopic long-term trends, including the spatial density of devices [30], [31], [33].”
Fang, H., Huang, K., Ye, H., Guo, C., Liang, L., Li, X., & Jin, S. (2025) · IEEE Transactions on Vehicular Technology · doi:10.1109/tvt.2025.3577365 · Cited by 7
Node-Variant Graph Filters in Graph Neural Networks
Gama, F., Anderson, B. G., & Sojoudi, S. (2022) · IEEE Data Science Workshop (DSW) · doi:10.1109/dslw53931.2022.9820575 · Cited by 10
Transferable Policies for Large Scale Wireless Networks with Graph Neural Networks
Eisen, M., & Ribeiro, A. (2020) · IEEE International Conference on Acoustics Speech and Signal Processing · doi:10.1109/icassp40776.2020.9054124 · Cited by 7
“GNNs scale well because they leverage the network’s interference pattern [23].”
BackgroundResultSelf-citationLearning Optimal Linear Precoding for Cell-Free Massive MIMO with GNN
Parlier, B., Salaün, L., & Yang, H. B. (2024) · Lecture notes in computer science · doi:10.1007/978-3-031-70378-2_5 · Cited by 5
Age of Information-Oriented Link Scheduling in Device-to-Device Networks
Wang, L., Wang, Q., Chen, H. H., & Zhou, S. (2025) · IEEE Transactions on Wireless Communications · doi:10.1109/twc.2025.3572690 · Cited by 5
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.