Efficient Power Allocation Using Graph Neural Networks and Deep Algorithm Unfolding

Chowdhury, A. R.; Verma, Gunjan; Rao, Chirag; Swami, Ananthram; Segarra, Santiago

doi:10.1109/icassp39728.2021.9415106

Cited by 17 publications

(6 citation statements)

References 22 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Substantial computational efficiency for the scenario with fixed numbers of UEs, macro BSs and small-cells, in comparison with the alternatives, has been achieved. A novel GNN architecture via algorithm unfolding for power allocation in an ad hoc network is proposed in [ 32 ]. The objective function of the throughput is decomposed into several parameters which process the vertices’ information iteratively, and are dependent on the channel gain (which is the vertex attribute).…”

Section: Graph-based Ra In Cellular Homogeneous and Het-netsmentioning

confidence: 99%

Graph-Based Resource Allocation for Integrated Space and Terrestrial Communications

Ivanov

Tonchev

Poulkov

et al. 2022

Sensors

View full text Add to dashboard Cite

Resource allocation (RA) has always had a prominent place in wireless communications research due to its significance for network throughput maximization, and its inherent complexity. Concurrently, graph-based solutions for RA have also grown in importance, providing opportunities for higher throughput and efficiency due to their representational capabilities, as well as challenges for realizing scalable algorithms. This article presents a comprehensive review and analysis of graph-based RA methods in three major wireless network types: cellular homogeneous and heterogeneous, device-to-device, and cognitive radio networks. The main design characteristics, as well as directions for future research, are provided for each of these categories. On the basis of this review, the concept of Graph-based Resource allocation for Integrated Space and Terrestrial communications (GRIST) is proposed. It describes the inter-connectivity and coexistence of various terrestrial and non-terrestrial networks via a hypergraph and its attributes. In addition, the implementation challenges of GRIST are explained in detail. Finally, to complement GRIST, a scheme for determining the appropriate balance between different design considerations is introduced. It is described via a simplified complete graph-based design process for resource management algorithms.

show abstract

Section: Graph-based Ra In Cellular Homogeneous and Het-netsmentioning

confidence: 99%

Graph-Based Resource Allocation for Integrated Space and Terrestrial Communications

Ivanov

Tonchev

Poulkov

et al. 2022

Sensors

View full text Add to dashboard Cite

show abstract

“…The previous methods are all based on data-driven neural networks, which have poor interpretability and scalability. Different from the previous works, A. Chowdhury et al proposed a hybrid data-model driven neural architecture inspired by the algorithmic unfolding of the iterative WMMSE, i.e., unfolded WMMSE (UWMMSE), to solve the optimal power allocation problem in a single-hop ad-hoc wireless network [53]. The optimization problem should be solved is…”

Section: A Power Controlmentioning

confidence: 99%

An Overview on the Application of Graph Neural Networks in Wireless Networks

He¹,

Xiong²,

Ou³

et al. 2021

Preprint

View full text Add to dashboard Cite

With the rapid enhancement of computer computing power, deep learning methods, e.g., convolution neural networks, recurrent neural networks, etc., have been applied in wireless network widely and achieved impressive performance. In recent years, in order to mine the topology information of graphstructured data in wireless network as well as contextual information, graph neural networks have been introduced and have achieved the state-of-the-art performance of a series of wireless network problems. In this review, we first simply introduce the progress of several classical paradigms, such as graph convolutional neural networks, graph attention networks, graph autoencoder, graph recurrent networks, graph reinforcement learning and spatial-temporal graph neural networks, of graph neural networks comprehensively. Then, several applications of graph neural networks in wireless networks such as power control, link scheduling, channel control, wireless traffic prediction, vehicular communication, point cloud, etc., are discussed in detail. Finally, some research trends about the applications of graph neural networks in wireless networks are discussed.

show abstract

“…To handle this problem, multiple GNN-based solutions are proposed [77,52,34,51,36,99,76,26,78]. In a series of studies [77,52,34,99], Random Edge Graph Neural Networks (REGNNs) are selected as the optimal solution for the power allocation and control optimization problem, with various system constraints.…”

Section: General Wireless Networkmentioning

confidence: 99%

“…REGNNs outperform baselines with an essential permutation invariance property, which are desirable in networks of growing size. For the optimal power allocation in a single-hop ad hoc wireless network, an iterative weighted minimum mean squared error method named UWMMSE is proposed, in which GNNs are used to learn the model parameters [51,36]. UWMMSE effectively reduces the computational complexity without harming the allocation performance, over the classic algorithm for power control.…”

Section: General Wireless Networkmentioning

confidence: 99%

“…Binary Classification [98] Decentralized GNN GCN [88], GIN [92] Channel Allocation [16] DRL with GCN ChebNet [87] Network Flow Optimization [97] Two-stage Topologyaware ML Framework MPNN [90] Power Allocation [77,52,34] REGNN GNN [86] Power Allocation [51,36] UWMMSE Method GCN [88] Power Control [99] REGNN GNN [86] Power Control [76] IGCNet GIN [92] Power Control and Beamforming [26] MPGNNs GIN [92], GCN [88] Power Control [78] Unsupervised…”

Section: Problem Paper Solution Gnnmentioning

confidence: 99%

See 1 more Smart Citation

Graph-based Deep Learning for Communication Networks: A Survey

Jiang

2021

Preprint

View full text Add to dashboard Cite

Communication networks are important infrastructures in contemporary society. There are still many challenges that are not fully solved and new solutions are proposed continuously in this active research area. In recent years, to model the network topology, graph-based deep learning has achieved state-of-the-art performance in a series of problems in communication networks. In this survey, we review the rapidly growing body of research using different graph-based deep learning models, e.g. graph convolutional and graph attention networks, in various problems from different communication networks, e.g. wireless networks, wired networks, and software-defined networks. We also present a well-organized list of the problem and solution for each study and identify future research directions. To the best of our knowledge, this paper is the first survey that focuses on the application of graph-based deep learning methods in communication networks. To track the follow-up research, a public GitHub repository is created, where the relevant papers will be updated continuously.

show abstract

Efficient Power Allocation Using Graph Neural Networks and Deep Algorithm Unfolding

Cited by 17 publications

References 22 publications

Graph-Based Resource Allocation for Integrated Space and Terrestrial Communications

Graph-Based Resource Allocation for Integrated Space and Terrestrial Communications

An Overview on the Application of Graph Neural Networks in Wireless Networks

Graph-based Deep Learning for Communication Networks: A Survey

Contact Info

Product

Resources

About