Two-Stage Heterogeneous Graph Neural Network for RIS-Aided Physical-Layer Security
Zihan Song, Yang Lu, Wei Chen, Bo Ai · 2026
Read Two-Stage Heterogeneous Graph Neural Network for RIS-Aided Physical-Layer Security by Zihan Song, Yang Lu, Wei Chen, Bo Ai (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2603-14655.txt.
Introduction
Song and co-authors secure RIS-aided wireless links with a two-stage graph network. A reconfigurable intelligent surface serves many users while eavesdroppers listen. Stage one draws the system as a bipartite graph of surface elements, users and eavesdroppers. Its output sets the surface phase shifts. Stage two uses a full graph to set beamforming and artificial noise. Both stages train without labels and scale with system size. The design beats rival graph methods and cuts run time a thousandfold versus convex solvers, they report.
Catalog record
| CREATOR | Zihan Song, Yang Lu, Wei Chen, Bo Ai |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 71106 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2603.14655
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Two-Stage Heterogeneous Graph Neural Network for RIS-Aided Physical-Layer Security Thanks: Zihan Song and Yang Lu are with the School of Computer and Information Technology, Beijing Jiaotong University, Beijing 100044, China (e-mail: [email protected], [email protected]). Thanks: Wei Chen and Bo Ai are with the School of Electronics and Information Engineering, Beijing Jiaotong University, Beijing 100044, China (e-mail: [email protected], [email protected]). Thanks: Zhiguo Ding is with the School of Electrical and Electronic Engineering (EEE), Nanyang Technological University, Singapore 639798 ([email protected]). Thanks: Arumugam Nallanathan is with the School of Electronic Engineering and Computer Science, Queen Mary University of London, London and also with the Department of Electronic Engineering, Kyung Hee University, Yongin-si, Gyeonggi-do 17104, South Korea (e-mail: [email protected]). Zihan Song, Yang Lu, Wei Chen, Bo Ai Abstract This paper investigates physical-layer security (PLS) enabled by graph neural networks (GNNs). We propose a two-stage heterogeneous GNN (HGNN) to maximize the secrecy energy efficiency (SEE) of a reconfigurable intelligent surface (RIS)-assisted multi-input-single-output (MISO) system that serves multiple legitimate users (LUs) and eavesdroppers (Eves). The first stage formulates the system as a bipartite graph involving three types of nodes—RIS reflecting elements, LUs, and Eves—with the goal of generating the RIS phase shift matrix. The second stage models the system as a fully connected graph with two types of nodes (LUs and Eves), aiming to produce beamforming and artificial noise (AN) vectors. Both stages adopt an HGNN integrated with a multi-head attention mechanism, and the second stage incorporates two output methods: beam-direct and model-based approaches. The two-stage HGNN is trained in an unsupervised manner and designed to scale with the number of RIS reflecting elements, LUs, and Eves. Numerical results demonstrate that the proposed two-stage HGNN outperforms state-of-the-art GNNs in RIS-aided PLS scenarios. Compared with convex optimization algorithms, it reduces the average running time by three orders of magnitude with a performance loss of less than 4 % 4\% . Additionally, the scalability of the two-stage HGNN is validated through extensive simulations. I Introduction I-A Background RIS has been widely regarded as a key wireless coverage enabler for next-generation mobile communication systems [ 31 ] . Specifically, RIS can intelligently adjust the phase of incident electromagnetic waves without additional radio frequency chains, thus effectively improving wireless channel quality, extending coverage, and mitigating signal interference [ 3 ] . Its passive operation enables energy-efficient deployment, which aligns well with the requirements of green communications. The channel reconfigurability of RIS can enlarge the channel diversity among receivers, which can be leveraged to enhance PLS [ 17 ] . As demonstrated in [ 9 ] , RIS-aided PLS can significantly improve the system SEE and secrecy sum rate (SSR) even when there are no direct communication links, and the performance gain increases with the number of reflecting elements. Despite the advantages of RIS, efficiently optimizing RIS-assisted communication systems is non-trivial due to the deeply coupled BS-related and RIS-related parameters. Existing optimization approaches for RIS can be categorized into three categories: 1) convex optimization (CVXopt)-based approach, 2) deep reinforcement learning (DRL)-based approach, and 3) learn-to-optimize approach. For example, the authors in [ 2 ] , [ 20 ] and [ 5 ] investigated multi-user MISO systems assisted by RIS, employing CVXopt-based approaches to maximize the system SEE and sum rate, respectively. However, as reviewed in [ 23 ] , most existing CVXopt-based methods mainly rely on alternating optimization (AO) or block coordinate descent (BCD) to tackle the deeply coupled variables, leading to high computational overhead and separate optimization losses. In contrast, DRL-based approaches have emerged as a promising joint-optimization solver for RIS-assisted systems. They have been validated to enable energy-efficient designs in RIS-assisted MISO [ 10 ] , simultaneous wireless information and power transfer [ 38 ] , and unmanned aerial vehicle mobile edge computing systems [ 1 ] . Nevertheless, DRL-based approaches learn from interactions in partially controllable wireless environments, thus suffering from limited generalization capability. This paper focuses on the data-driven learn-to-optimize approach, which has been widely adopted as an efficient optimization solver for various wireless systems [ 27 ] . Specifically, existing studies have adopted multi-layer perceptrons (MLPs) [ 24 , 4 ] and convolutional neural network (CNNs) [ 21 , 35 ] , respectively, to realize joint optimization for RIS-assisted systems. However, MLPs and CNNs are inherently limited by their fixed input and output dimensions, thus failing to provide satisfactory generalization performance in dynamic RIS-assisted systems. By contrast, GNNs can effectively overcome the limitation [ 19 , 18 ] . As analyzed in [ 25 ] , GNNs are well matched to the topology of wireless networks, thereby not only outperforming other models but also achieving strong adaptability to wireless dynamics. For example, GNNs have been shown to enable real-time, near-optimal and scalable inference for MISO systems, aiming to maximize the sum rate (SR) [ 13 ] , energy efficiency (EE) [ 7 ] and SEE [ 26 ] , respectively. Compared with non-RIS and non-PLS systems, RIS-aided PLS involves complicated heterogeneity among distinct entities, such as the BS, RIS, LUs, and Eves. Therefore, it is of great importance to align the deep learning (DL) model with both the system and the task. Existing works handle the “heterogeneity” via two perspectives, i.e., framework and neural architecture. For the framework, the authors in [ 32 , 8 ] developed a multi-stage GNN to sequentially generate the phase shift matrix and beamforming vectors. This framework inherits the algorithmic structure of conventional CVXopt-based approaches and enhances the model’s interpretability. For the neural architecture, the authors in [ 33 , 28 , 34 , 16 ] exploited HGNNs to capture the heterogeneity within RIS-assisted systems. These HGNNs exhibit superior scalability with respect to the system entities. I-B Contribution Existing works have demonstrated that the model’s expressive capability can be enhanced from the perspectives of both framework and neural architecture. However, no existing work has yet integrated the advantages of these two aspects. To fill this gap, this paper proposes to utilize the two-stage framework and HGNNs for RIS-aided physical-layer security, which involves heterogeneous system entities (i.e., BS, RIS, LUs, and Eves) as well as dual-functional services. The contributions of this paper are summarized as follows. • We consider a typical RIS-aided PLS system, in which a multi-antenna BS serves multiple LUs in the presence of Eves via both direct and cascaded links. The transmitted signals consist of both information-bearing signals for LUs and AN to disrupt Eves. We formulate an SEE maximization problem by jointly optimizing the beamforming and AN vectors, as well as the RIS phase shift matrix. • We propose a two-stage HGNN to solve the formulated problem. Specifically, the first stage models the considered system as a bipartite graph and employs both edge-based and edge-free graph attention operators to map channel state information (CSI) to the phase shift matrix. The second stage models the system as a fully-connected graph with heterogeneous node types and adopts an edge-free graph attention operator and a semantic attention mechanism to map augmented effective CSI to the beamforming and AN vectors. The outputs of both stages are fed into an unsupervised loss function for end-to-end joint training. • Analysis results are presented to validate that the proposed two-stage HGNN is scalable with respect to the