The proposed architecture provides a scalable foundation for adaptive cell selection in next-generation intelligent transportation systems and shows consistent improvements in association stability, handover reliability, and overall network performance compared with existing deep learning and graph-based methods.
Abstract
Efficient cell association remains a fundamental challenge in fifth-generation (5G) vehicle-to-everything (V2X) systems due to rapid topology changes, heterogeneous deployments, and stringent latency requirements. Conventional learning-based approaches often rely on shallow representations or independent optimization strategies, limiting their adaptability in dense and highly dynamic environments. To address these issues, this study introduces a multi-level graph representation framework that models interactions between vehicles and base stations across hierarchical spatial structures. The proposed approach integrates contextual node embedding with attention-driven graph learning to capture mobility patterns, signal characteristics, and network load dependencies. Additionally, a training-stage optimization mechanism is incorporated to refine attention parameters, improving convergence behavior without increasing inference complexity. The framework is evaluated using a real-world vehicular mobility dataset, demonstrating consistent improvements in association stability, handover reliability, and overall network performance compared with existing deep learning and graph-based methods. Experimental results show gains in accuracy (94.17%) and F1-score (93.93%), indicating enhanced decision robustness under dynamic conditions. Although validation is conducted on an urban dataset, the proposed architecture provides a scalable foundation for adaptive cell selection in next-generation intelligent transportation systems.
The emergence of 5G and 6G advanced ecosystems demands highly adaptive resource management to orchestrate the specialised requirements of eMBB, URLLC, and mMTC network slices. In dense multi-cell environments, capturing complex spatial interdependencies and mitigating dynamic interference is paramount for maintaining Quality of Service (QoS). This paper introduces a robust GNN-DQN framework designed for Rate Splitting Multiple Access (RSMA) based networks. By representing the network topology as a graph, the framework leverages Graph Neural Networks (GNNs) to extract highdimensional spatial features and model inter-cell interference patterns. These insights enable a Deep Q-Network (DQN) agent to perform intelligent resource partitioning and dynamic power splitting of the RSMA common stream. Experimental results demonstrate that the proposed GNN-DQN framework achieves a connectivity success ratio exceeding 90% across all slices, representing an average improvement of over 60% compared to non-graph-based reinforcement learning and supervised baselines. Notably, the framework demonstrates exceptional spectral efficiency, maintaining near-total connectivity while utilising less than 10% of the normalised system bandwidth, a 4× reduction in resource overhead compared to traditional methods. Furthermore, the GNN-driven architecture ensures stable convergence during training, yielding a 1.6× higher system reward score. Our findings validate GNN-DQN as a high-performance, scalable, and resource-efficient paradigm for intelligent orchestration in 5G and 6G networks.
Aya Kh. Ahmed, Nadia Al-Aboody, Hamed S. Al-Raweshidy· International Conference on...· 0 citations
An innovative combination of Multi‐Relational Skeleton Graph Attention Networks is proposed to cater to the challenge of understanding multiple relational behaviors in data traffic and is optimized with the Greater Cane Rat Optimization Algorithm to enhance learning efficiency and the precision of congestion control.
R. Mohanapriya, M. Gunavathie, S. Susmi et al.· International Journal of Com...· 0 citations
IC-GMRO is presented, a conditional information-bottleneck graph-clustering framework for structured representation learning in multi-agent resource optimization and distinguishes representation-level information guarantees from the idealized potential and projected-dual arguments used only to motivate the practical neural updates.
Routing optimization in cloud-edge collaborative networks faces a fundamental conflict between global strategic planning and local real-time responsiveness, further complicated by structural heterogeneity and stochastic traffic patterns. Traditional protocols lack adaptivity, while existing Deep Reinforcement Learning (DRL) approaches based on Graph Neural Networks (GNN) struggle with limited receptive fields and over-smoothing issues in large-scale topologies. In this paper, we propose HAT-Route, a Transformer-driven hierarchical routing framework supported by the Network Digital Twin (NDT). Our contributions are threefold: 1) We establish a cloud-edge collaborative architecture operating under the Centralized Training and Decentralized Execution paradigm. This architecture balances the trade-off between global optimization and real-time inference. 2) We introduce FlowFormer, a Spatiotemporal Transformer for the NDT. FlowFormer integrates a novel Edge-Conditioned Spatial Attention (EC-SAT) mechanism to capture physical link constraints and distinguish between congestion and Head-of-Line (HOL) blocking. 3) We design HAT-Route, a hierarchical DRL agent that utilizes Graph Transformers for global policy learning in the cloud, coupled with knowledge distillation to deploy lightweight policies at the network edge. Extensive experiments demonstrate that our framework outperforms traditional protocols and GNN-based baselines in terms of QoS optimization, training stability, scalability, and generalization capability on large-scale network topologies.
Bin Dai, Yuntao Wang, Jianhai Zheng· IEEE Transactions on Network...· 0 citations
Traffic forecasting is highly challenging due to complex and nonlinear spatial and temporal dependencies. Self-attention mechanisms have been widely adopted to model dynamic and long-range dependencies, achieving state-of-the-art performance, but suffer from limited scalability due to quadratic computational and memory complexity. To address this, we propose an Efficient Multi-Attention Graph Network (EMAGN) that linearises the spatial attention mechanism itself, inspired by the theory of fast high-dimensional Gaussian filtering. Two learned clustering matrices C_k and C_v adaptively group key and value vectors into M super-clusters, reducing complexity from O(N^2 d) to O(NMd) without sacrificing the flexibility of attention for dynamic dependency modelling. Experimental results on PEMS-BAY and METR-LA show that EMAGN achieves accuracy within 2.7-3.2% MAE of full-attention GMAN while reducing training time by 32%, inference time by 38%, and GPU memory by 58%. Critically, at K=16 attention heads, full-attention GMAN runs out of memory on a standard 11 GB GPU entirely while EMAGN continues to operate, demonstrating a categorical expansion of feasible model configurations. EMAGN also surpasses Linformer and Performer in both accuracy and efficiency within the same backbone, owing to its traffic-network-aware adaptive clustering.
Mingxing Xu, Rakesh Chowdary Machineni, Ke Liu et al.· 0 citations
Trajectory prediction serves as a pivotal component within the autonomous driving technology stack. However, predicting trajectories in heterogeneous traffic environments remains a formidable challenge, primarily due to the intricate interactions among agents and the significant distinctness in kinematic patterns across different object categories. To address these issues, we propose a novel trajectory prediction framework named Multi-graph Fusion and Future Interaction-Aware Network (MGFI). Specifically, we first construct a multi-graph spatio-temporal feature encoding module designed to comprehensively capture complex interaction information among traffic participants. This module employs four independent graph neural network branches on heterogeneous nodes. To synthesize the node features derived from these branches, we introduce a hierarchical feature fusion module that selectively aggregates features from the aforementioned multi-graph features. Finally, the fused node representations are fed into a sequence prediction module, which is further augmented by anticipated future interaction features to enhance prediction accuracy. Extensive experiments conducted on ApolloScape dataset demonstrate that the proposed MGFI significantly outperforms state-of-the-art trajectory prediction algorithms in terms of both prediction accuracy and robustness.
Changxu Zhu, Shibei Xue· International Conference on...· 0 citations