EdgeSAC: Graph Neural Soft Actor-Critic for Hierarchical IoV Resource Management

Graph-Centric Deep Q-Learning for Interference-Aware Resource Allocation in Rsma-Enabled 5G Slicing

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.

Scene-Adaptive Task Offloading in Heterogeneous Edge Networks via Graph Neural Network-Enhanced Deep Reinforcement Learning

SAGE (Scene-Adaptive Graph-Enhanced offloading), a task-offloading framework that combines a heterogeneous graph neural network (HeteroGNN) with a dueling double DQN meta-controller and a mixed-integer linear programming (MILP) solver, is proposed, demonstrating strong scene-adaptive decision-making capability.

Optimizing Computation Offloading in IoV Environments: A Hybrid Learning Automata and Deep Recurrent Q-Network Approach

Graph-Enhanced Critic Learning for Cooperative Spectrum Access in Multi-UAV Cognitive Radio Networks

Cooperative spectrum access in multi-UAV cognitive radio networks requires decentralized control under dynamic primary-user activity, interference coupling, and partial observability. Standard multi-agent deep deterministic policy gradient (MADDPG) follows centralized training with decentralized execution, but its centralized critic usually processes the joint state-action context as a flat vector and therefore does not explicitly encode inter-UAV topology. This paper proposes a graph-enhanced centralized critic that injects topology-aware relational embeddings into value estimation while leaving decentralized actors unchanged. Under a unified eight-seed dynamic-spectrum protocol, the weighted radial basis function (weighted-RBF) graph critic improves mean reward from 4238.63 to 4346.55 and reduces collision rate from 0.1004 to 0.0383 relative to vanilla MADDPG. A stronger TD3-style multi-agent baseline (MATD3) reaches 4348.47 reward and 0.0495 collision rate, while MATD3 with the weighted-RBF graph critic further improves reward to 4379.29 and lowers collision rate to 0.0346. Additional topology ablation, high-interference, complexity, and scalability analyses show that graph-enhanced critic learning is most useful for collision control and interference-heavy operation, with centralized-training overhead and dense 16-secondary-user settings as practical limitations.

Conditional Information-Bottleneck Graph Clustering for Structured Representation Learning in Dynamic Vehicular ISAC Networks

Dynamic vehicular integrated sensing and communication (ISAC) requires representations that remain compact, decision-relevant, and structurally stable as mobility rewires interference and sensing relations. This paper presents IC-GMRO, a conditional information-bottleneck graph-clustering framework for structured representation learning in multi-agent resource optimization. At each control epoch, vehicles, roadside units, targets, and typed interactions form a temporal heterogeneous graph. A context-conditioned variational bottleneck suppresses nuisance variation while retaining action-relevant information; balanced soft graph clusters then convert the latent space into reusable coordination codes. Feasibility-masked policies jointly select association, beam, resource block, transmit power, and sensing-time ratio. The analysis distinguishes representation-level information guarantees from the idealized potential and projected-dual arguments used only to motivate the practical neural updates. Controlled simulations and component ablations show improved utility, sensing success, latency robustness, and cross-density robustness relative to greedy, flat, and graph-only baselines.

Domain-Guided Soft Actor–Critic for Network Slicing in Cell-Free Massive MIMO Systems

Cell-free massive multiple-input multiple-output (mMIMO), which eliminates cell edge effects and enhances coverage and resource utilization, is suited for industrial Internet of things (IIoT) applications. In user-centric cell-free mMIMO-based IIoT networks, joint optimization of network slicing and access point (AP) selection is crucial for meeting diverse quality-of-service (QoS) requirements. However, the joint optimization is challenging due to the coupling of resource allocation decisions and typically imperfect channel state information. In this paper, we formulate the joint AP selection and network slicing problem as a constrained Markov decision process (CMDP) with a hybrid action space, and propose a deep reinforcement learning (RL) algorithm, domain-guided hybrid soft actor-critic for CMDP (DG-HSA2C), to maximize the long-term proportional fairness in UE transmission rates while ensuring their QoS across slices. DG-HSA2C integrates CMDP-based RL into a hybrid action space by extending the Lagrangian multiplier method. To mitigate reward hacking, our algorithm incrementally predicts future states and incorporates a domain-adaptation mechanism, enhancing fairness in resource allocation and balancing performance across slices. Simulations verify our algorithm’s effectiveness in achieving rate fairness among UEs and mitigating reward hacking under the balance of QoS and rewards.