Simulation results under dynamic user mobility, stochastic task arrivals, and varying primary-user activity show that SCOPE improves latency, energy efficiency, service-level constraint satisfaction, and throughput compared with existing scheduling methods.
Abstract
Edge-assisted cognitive radio networks require efficient scheduling mechanisms to jointly manage opportunistic spectrum access, task offloading, energy consumption, and latency constraints. Existing multi-agent scheduling approaches often rely on fixed penalty terms or average queue-based constraints, which may not effectively control service-level violations under uncertain spectrum availability and dynamic edge-resource contention. This work proposes SCOPE, a Safe Causal-Graph Primal–Dual multi-agent scheduling framework for energy- and latency-constrained edge-assisted cognitive radio networks. The major strength of SCOPE is its integrated design, where belief-state augmentation improves decision-making under imperfect spectrum sensing, dual-relational causal graph coordination separately models’ interference coupling and computation-resource contention, and CVaR-based primal–dual optimization regulates tail-risk violations of latency and energy constraints. The framework follows a centralized-training and decentralized-execution structure, enabling coordinated learning during training while supporting scalable decentralized scheduling during deployment. Simulation results under dynamic user mobility, stochastic task arrivals, and varying primary-user activity show that SCOPE improves latency, energy efficiency, service-level constraint satisfaction, and throughput compared with existing scheduling methods. Ablation analysis further confirms the individual contribution of belief modeling, graph coordination, and risk-sensitive constraint enforcement.
Simulations show that HeLyMARL is the only method that sustains the throughput-fairness balance together with uninterrupted service throughout the horizon, outperforming conventional MARL, Lyapunov-based, and constrained MARL benchmarks without premature budget exhaustion.
Yeonseo Jeong, Wonhyeok Ko, Sungweon Hong et al.· 0 citations
Scheduling AI inference across heterogeneous edge-cloud chips requires balancing energy, latency, cost, and thermal feasibility. This study presents a chip-aware scheduling framework, rather than a new multi-objective reinforcement learning theory. A 53-dimensional state describes directed-acyclic-graph tasks, CPU/GPU/NPU/FPGA status, network conditions, and queue slack. A fuzzy controller adjusts energy, latency, and cost priorities, while a hybrid-action Proximal Policy Optimization policy jointly selects the offloading target, physical chip, and dynamic-voltage-and-frequency-scaling coefficient. Dependency, deadline, thermal, and bandwidth constraints are enforced through action masks and residual penalties. In traffic-video and industrial-inspection simulations, the framework achieved 5.8 TOPS/W, 132 ms average latency, a normalized cost coefficient of 0.17 per task, and 85.2% Pareto coverage. Under an equal 1000-episode and 30-seed budget, PPO reached 95% of its asymptotic improvement in 580 ± 45 episodes, obtained a final normalized return of -0.15 ± 0.03 and an HV of 0.87 ± 0.02, and produced no divergent seed. EdgeCloudSim/iFogSim2 replication and hardware-in-the-loop testing preserved the advantage over DTRL; in hardware-in-the-loop tests, the framework required 31.5 ± 3.5 J per task and 142 ± 8.0 ms, compared with 35.8 ± 4.0 J and 155 ± 9.5 ms for DTRL. The results indicate that task-aware objective adaptation and chip-level control can reduce wasteful offloading and thermal stress while preserving service timeliness.
Shijia Shao, Pei-Qing Ye, Biao Zhao et al.· Journal of King Saud Univers...· 0 citations
This study jointly optimizes task offloading and system resource scheduling to minimize the long-term delay–energy cost of NOMA-MEC systems using a master-refined multi-agent proximal policy optimization algorithm.
A graph-enhanced centralized critic is proposed that injects topology-aware relational embeddings into value estimation while leaving decentralized actors unchanged, with centralized-training overhead and dense 16-secondary-user settings as practical limitations.
Changjing Sun, Zhenhua Wang, Yangzhi Li et al.· IEEE Access· 0 citations
Cloud service platforms are increasingly extended to cloud-edge continua to support latency-sensitive and computation-intensive applications. In such distributed service environments, heterogeneous and time-varying compute capacity across edge sites creates strong competition among users, making it difficult to jointly achieve low delay and energy consumption, sustainable provider profit, and fair resource sharing. Although existing studies have investigated efficiency optimization, pricing mechanisms, and fairness-aware resource allocation, the joint coordination of adaptive pricing incentives and long-term fairness in dynamic multi-agent cloud-edge systems remains insufficiently explored. To address this issue, we develop a fairness-aware pricing and service-routing framework for multi-user multisite cloud-edge systems, and propose a heterogeneous multiagent learning method in which user agents learn service-routing decisions while service-node agents jointly adapt pricing and CPU-allocation policies under a fairness-aware utility design. The resulting coupled decision process is formulated as a Multi-Agent Markov Decision Process and implemented using a Multi-Agent Actor-Critic framework under centralized training and decentralized execution. Simulation results show that the proposed method reduces p95 delay and worst-user delay by up to 39.1% and 52.9%, respectively, while improving provider-side profit by up to 59.5% relative to the strongest competing baselines.
Yun Xia, Gang Zhou, Lirui Pan et al.· IEEE International Conferenc...· 0 citations
Multi-uncrewed aerial vehicle (UAV) cooperative mobile edge computing (MEC) systems present significant challenges owing to task causal dependencies, dynamic channel variations, and multi-dimensional resource coupling. In this study, a multi-UAV cooperative MEC system with task causality constraints and time-varying wireless channels is considered, and the joint optimization of task offloading, task migration, dynamic UAV clustering, and continuous UAV trajectory planning is investigated. The objective is to minimize the long-term weighted sum of system latency and energy consumption while ensuring task queue stability. Lyapunov optimization is first introduced to transform the formulated stochastic mixed-integer problem into a deterministic per-slot optimization. Afterward, a dual-timescale graph-enhanced multi-agent proximal policy optimization (DT-HGMAPPO) framework is proposed to coordinate long-timescale UAV clustering and trajectory planning with short-timescale task offloading and migration. Specifically, this framework decouples the problem by utilizing dual-layer weighted hypergraph matching (DL-WHM) for joint clustering and association, a dynamic priority scoring (DPS) mechanism for intra-cluster load balancing, and a graph-enhanced MAPPO algorithm for trajectory optimization. Simulation results reveal that the proposed DT-HGMAPPO algorithm outperforms conventional multi-agent deep reinforcement learning baselines in terms of both convergence speed and policy stability. It achieves a final reward that is at least 18.0% greater than that of other multi-agent algorithms. Moreover, the proposed framework reduces total system cost by 33.3% compared with MASAC and 18.9% compared with MATD3, thereby achieving a superior delay-energy trade-off while ensuring queue stability.
Jiaming Zhang, Hong Zhao, Lanhua Li et al.· IEEE Transactions on Cogniti...· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.
