H2-LBM: A Hierarchical Hybrid Deep Reinforcement Learning Framework for L7 Load Balancing and Global Traffic Scheduling in Multi-Cloud LLM Serving

DEEP REINFORCEMENT LEARNING-BASED INTELLIGENT TASK SCHEDULING FRAMEWORK FOR CLOUD DISTRIBUTED SYSTEMS

This paper proposes an innovative Deep Reinforcement Learning-based Intelligent Task Scheduling Framework (DRITS) designed to optimize task allocation and resource utilization in cloud distributed systems and establishes DRL-based intelligent scheduling as a promising solution for next-generation cloud computing infrastructure management.

Deep reinforcement learning for cost-efficient resource management in hybrid cloud environments

Sensitivity and ablation studies confirm stable learning and controllable latency-cost trade-offs, demonstrating that lightweight RL can effectively deliver cost-efficient, adaptive autoscaling in hybrid cloud environments.

STQ-Scheduler: A Secure and Throughput-Aware Deep Reinforcement Learning Framework for QoE-Driven Resource Scheduling in Distributed Video Streaming Systems

. With the rapid growth of large-scale video streaming services, cloud gaming, and edge-assisted media delivery, ensuring high Quality of Experience (QoE) under dynamic network conditions and heterogeneous edge infrastructures has become a critical challenge. In practical systems, issues such as burst traffic, cross-edge latency variability, and noisy or unsafe data streams often lead to suboptimal resource utilization and degraded user experience. Meanwhile, conventional rule-based schedulers and static data processing pipelines are unable to jointly address the challenges of efficient model training and adaptive resource allocation. To tackle these problems, this paper proposes STQ-Scheduler, a secure and throughput-aware deep reinforcement learning framework that integrates high-throughput data processing, Transformer-based QoE prediction, and Proximal Policy Optimization (PPO)-based resource scheduling. The framework incorporates real-time data cleaning, anomaly filtering, and online feature transformation to ensure data quality and prevent data processing from becoming a bottleneck in distributed training. Furthermore, a multi-objective reward function is designed to jointly optimize QoE, latency, throughput, and system cost, enabling adaptive scheduling decisions across distributed edge nodes. Experimental results demonstrate that STQ-Scheduler significantly outperforms baseline methods in distributed video streaming environments. Specifically, it reduces average latency from 215 ms to 162 ms and improves throughput from 910 req/s to 1075 req/s under normal workloads. Under burst conditions, it maintains over 1020 req/s while reducing SLA violation rates from 10.5% to 3.9%, confirming its robustness and effectiveness in QoE-driven resource scheduling.

Hierarchical Meta-Graph Reinforcement Learning for Collaborative GenAI Model Caching and Inference Scheduling

Enabling collaborative generative AI (GenAI) inference at the network edge is challenging due to limited caching capacity, heterogeneous computing resources, and highly dynamic, latency-sensitive service demands. In this paper, we investigate the joint optimization of GenAI model caching, inference offloading, and resource allocation in a collaborative cloud–edge–end architecture. To address the strong coupling between long-term caching decisions and short-term scheduling dynamics, we propose a Hierarchical Meta-Graph Reinforcement Learning framework, termed HMGRL. Specifically, a heat-greedy model caching strategy is developed to capture time-varying model popularity and to reduce switching overhead on a slow timescale, while a graph-enhanced dueling deep reinforcement learning algorithm with prioritized experience replay enables topology-aware collaborative inference offloading and resource allocation on a fast timescale. Extensive simulations demonstrate that HMGRL consistently outperforms representative baselines in terms of system utility, cache and computing-resource utilization, convergence stability, and performance robustness. These results validate the effectiveness of the proposed hierarchical learning framework for practical GenAI applications at the network edge.

A Deep Reinforcement Learning Framework for Scalable Cloud Service Orchestration

Multi-Timescale Latent-Action DRL for Joint Optimization in Edge-Cloud Networks

A two-timescale multi-layer deep reinforcement learning framework with a latent action space (2T-MDRL-LA) to jointly optimize service placement, user association, computational delegation, task offloading, and user transmit power and achieves near-optimal performance compared to branch-and-bound solutions.