Skyline-Enhanced Deep Reinforcement Learning Approach for Energy-Efficient and QoS-Guaranteed Multi-Cloud Service Composition

Cloud computing has experienced rapid growth in recent years and become a critical paradigm. Combining multiple cloud services to satisfy complex user requirements research hotspot computing. Service composition multi-cloud environments is characterized by high energy consumption, which brings attention the importance of consumption cross-cloud service composition. Nonetheless, prior mainly focused on finding that maximizes quality (QoS) overlooks generated during invocation. Additionally, dynamic nature challenges adaptability scalability methods. Therefore, we propose skyline-enhanced deep reinforcement learning approach (SkyDRL) address these challenges. Our defines an model for environments. The branch bound skyline algorithm leveraged reduce search space training time. enhance basic Q-network (DQN) incorporating double DQN overestimation problem, Dueling Network Prioritized Experience Replay speed up improve stability. We evaluate our proposed method using comparative experiments with existing results demonstrate effectively reduces while maintaining good problems. According experimental results, outperforms approaches demonstrating savings ranging from 8% 35%.