无服务器环境下基于DQN和用户意图的函数放置研究  

Research on function placement based on DQN anduser intention in serverless environment

作　　者：徐东红[1,2] 花旭 Xu Donghong;Hua Xu(School of Computer Science&Technology,China University of Mining&Technology,Xuzhou Jiangsu 221116,China;Mine Digitization Engineering Research Center of Ministry of Education,Xuzhou Jiangsu 221116,China)

机构地区：[1]中国矿业大学计算机科学与技术学院,江苏徐州221116 [2]矿山数字化教育部工程研究中心,江苏徐州221116

出　　处：《计算机应用研究》2023年第12期3747-3753,3759,共8页Application Research of Computers

基　　金：国家自然科学基金资助项目(61702405);OPPO研究资金资助项目(H7D220060)。

摘　　要：随着无服务器计算技术的快速发展,无服务器计算以其按需付费、自动扩展等特性吸引了越来越多用户的关注,研究人员正在探索将无服务器架构应用于日渐完善的边缘计算系统。然而,在分布式的无服务器系统中如何有效地进行无服务器函数放置并为用户提供优质的服务仍是该领域一个亟需解决的问题。采用基于用户意图的DQN(deep Q-network)无服务器函数放置算法来解决分布式无服务器系统中的函数放置问题。该算法将用户意图作为学习目标,利用DQN算法进行函数放置的优化。仿真实验表明,基于用户意图的DQN无服务器函数放置算法在计算时延和放置开销方面显著降低。相较于传统方法,该算法能够有效满足用户的不同需求,并取得了较好的函数放置效果,在分布式无服务器系统中具有较高的适应性和性能优势。With the rapid development of serverless computing technology,serverless computing has attracted more and more users’attention with its pay-as-you-go,automatic extension and other features.Researchers are exploring serverless architectures for increasingly sophisticated edge computing systems.However,how to effectively place serverless functions in distributed serverless systems and provide users with high-quality services is still an urgent problem in this field.This paper applied user intent-based DQN serverless function placement algorithm to solve the function placement problem in distributed serverless systems.The algorithm took user intention as the learning objective and used DQN algorithm to optimize function placement.Simulation experiments show that,the user intent-based DQN serverless function placement algorithm achieves significant reduction in computing delay and placement cost.Compared with traditional methods,this algorithm can effectively meet the different needs of users and achieve better function placement effect,and has higher adaptability and performance advantages in distributed serverless systems.

关 键 词：云计算 无服务器计算 深度强化学习 用户意图 函数放置 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]