摘要
网络功能虚拟化环境下,为满足用户不同需求并提高资源利用效率,将虚拟网络功能映射与调度联合考虑。首先,通过构建网络低时延、虚拟机资源高利用及能量低损耗的多目标优化模型,设计了一种强化学习(RL)联合第三代非支配排序遗传算法(NSGAⅢ)的优化方法RL-NSGAⅢ;然后,采用两段式初始化技术求得高质量初始解,利用强化学习的优势自适应调节交叉变异参数,以保持种群多样性;最后,基于参考点的第三代非支配排序遗传算法,将虚拟网络功能映射至虚拟机并进行调度服务,得到多目标优化策略。仿真结果表明:相较于已有的NSGAⅢ、NSGAⅡ和MOPSO算法,采用RL-NSGAⅢ算法计算得到的时延降低了17%~28%,节点负荷提高了9%~19%,能量损耗降低了12%~26%,表明所提算法在提高网络速率和降低网络运营支出上的有效性。
In order to meet different needs of users and improve resource utilization efficiency in the network function virtualization environment,this paper examines virtual network function mapping and scheduling together.Firstly,an optimization method RL-NSGAⅢ based on reinforcement learning(RL)combined with the third-generation non-dominated sorting genetic algorithm(NSGAⅢ)is designed by constructing a multi-objective optimization model with low network delay,high utilization of virtual machine resources and low energy loss.Then,the two-stage initialization technique is used to obtain a high-quality initial solution,and RL’s advantage is used to adjust the crossover and variation parameters adaptively to maintain the diversity of the population.Finally,NSGAⅢ based on reference points is used to map virtual network functions for virtual machines and perform scheduling services to obtain a multi-objective optimization strategy.The simulation results show that compared with the existing methods NSGAⅢ,NSGAⅡ and MOPSO,the RL-NSGAⅢ method reduces latency by 17% to 28%,improves node load by 9% to 19%,and reduces energy loss by 12% to 26%,which verifies the effectiveness of the proposed method in improving network speed and reducing network operating expenses.
作者
刘光远
曹晶仪
杜婕
LIU Guangyuan;CAO Jingyi;DU Jie(School of Information Science and Technology,Shijiazhuang Tiedao University,Shijiazhuang 050043,China;Hebei Provincial Key Laboratory of Electromagnetic Environmental Effects and Information Processing,Shijiazhuang 050043,China)
出处
《西安交通大学学报》
EI
CAS
CSCD
北大核心
2024年第8期175-184,共10页
Journal of Xi'an Jiaotong University
基金
国家自然科学基金资助项目(62106157)
国家重点研发计划资助项目(2018YFB1701403)。
关键词
网络功能虚拟化
虚拟网络功能
映射与调度
强化学习
第三代非支配排序遗传算法
network function virtualization
virtual network function
mapping and scheduling
reinforcement learning
the third generation of non-dominated sorting genetic algorithm
