网络功能虚拟化中的性能感知服务组合优化  

Performance-Aware Service Composition Optimization in Network Function Virtualization

作　　者：张晓红[1] 黄继海[1] 

机构地区：[1]中州大学信息工程学院,郑州450005

出　　处：《电讯技术》2015年第8期919-925,共7页Telecommunication Engineering

摘　　要：服务功能链是网络功能虚拟化的重要支撑,为了构建满足功能和性能需求的服务功能链,需要建立服务的性能模型,从而产生基于性能的服务组合优化问题。一种基于性能模型的服务组合优化问题被建模,并针对复杂约束情况下的无效解干扰,提出了改进的模拟退火算法,该算法包含基于层次属性的产生函数和基于偏离度的目标函数。仿真结果表明,该算法提高了21%的服务组合成功率,同时降低了组合成本和时间消耗。仿真结果验证了所提算法的有效性。Service function chain is a great support to Network Function Virtualization （NFV）. To construct a service function chain satisfying function and performance demands, a performance model is required for the service entities, and then the performance-aware service composition optimization problem（SCOP） is coming. A performance model is proposed to analyze the SCOP, and an improved Service Composition Mode search algorithm based on simulated annealing（SCM-SA） is designed to lighten the impact of inva- lid solutions which do not satisfy the complex constraints. The SCM-SA algorithm contains a generation function based on level attribute and a cost function based on offset degree. Simulation results show that the SCM-SA algorithm increases the service composition success rate（CSR） by 21% while decreasing the composition cost and time cost. Simulation results confirm the validity of the proposed algorithm.

关 键 词：网络功能虚拟化 服务功能链 服务组合 性能模型 模拟退火 

分 类 号：TN915.8[电子电信—通信与信息系统]