网络功能虚拟化中的性能感知服务组合优化

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。

Performance-aware service composition optimization in network function virtualization

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 invalid 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.