服务承载网构建算法研究

发布时间：2018-02-28 06:48

本文关键词： 可重构柔性网络服务承载网虚拟网络映射遗传算法粒子群优化算法网络中心性分析 K-核分解　出处：《解放军信息工程大学》2014年硕士论文　论文类型：学位论文

【摘要】：多年来,互联网应用上的创新层出不穷,大量个性化、小规模业务不断涌现,并逐步从小规模个性化应用向规模化应用发展。单一网络体系架构或者服务模式支撑多样化网络业务运营需求显得非常困难。为此,可重构柔性网络技术体系通过构建服务承载网(Service Carrying Network,SCN)的方式,同时支持多个异质网络并存,从而为用户提供多样化的网络服务。因此SCN的构建机制必然成为可重构柔性网络技术研究的关键问题之一。论文依托国家863项目“可重构路由构件组研制”、“可重构柔性试验网组网设备工程化实施”子课题。主要针对项目提出的三种聚类后的服务承载网特点及构建要求进行归纳总结。并分别针对三类服务承载网的构建算法展开了深入研究。具体研究工作如下:1.公共服务承载网具有常态性存在、规模大、管理简单、容易聚类的特点,针对上述特点本文提出了基于混合群智能优化的服务承载网构建算法。该算法利用群智能优化使映射节点方案在不断向最优方案接近的同时克服局部最优现象,通过带宽均衡分布的多商品流算法克服了节点间连通性问题。通过仿真表明,该算法具有较高的构建成功率,和算法D-ViNE-SP、D-ViNE-LB和VNE-R-PSO相比,底层网络收益开销比分别提高了55%、23%和6%,增加了可重构柔性网络的收益。2.个性服务承载网具有长期租用、规模可控、有特殊功能要求的特点。本文研究承载提供服务的分布式系统的个性服务承载网,在构建此类承载网时要求降低构建开销,增加外节点接入带宽。针对上述要求提出了基于网络中心性分析服务承载网构建算法。该算法量化了节点在全局内与其他节点的连通情况和路径被映射占用的可能性,为增加外节点接入带宽提供了参考。并依据最大最小匹配原则构建服务承载网。通过仿真表明,与算法VNE-ICC和算法VNE-Greedy相比,算法外节点的接入带宽增加了14%和32%,提高了分布式系统的服务质量。3.即时服务承载网具有建网时间和存在时间短、用户规模小等特点,针对上述特点本文提出了基于K-核分解的服务承载网构建算法。算法通过一个K-核分解类似过程构建中继传输子网,压缩了解空间。候选节点通过连接中继传输子网进而构建承载网。通过仿真表明,该算法服务承载网构建时间仅为算法VNE-R-PSO的50%,D-ViNE-LB的23%,D-ViNE-SP的27%,提高了服务承载网构建的即时性。
[Abstract]:Over the years, innovations in Internet applications have sprung up, a large number of personalization, and small-scale businesses have sprung up. And gradually from small-scale personalized applications to large-scale applications. A single network architecture or service model to support the diversified network business operation requirements is very difficult. The reconfigurable flexible network technology architecture supports the coexistence of multiple heterogeneous networks through the construction of Service Carrying Network (SCNs). Therefore, the construction mechanism of SCN must become one of the key problems in the research of reconfigurable flexible network technology. This paper relies on the national 863 project, "Research and development of reconfigurable routing component group", "reconfigurable routing component group", "reconfigurable routing component group" and "reconfigurable routing component group". Construction of flexible test network construction equipment engineering implementation "sub-project. Mainly in view of the project proposed three kinds of clustering service carrying network characteristics and construction requirements are summarized, and respectively for the construction of three types of service carrying network calculation." The concrete research work is as follows: 1.The public service carrying network has the normal existence. Large scale, simple management, easy clustering, In view of the above characteristics, this paper proposes a service carrying network construction algorithm based on hybrid swarm intelligence optimization, which uses swarm intelligence optimization to make the mapping node scheme approach to the optimal scheme and overcome the local optimal phenomenon. The problem of connectivity between nodes is overcome by a multi-commodity flow algorithm with balanced bandwidth distribution. The simulation results show that the algorithm has a high success rate, compared with D-ViNE-SPND-ViNE-LB and VNE-R-PSO. The ratio of income to expense of the bottom network increases by 55% and 6% respectively, and the income of the reconfigurable flexible network is increased. 2. The personalized service carrying network has long-term rental and controllable scale. This paper studies the individualized service carrying network of distributed system which provides service, and it is necessary to reduce the cost of building this kind of service network. In order to increase the access bandwidth of external nodes, an algorithm based on network centrality analysis service bearer network is proposed. The algorithm quantifies the connectivity between nodes and other nodes in the global context and the possibility that the path is occupied by mapping. It provides a reference for increasing the access bandwidth of external nodes, and constructs the service carrying network according to the principle of maximum and minimum matching. The simulation results show that, compared with the algorithm VNE-ICC and algorithm VNE-Greedy, The access bandwidth of the nodes outside the algorithm is increased by 14% and 32, and the quality of service of the distributed system is improved .3.Instant service carrying network has the characteristics of short time of network construction, short time of existence, small scale of users, etc. In view of the above characteristics, this paper proposes a service carrying network construction algorithm based on K- kernel decomposition, which constructs relay transmission subnet through a similar process of K- kernel decomposition. The candidate nodes construct the carrier network by connecting the relay transmission subnet. The simulation results show that the construction time of the service carrying network is only the 235D ViNE-LB of VNE-R-PSO, and the 27th of D-ViNE-SP, which improves the instantaneity of the construction of the service bearer network.
【学位授予单位】：解放军信息工程大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP393.09;TP18

【参考文献】