【作者】 刘强；

【导师】 唐良瑞； 赵兵；

【作者基本信息】 华北电力大学 ， 电子与通信工程（专业学位）， 2015， 硕士

【摘要】 通信业务的多元化、宽带化和IP化的发展对光传输技术和网络技术提出了更高的要求。作为光网络技术中至关重要的组成部分,业务疏导算法是影响网络接入性能和业务QoS性能的首要因素。因此设计适用于光网络,且能提升网络接入性能并充分保障业务QoS性能的业务疏导算法是一个十分有意义的研究课题。为解决区分业务模型的静态业务疏导问题,将其转化成多目标优化问题,以网络的建设成本、时延代价和风险性为优化目标建立疏导模型,提出区分业务模型的多目标优化疏导算法。算法将各业务的疏导方式作为决策变量,建立最优子种群和次优子种群,利用局部搜索和全局搜索相结合的方法寻找一组最优解。仿真实验证明,该算法能有效提高寻优的收敛速度,侧重于平衡三个优化目标的性能,以较小的网络建设成本为代价,大幅降低网络时延代价和网络风险性。为解决持续时间已知模型的动态业务疏导问题,以业务平均时延、业务阻塞率和业务量损失率为优化目标建立疏导模型,提出持续时间已知模型的高等级优先疏导算法。算法针对业务QOS性能要求的不同,采用不同的疏导策略,同时加入抢占机制。仿真实验证明,该算法能以部分低等级业务的阻塞为代价保障高等级业务的接入,在此基础上大幅提高了业务的实时性并减少业务量损失。更多还原

【Abstract】 Optical transmission technology and network technology are made demands as the multiple, broadband and IP-based development of communication services. Traffic grooming algorithm is the primary factor in the performance of access and QoS as a vital part of optical network technology. It is a meaningful project that designing a traffic grooming algorithm which is suitable for optical network and can promote the performance of access and QoS.To solve the static traffic grooming problems for differentiated services model, the problem is turn to multi-objective optimization problem, the grooming model is established based on the optimization objectives of the construction cost, price of delay and risk for network and the multi-objective optimization for differentially services grooming algorithm is proposed. Grooming ways for services are taken as decision variables, optimal subpopulations and sub-optimal subpopulation are set up, a group of optimal solutions is acquired through the combination of local search and global search. Simulation experiment shows that rate of convergence of optimizing is effective improved, the price of delay and the risk for network is reduced greatly at the expense of little network construction cost under the emphasis on the performance balance of the three optimization objective.To solve the dynamic traffic grooming problems for holding time aware services model, the grooming model is established based on the optimization objectives of the average delay, block rate of services and loss rate of traffic volume and the high-grade prior grooming for holding time aware services algorithm is proposed. Different grooming strategies are adopted for different service requirements, minimizing loss of services algorithm is introduced. The simulation improves that the block of low-grade services is sacrificed to guarantee the access of high-grade services, at the same time, the real-time performance is improved and the loss of traffic volume is decreased sharply.更多还原