【作者】 赵永利；

【导师】 张杰；

【作者基本信息】 北京邮电大学 ， 电磁场与微波技术， 2010， 博士

【摘要】 随着互联网与物联网技术的飞速发展,光传送网的规模不断扩大、业务种类不断增多,未来的传送网面临着规模化和动态化需求,分层分域成为未来光网络的发展趋势。多层多域光网络在解决传送网可扩展性和灵活性的同时,也面临着优质化需求。在未来传送网规模化、动态化和优质化需求的驱动下,多层多域光网络仍存在着体系扩展性、业务复杂性、路由协同性和网络高效性等技术问题。以这些问题为出发点,本文在国家863目标导向类课题“可大规模扩展的多层多域智能光网络关键技术与实验系统”的支撑下,主要研究多层多域智能光网络的关键技术,特别是研究多层多域光网络中路由与资源优化问题,并取得了若干具有创新性的研究成果。本论文的主要工作和创新点可以总结为：提出了一种多层多域光网络新型路由体系结构,围绕此体系结构,在理论数学模型、路由优化算法和资源使用策略三个方面展开研究,并在多层多域光网络仿真平台上分别进行了性能验证。具体包括以下几个方面：第一,针对多层多域光网络存在的体系扩展性、业务复杂性、路由协同性以及网络高效性等方面的问题,提出了一种多层多域多约束GMPLS/ASON光网络的新型体系结构—DREAMSCAPE,并对DREAMSCAPE的协议与机制进行了研究设计,利用OMNET++仿真工具对基于DREAMSCAPE的双路由方案进行了仿真实验。实验结果表明,相对于分级路由方案和基于PCE的后向回溯路由方案,双路由方案在阻塞率和建路时延方面具有较大优势。第二,在理论数学模型方面,从局部的节点交换结构和全网的拓扑复杂性角度进行研究。建立了多粒度光网络的阻塞率分析模型,首次考虑了多粒度节点不具备波长变换能力和具备稀疏波长变换能力两种情况；然后研究了光网络的复杂性问题,发现并总结了光网络中涌现出的各种复杂性规律,在上述复杂性规律的基础上,设计了基于时空复杂性的光网络业务模型,较好地解决了在什么时候建立连接和与谁建立连接的问题。第三,在路由优化算法方面,针对BRPC算法在大规模网络中收敛时间过长的问题,提出了一种双向回溯的域问路径计算方法,在降低路径计算收敛时间的同时,还能够提高网络的可靠性；针对WSON光网络中路由与波长分配算法的复杂性需求,提出了基于PCE的解决方案,分解为三个子问题进行解决,即路由问题、资源预留问题和波长分配问题,具体包括一种基于PCE的路由架构、两种基于PCE的路由模式、两种基于PCE的资源预留方案和8种基于PCE的波长分配方案,并利用离散事件仿真工具OMNeT++进行了性能验证。第四,在资源使用策略方面,首次提出了一种资源预置的思想,并设计了相应的预置拓扑生成算法和基于预置拓扑的动态选路算法。仿真结果表明,通过资源预置策略,平均跳数和连接建立时间得到了显著降低；然后针对多层光网络中的能耗问题,基于不同的业务连接类型建立了动态多层光网络中的能耗分析模型,以此模型为基础设计了几种基于PCE的流量疏导算法和考虑生存性的流量疏导算法,并进行了仿真验证。更多还原

【Abstract】 With the rapid development of Internet and Internet of things, the scale of optical transport networks is getting larger and larger, and the service types are also getting more and more.Large scale and dynamic are the requirement of future optical networks, and multi-layer and multi-region are the inevitable evolution trends. Although the scalability and flexibility of optical transport networks can be solved by multi-layer and multi-region, there is still quality requirement. Under the driven of large scale, dynamic and quality requirement, there are still a series of problems in multi-layer and multi-region optical networks, such as architecture scalability, service complexity, routing cooperation and network efficiency. To solve these problems, this paper mainly researches on the key technologies in multi-layer and multi-region intelligent optical networks, especially the problem of routing and resource optimization, with the support of the object-oriented project of the national high-tech research and development program "Key technologies and experiment system in large scalable multi-layer and multi-region intelligent optical networks".A number of innovative research achievements have been gained.The main work and innovative contributions of this paper can be summarized as:a novel routing engine architecture of multi-layer and multi-region is proposed. Based on this architecture, mathematical models optimized routing algorithms and resource utilization strategies are studied and simulated on the simulation platform of multi-layer and multi-region optical networks.The details are listed as follows. Firstly, to resolve the problems of architecture scalability, service complexity, routing cooperation and network efficiency in multi-layer and multi-region optical networks, a novel routing engine architecture of multi-layer and multi-region multi-constraints optical networks (DREAMSCAPE) is proposed, which includes group engine and unit engine, and the protocols and mechanisms of DREAMSCAPE are designed. The performance of dual routing schemes based on DREAMSCAPE has been simulated and compared with hierarchical routing and backward recursive PCE-based computation.Numeric results show that DRE can reduce average time of path computation and setup without increasing blocking probability compared with BRPC.Secondly, in the respect of mathematical model, the switching architecture in the local node and the complexity of the entire network are studied.The blocking probability analytical models for MG-OXC based optical networks are built and both scenarios without and with sparse wavelength conversion capabilities are considered for the first time. Some complexity laws in optical networks are discovered and summarized, based on which, a novel traffic model for optical networks considering the time-space complexity is designed which satisfactorily solves the problems that when to create connections and whom to be connected.Thirdly, in the respect of optimized routing algorithms, a dual-end recursive PCE-based computation (DRPC) procedure is proposed to compute shortest constrained inter-domain traffic engineering label switched paths which can reduce the connection establishing time and improve the network reliability. For the complexity requirement of routing and wavelength assignment algorithms in WSON, a PCE-based solution is proposed, which consists of three sub problems which are routing, resource reservation, and wavelength assignment. The solution includes a PCE-based routing framework, two PCE-based routing models, two PCE-based resource reservation schemes, and eight PCE-based wavelength assignment schemes, the performances of which have been validated using discrete event simulation tool OMNeT++.Fourthly, in the respect of resource utilization strategies, the concept of resource pre-configuration is proposed for the first time.A planning scheme of resource pre-configuration that forms a pre-configured topology different from physical topology and virtual topology has been introduced for dynamic optical networks. Simulation results show the connection establishing time could be significantly reduced.In addition, for the energy-orient problems in multi-layer optical networks, a power consumption model is built based on different service LSPs. Several PCE-based and survivable traffic grooming algorithms are designed based on the model,then simulated and analyzed using discrete event simulation tool OMNeT++.更多还原