智能光网络的路由与资源分配及调度研究

发布时间：2018-05-26 19:20

本文选题：智能光网络 + 光传送网　；参考：《北京邮电大学》2014年博士论文

【摘要】：随着IP宽带业务、移动互联网和物联网的飞速发展,带宽需求迅猛增长,超大容量、动态灵活、高可扩展、高谱效率等成为下一代光传送网技术应具备的主要特征。业务的丰富和多样化,要求灵活高效的网络资源配置,以及弹性合理的网络控制和管理,对下一代光传送网的资源分配、控制管理与优化技术提出了挑战。传统波长交换光网络资源分配粒度较大,带宽固定,网络控制和管理可扩展性不高,运营成本较高。在智能光网络中引入带宽可变的光传输技术和光交换技术,能够灵活高效地利用频谱资源。智能光网络还需要与基于通用多标签协议标签交换(Generalized Multiprotocol Label Switching, GMPLS)架构的控制技术相结合,对网络资源进行合理管理和调度,应用路径计算单元(Path Computation Element, PCE)进行多约束条件下的路径计算,有效降低网络阻塞率,提高资源利用率。为了应对智能光传送网在演进发展中的上述问题,本文从动态灵活的资源分配和高效可扩展的网络控制管理两方面入于,研究了可变带宽光交换的选路与频谱资源分配问题、拥塞避免的路径计算问题、成本优化的业务调度问题和网络可扩展性的评估问题,主要创新工作如下：第一,针对智能光网络在可变带宽光交换下复杂的动态路径计算及频谱分配问题,特别是调制方式选择问题,将链路物理损伤信息与调制级别自适应变换(Modulation Level Conversion, MLC)相结合,提出了路由、测制方式和频谱分配的一体化算法(Link-MLC等)。仿真结果表明该算法能够改善多粒度场景下的网络阻塞率并提高资源利用率第二,针对智能光网络在大规模情况下流量分配不均的问题,设计了两种可避免拥塞(Congestion Avoidance, CA)的动态路由算法。仿真结果表明,这两种算法有利于减少流量分配不均现象,进而实现流量均衡。第三,针对智能光网络在多域情况下的可扩展性评估问题,指出路由协议泛洪是影响多域可扩展性的最主要因素,设计了收敛时间的测试方案。通过实验验证了其有效性,并将此测试方案应用在多层多域光网络实验平台上。第四,为了降低多层智能光网络的运营成本,通过分析多层光网络模型以及调度业务的特点,提出了两种调度业务算法。结果表明,设备重用(Transponder/Regenerator Reuse, TRR)算法降低运营成本效果较好；使用同一种业务调度算法时,贪心算法比禁忌搜索算法运行时间短,但禁忌搜索算法能得到更接近最优成本的解。
[Abstract]:With the rapid development of IP broadband services, mobile Internet and Internet of things, the demand for bandwidth is growing rapidly, the large capacity, dynamic flexibility, high scalability and high spectral efficiency become the main characteristics of the next generation optical transport network technology. The rich and diversified services require flexible and efficient network resource allocation and flexible and reasonable network control and management, which challenge the resource allocation, control management and optimization technology of the next generation optical transport network. Traditional wavelength switched optical networks have large granularity of resource allocation, fixed bandwidth, low scalability of network control and management, and high operating cost. The introduction of optical transmission technology and optical switching technology with variable bandwidth in intelligent optical networks can make use of spectrum resources flexibly and efficiently. Intelligent optical network also needs to combine with the control technology based on generic multi-label protocol label switching generalized Multiprotocol Label Switching, GMPLS) architecture to manage and schedule the network resources reasonably. Path computing unit (path Computation Element, PCE) is applied to path computation under multi-constraint conditions, which can effectively reduce network blocking rate and improve resource utilization. In order to deal with the above problems in the evolution of intelligent optical transport networks, this paper studies the routing and spectrum resource allocation of optical switching with variable bandwidth from two aspects: dynamic and flexible resource allocation and efficient and scalable network control management. Congestion avoidance path computing problem, cost optimization service scheduling problem and network scalability evaluation problem, the main innovation work is as follows: First, aiming at the complex dynamic path calculation and spectrum allocation problem of intelligent optical network under variable bandwidth optical switching, especially the problem of modulation mode selection, the link physical damage information is combined with modulation level adaptive modulation Level Conversion, MLC). An integrated algorithm for routing, measurement and spectrum allocation is proposed, such as Link-MLC and so on. Simulation results show that the proposed algorithm can improve the network blocking rate and resource utilization in multi-granularity scenarios. Secondly, aiming at the problem of uneven traffic distribution in large scale in intelligent optical networks, two dynamic routing algorithms are designed to avoid congestion contention (CAA). The simulation results show that these two algorithms can reduce the uneven distribution of traffic and realize traffic balance. Thirdly, aiming at the scalability evaluation of intelligent optical networks in multi-domain, it is pointed out that routing protocol flooding is the most important factor affecting multi-domain scalability, and a test scheme of convergence time is designed. The experimental results show that the proposed scheme is effective and applied to the multi-layer and multi-domain optical network experiment platform. Fourthly, in order to reduce the operation cost of multi-layer intelligent optical network, two scheduling algorithms are proposed by analyzing the model of multi-layer optical network and the characteristics of scheduling services. The results show that the transponder / Regenerator Reuse( TRR) algorithm is more effective in reducing the operation cost, and the greedy algorithm has a shorter running time than the Tabu search algorithm, but the Tabu search algorithm can get a solution close to the optimal cost.
【学位授予单位】：北京邮电大学
【学位级别】：博士
【学位授予年份】：2014
【分类号】：TN929.1

【相似文献】