业务驱动的频谱灵活光网络资源优化技术研究
发布时间：2019-01-20 18:42
【摘要】：随着大数据、云计算的快速发展,数据中心互联驱动的新一代光联网技术受到广泛关注。为满足大带宽、低时延、低能耗的业务传送要求,支持弹性带宽提供的频谱灵活光网络成为光通信领域的研究焦点之一。同时在数据中心互联驱动下,其未来发展呈现两大特点,即融合分组、电路(光通道)等多层网络资源的垂直融合趋势,以及融合计算／存储、IP/光通信等异构网络资源的水平融合趋势。如何实现业务驱动的频谱灵活光网络资源优化是一项关键研究课题。 本论文围绕光网络的频谱灵活需求和融合发展趋势,重点从多层流量疏导、频谱资源共享、网络虚拟化和智能控制等方面对频谱灵活光网络的资源优化技术进行研究,主要工作和创新成果如下： 第一,多层网络流量疏导是“垂直融合”下资源优化的关键技术。论文提出了针对不同弹性光收发机技术下的能耗最小化流量疏导ILP方法,设计了最大化光层疏导(MOLG)和最大化电层疏导(MELG)两种启发式算法。仿真结果表明,弹性光收发机的可切片特性能够有效地降低网络中耗能元件的数量,从而降低网络能耗。此外,针对可切片弹性光收发机技术下的动态流量疏导问题,论文提出了一个辅助图模型和两种频谱资源预留方法。通过设置辅助图中边的权重值来实现不同优化目标的流量疏导策略。论文分析了四种流量疏导策略在两种资源提供状态下的业务阻塞率、业务平均收发机个数、业务平均物理跳数和虚拟跳数等。 第二,针对“垂直融合”下带宽时变业务造成的频谱资源浪费问题,借助认知无线电的思想建立了时间相关的频谱资源共享模型(T-SRS)。该模型通过将带宽时变业务空闲出来的频谱资源动态地分配给其他用户使用,从而提升频谱资源的利用率。通过引入干扰因子来评价该模型的频谱共享效率,并设计了针对T-SRS模型的干扰因子最小化的ILP方法。同时为了验证模型的可扩展性,论文提出了最大保持时间优先的最短路径与最远距离频谱分配(MHF-SPFF)以及最大保持时间优先的平衡干扰因子路由与最远距离频谱分配(MHF-BIFF)两类启发式算法。仿真结果表明,MHF-BIFF比MHF-SPFF具有更高的资源利用效率。 第三,针对“水平融合”下的数据中心网络虚拟化问题,论文通过引入链路可切片能力、光收发机可切片能力、光交换节点可切片能力的概念,首次提出了在频谱灵活光网络中基于网络元素可切片的虚拟资源描述方法,并根据该方法设计了基于链路可切片(LS-based)以及基于交换节点可切片(NS-based)的两类虚拟网络映射策略。两种映射策略与最基本的虚拟网络映射策略(Baseline)进行对比。仿真结果表明,与Baseline相比,基于网络元素可切片的虚拟网络映射策略具有更低的阻塞率和更高成本收益,其中NS-based表现最佳。 第四,针对两种融合趋势下的频谱灵活光网络智能控制问题,论文提出了基于OpenFlow的频谱灵活光网络控制平面架构。设计了针对频谱灵活光网络的OpenFlow协议扩展。搭建了基于OpenFlow的频谱灵活光网络控制平台并进行了光路建立和带宽动态调整的实验。通过网络协议软件Wireshark分析了光路建立和调整的OpenFlow协议流程,统计了光路建立和调整时延,验证了基于OpenFlow集中式光网络控制架构的可行性。
[Abstract]:With the rapid development of large data and cloud computing, the new generation of optical networking technology driven by the data center is widely concerned. In order to meet the service transmission requirements of large bandwidth, low time delay and low energy consumption, the flexible optical network supported by the elastic bandwidth is one of the research focuses in the field of optical communication. At the same time, under the data center interconnection drive, its future development presents two big characteristics, namely, the vertical integration trend of the multi-layer network resources such as the fusion packet and the circuit (optical channel), and the horizontal integration trend of the heterogeneous network resources such as the fusion calculation/ storage, the IP/ optical communication and the like. How to realize the flexible optical network resource optimization of the business-driven spectrum is a key research topic. This paper focuses on the research, main work and innovation achievements of the spectrum flexible optical network from the aspects of multi-layer traffic channel, spectrum resource sharing, network virtualization and intelligent control. Next: First, multi-layer network traffic is the gateway of resource optimization under the 鈥渧ertical fusion鈥，
本文编号：2412298
[Abstract]:With the rapid development of large data and cloud computing, the new generation of optical networking technology driven by the data center is widely concerned. In order to meet the service transmission requirements of large bandwidth, low time delay and low energy consumption, the flexible optical network supported by the elastic bandwidth is one of the research focuses in the field of optical communication. At the same time, under the data center interconnection drive, its future development presents two big characteristics, namely, the vertical integration trend of the multi-layer network resources such as the fusion packet and the circuit (optical channel), and the horizontal integration trend of the heterogeneous network resources such as the fusion calculation/ storage, the IP/ optical communication and the like. How to realize the flexible optical network resource optimization of the business-driven spectrum is a key research topic. This paper focuses on the research, main work and innovation achievements of the spectrum flexible optical network from the aspects of multi-layer traffic channel, spectrum resource sharing, network virtualization and intelligent control. Next: First, multi-layer network traffic is the gateway of resource optimization under the 鈥渧ertical fusion鈥，
本文编号：2412298
本文链接：https://www.wllwen.com/kejilunwen/wltx/2412298.html
