Architectures and algorithms for service oriented optical networks

一些新兴的应用和服务，特别是基于海量数据传输的应用和服务，需要传统网格基础设施无法提供的能力。后者通常将数据包路由网络用作非确定性资源。光网络技术的不断发展与动态预配置机制相结合，有望支持这些类型的高级应用和服务。然而，光网络提供的巨大带宽仍然没有得到有效利用，需要适当的杀手级应用。我的研究计划的全球目标是研究新的网络概念，并在面向服务的光网络背景下提出创新的解决方案。我将研究下一代光网络的新体系结构和协议，以支持网络和服务虚拟化。新兴的大规模应用需要已经被称为“确定性”的网络服务，这些服务意味着定义和保证的服务级别参数(例如，最小带宽、丢失率、延迟和抖动)。此外，这些服务需要跨多个域的异类网络进行配置。为了创建一个支持具有不同服务质量保证的端到端虚拟链路的虚拟化子网络的光网络，我将首先研究一种考虑流量动态的新型光控制平面。其目的是提供一种新的具有动态多粒度连接的智能控制平面策略。这些连接的路由和带宽分配将在以下约束条件下执行：Qos保证交互、计算资源和网络状态信息的可用性。我的新控制平面将在高级信令和低级信令之间动态切换，前者用于长期应用程序，后者用于更精细的应用程序粒度。我在本申请中提出的研究项目主要研究三个方面：1)低层信令，如OBS(光突发交换)信令；2)高层信令，如UCLP(用户控制光路)；3)面向服务的光网络中的多粒度资源优化，以支持网络虚拟化。

Several emerging new applications and services, especially the ones based on massive data transmission, require capabilities non-offered by traditional Grid infrastructure. The later usually exploits packet-routed networks as non-deterministic resources. The constant evolution of optical network technologies combined with the dynamic provisioning mechanisms holds the promise to support these types of advanced applications and services. Nevertheless, the huge bandwidth offered by optical networks is still inefficiently exploited and needs appropriate killer applications. The global objective of my research program is to investigate new networking concepts and to propose innovative solutions in the context of service-oriented optical networks. I will study new architecture and protocols for the next generation of optical networks to support network and service virtualization. The emerging large scale applications require what have been referred to as "deterministic" network services that imply defined and guaranteed service-level parameters (e.g., minimum bandwidth, loss rate, latency and jitter). In addition, those services need to be provisioned across multi-domain heterogeneous networks. In order to create an optical network supporting virtualized sub-networks with different QoS-guaranteed end-to-end virtual links, I will first investigate a novel optical control plane that takes into account traffic dynamics. The aim is to provide a new intelligent control plane strategy with dynamic multi-granularity connections. The routing and bandwidth allocation for those connections will be performed under constraints: QoS guarantees interaction, computing resources and network state information availability. My new control plane will dynamically switch between high-level signaling for long-lived applications and low-level approach for finer applications granularity. My research project proposed in this application investigate three main tracks: 1) low-level signaling such as OBS (Optical Burst Switching) signaling 2) high-level signaling such as UCLP (User Controlled LightPath) and 3) multi-granularity resources optimization in service oriented optical networks to support network virtualization.