SGER: Exploratory Research on Integrative Physical/Logical Layers Performance Scaling in Optical Packet Switching Systems

以数据为中心的业务量的爆炸性增长和对各种业务的需求正在推动光通信系统向分组交换网络的迁移。光分组交换（OPS）网络提供了独特的组合，包含了光域的巨大容量，同时提供了由单个分组路由提供的多功能连接。OPS网络在逻辑层的关键性能指标，包括吞吐量、延迟、可伸缩性和丢包率，已经在各种体系结构中得到了广泛的研究。然而，物理层性能是衡量包交换网络在光域实现可行性的关键指标，在这种情况下，人们并没有很好地理解。在OPS网络中，光包通常通过复杂的互连拓扑网络进行自路由。数据包可能采取的确切路径通常并不为人所知，因为路由机制可能会在统计上执行以实现公平性和负载平衡。此外，包内携带的数据结构可以是多种多样的，并且包括以不同比特率和调制格式编码的数据。因此，尽管逻辑拓扑可以根据网络层性能指标进行扩展，但并不一定意味着具有复杂光包传播的物理层也可以根据保持端到端信号完整性进行扩展。在这项探索性研究中，PI将致力于实现一种评估OPS网络中集成物理/逻辑层性能的方法，该方法真正涵盖了各种缩放指标集，并捕获了这些网络中光信号传播的复杂性。为了实现这些目标，一项实验研究将采用一个独特的、集成的系统，该系统由一个包含36个交换元件的完全连接的12端口OPS网络组成。该实验测试平台将实现数值和分析建模之间的直接耦合，并通过完整的实现OPS网络端到端传播实际数据。智力优势：提出的探索性研究将建立一种新的方法来弥合OPS通信系统中网络逻辑层和物理层之间的差距。这些活动将为整合目前两个独立学科的新兴研究领域奠定基础。逻辑和物理层的合并将使综合系统方法能够重新理解OPS网络的关键性能指标。与电子系统不同，光通信系统的物理层受到光场与传输介质、开关元件和放大器的相互作用（线性和非线性）引起的许多损伤。因此，整个复杂系统的性能不能解耦到其物理和逻辑组件层，必须以集成的方式进行研究。然而，执行这项调查的工具尚不存在。在这个提出的探索性研究计划中，PI将在一个完全实现的光分组交换网络元素中进行实际的流量路由实验。这些实验将强调物理层的可扩展性，因为每个超高容量端口的传入数据包可能包含跨越WDM c波段的有效载荷数据。整个有效载荷通过网络元件透明地端到端路由，因此可以包括沿多个有效载荷波长编码的各种调制速度和格式集。这些通过直接耦合物理层和逻辑层的实验研究，将能够为OPS网络的性能创建一个真正集成的系统模型。更广泛的影响：显然，光通信系统的迁移正朝着动态网络的方向发展，这是由分组数据中心流量的爆炸性增长所驱动的。OPS网络提供了利用光波通信的巨大容量的潜力，同时通过单个分组路由向众多流量目的地提供细粒度连接。透明的端到端有效负载路径支持的各种数据结构、编码和调制方案集创建了一个网络元素，可以与许多新一代网络无缝地发展。用于演示集成物理/逻辑层性能可扩展性的测试平台实验和建模将使未来超高容量分组交换系统的智能设计成为可能。

0532762BergmanThe explosive growth in data centric traffic and demand for diverse services are driving the migration ofoptical communication systems toward packet switched networks. Optical packet switched (OPS)networks offer the unique combination of encompassing the enormous capacity of the optical domainwhile providing versatile connectivity afforded by individual packet routing.The key performance metrics of OPS networks at the logical layer, including throughput, latency,scalability, and packet-loss-rates have been studied extensively for various architectures. However, thephysical layer performance, a critical measure of the feasibility of packet switched networks intended forimplementation in the optical domain, is not well understood in this context. In OPS networks opticalpackets are typically self-routed through a complex web of interconnection topologies. The exact path anypacket may take is not often known, as the routing mechanism may be performed statistically to achievefairness and load balancing. Furthermore, the data structure carried within the packet may be diverse andinclude data encoded in different bitrates and modulation formats.Thus, whereas it may be shown that the logical topology scales in terms of the network layer performancemetrics, it does not necessarily follow that the physical layer with complex optical packet propagationalso scales in terms of maintaining end-to-end signal integrity.In this proposed exploratory research the PI will aim to achieve a methodology for evaluating theintegrated physical/logical layers performance in OPS networks that truly encompasses the diverse setof scaling metrics and captures the complex nature of optical signal propagation in these networks. Toaccomplish these goals an experimental investigation will be performed employing a unique, integratedsystem of a fully connected 12-port OPS network containing 36 switching elements. This experimentaltest-bed will enable direct coupling between numerical and analytical modeling with realistic datapropagated end-to-end through a complete implemented OPS network.Intellectual Merit: The proposed exploratory research will establish a novel methodology for bridgingthe gap between the network logical and physical layers in OPS communication systems. These activitieswill create the groundwork for an emerging field of research that integrates two currently separatedisciplines. The merging of the logical and physical layers will enable an integrated systems approach to anew understanding of the critical performance metrics for OPS networks. Unlike electronic systems, thephysical layer of optical communication systems is subject to numerous impairments arising from theinteractions (linear and nonlinear) of the optical field with the transport medium, switching elements, andamplifiers. Thus, the total complex system performance cannot be decoupled into its physical and logicalcomponent layers and must be studied in an integrated fashion. The tools to perform this investigationhowever do not yet exist. In this proposed exploratory research program the PI will perform realistic trafficrouting experiments in a completely implemented optical packet switched network element. Theseexperiments will stress the physical layer scalability as incoming packets to each of the ultra-highcapacity ports may contain payload data that spans the WDM C-band. The entire payload is transparentlyrouted end-to-end through the network element and thus may include a diverse set of modulation speedsand formats encoded along the multiple payload wavelengths. These experimental investigations bydirectly coupling the physical and logical layers, will enable the creation of a truly integrative systemsmodel for the performance of OPS networks.Broader Impact: Clearly, the migration of optical communication systems is headed toward dynamicnetworks driven by the explosive growth of packetized data centric traffic. OPS networks offer thepotential of exploiting the enormous capacity of lightwave communications while delivering fine grainedconnectivity to a multitude of traffic destinations via individual packet routing. The diverse set of datastructures, encoding, and modulation schemes enabled by the transparent end-to-end payload path createsa network element that can potentially seamlessly evolve with many new network generations. Test-bedexperiments and modeling employed to demonstrate the integrated physical/logical layers performancescalability will enable intelligent design of future ultra-high capacity packet switched systems.