Stokes Vector Modulation for Terabit-Class Data Center Networks

适用于太比特级数据中心网络的斯托克斯矢量调制

• 批准号: 1609389
• 负责人: Mark Feuer
• 金额: $ 30.72万
• 依托单位: CUNY College of Staten Island
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2016
• 资助国家: 美国
• 起止时间: 2016-08-01 至 2022-09-30
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=1609389&HistoricalAwards=false
• 关键词: Stokes; Vector; Modulation; Terabit; Class

Data centers have become the nerve centers of the modern information-driven economy, relying on large networks of fiber-optic links operating at data rates up to 10 gigabits per second. Going forward, these fiber-optic links must scale to higher speeds at an acceptable cost in both dollars and in watts, but today's solutions, using on-off keyed signaling and power-sensing receivers, will not meet cost and dissipation targets as rates exceed 40 gigabits per second. Neither will fiber-optic transceivers developed for telecommunication networks, as these require complex receivers, expensive lasers in both transmitter and receiver, and power-hungry digital signal processing. To close the performance gap, this research program will investigate fiber-optic links in which the polarization of the transmitted light is switched to carry the data. Polarization shift keying and Stokes vector modulation (which switches both the polarization and amplitude) can transmit multiple bits of data per transmitted symbol, greatly accelerating the link's data rate while maintaining tolerance to fiber loss and other signal impairments. Research tasks will include theoretical study of noise impacts, optimization of advanced transmitter and receiver designs, simulation of polarization-modulated communication links, and experimental study in a lab testbed that incorporates dynamic optical networking. The timely and practical focus of this project will offer graduate and undergraduate students excellent opportunities to develop into productive practitioners in communications, fiber optics, and data center design. Beyond its direct impact on the student researchers, the program will offer additional educational benefits to the broader community by presenting lab tours and demonstrations to college and high-school students who might otherwise lack an opportunity to experience modern fiber-optic technology. Finally, by advancing the state of the art in data center technology, the project will contribute to economic growth as well as helping to keep the U.S. innovation pipeline well-filled. The primary technical objective of this research program is to advance the understanding of multi-dimensional modulation techniques based on polarization-shift keying and Stokes vector modulation, as applied to fiber-optic transceivers for terabit-class data center networks (i.e., photonic networks in which each wavelength channel carries about 1 terabit per second of data). Such research is critically needed because no existing technique has been found that offers both the data throughput and the cost needed to support Big Data applications in the 2020-2030 timeframe. Scaling of massive data centers to meet future demands will require a radical shift in optical data link technology. Current development based on pulse amplitude modulation will not be able to reach the 1 terabit per second level, because it is based on a one-dimensional symbol space. Coherent lightwave systems developed for long-distance telecommunications offer multi-dimensional symbol spaces, but they are too costly and power-hungry, and their slow setup times will inhibit the rapid dynamic networking needed for data center networks. Polarization-based modulation formats can provide 2-D, 3-D, or even 4 D symbol spaces using direct (i.e. non-coherent) detection, once critical challenges are overcome. This program will study fundamental issues of symbol constellations and noise mapping into Stokes space, as well as low-complexity digital signal processing and multiple-input, multiple-output techniques to enhance both receivers and transmitters. Alternative Stokes vector receiver designs will be compared theoretically and experimentally for both unimpaired link budgets and impairment tolerance. In support of dynamic topologies, performance under rapid optical switching/routing will be directly tested. At all stages, the work will be socialized through publications, presentations, and bilateral collaborations, while preparing students to enter the industrial and academic communities.

数据中心已成为现代信息驱动型经济的神经中枢，依赖于以高达每秒10千兆比特的数据速率运行的大型光纤链路网络。 展望未来，这些光纤链路必须以可接受的美元和瓦特成本扩展到更高的速度，但当今使用开关键控信令和功率传感接收器的解决方案将无法满足成本和功耗目标，因为速率超过40千兆比特每秒。 为电信网络开发的光纤收发器也不会，因为这些需要复杂的接收器，发射器和接收器中的昂贵激光器以及耗电的数字信号处理。 为了缩小性能差距，该研究计划将研究光纤链路，其中传输光的偏振被切换以携带数据。 偏振移位键控和斯托克斯矢量调制（切换偏振和幅度）可以在每个传输符号上传输多位数据，大大加快了链路的数据速率，同时保持对光纤损耗和其他信号损伤的容忍度。 研究任务将包括噪声影响的理论研究，先进的发射机和接收机设计的优化，偏振调制通信链路的模拟，以及在实验室测试平台，结合动态光网络的实验研究。 该项目的及时和实用的重点将为研究生和本科生提供极好的机会，发展成为通信，光纤和数据中心设计的生产实践者。 除了对学生研究人员的直接影响外，该计划还将通过向大学和高中学生提供实验室图尔斯参观和演示，为更广泛的社区提供额外的教育好处，否则他们可能没有机会体验现代光纤技术。 最后，通过推进数据中心技术的发展，该项目将有助于经济增长，并有助于保持美国创新管道的良好状态。 该研究计划的主要技术目标是促进对基于偏振移位键控和斯托克斯矢量调制的多维调制技术的理解，如应用于太比特级数据中心网络的光纤收发器（即，光子网络，其中每个波长信道承载大约每秒1太比特的数据）。 这种研究是非常必要的，因为没有发现现有的技术可以提供在2020-2030年时间框架内支持大数据应用所需的数据吞吐量和成本。 为了满足未来的需求，大规模数据中心的扩展将需要光学数据链路技术的根本转变。 目前基于脉冲幅度调制的发展将无法达到每秒1太比特的水平，因为它是基于一维符号空间的。 为长距离电信开发的相干光波系统提供多维符号空间，但它们过于昂贵和耗电，并且它们缓慢的建立时间将抑制数据中心网络所需的快速动态联网。 一旦克服了关键的挑战，基于偏振的调制格式可以使用直接（即，非相干）检测来提供2-D、3-D或甚至4D符号空间。 该计划将研究符号星座和斯托克斯空间噪声映射的基本问题，以及低复杂度数字信号处理和多输入多输出技术，以增强接收器和发射器。 替代斯托克斯矢量接收机的设计将进行比较，理论和实验两个未受损的链路预算和损伤容限。为了支持动态拓扑，将直接测试快速光交换/路由下的性能。 在所有阶段，工作将通过出版物，演示文稿和双边合作进行社会化，同时为学生进入工业和学术界做好准备。