ACED Fab: Co-Design of Novel Electronic-Photonic Systems for Energy-Efficient Coherent Optical Interconnects

ACED Fab：用于节能相干光互连的新型电子-光子系统的协同设计

• 批准号: 2314868
• 负责人: Samuel Palermo
• 金额: $ 40万
• 依托单位: Texas A&M Engineering Experiment Station
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2023
• 资助国家: 美国
• 起止时间: 2023-07-01 至 2026-06-30
• 项目状态: 未结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=2314868&HistoricalAwards=false
• 关键词: ACED; Fab; Co; Design; Novel

Dramatic improvements in datacenters and high-performance computing systems’ interconnect bandwidth-density and energy-efficiency are necessary to support advances in machine learning, artificial intelligence, sensor systems, and 5G/6G workloads. However, there are fundamental limitations to scaling data rates in conventional intensity-modulated direct detection (IMDD) optical links due to the extreme baud rates, i.e. changes in signals per second. Coherent optical interconnects offer a potential solution, as they modulate both the amplitude and phase of the optical carrier and utilize dual polarization (DP) operation to allow for a dramatic increase in bandwidth-density per wavelength. While coherent optical links are spectrally-efficient, key challenges include limited silicon photonic modulator bandwidth, high-power transceivers due to independent design of the photonic devices and front-end circuitry, sensitivity to photonic device fabrication variations, and high-power receiver-side optical carrier recovery that is commonly performed in a complex digital signal processor (DSP) block. This proposal addresses these important issues by co-designing high-bandwidth photonic devices and advanced-node CMOS front-ends that can adapt to variations in optical device performance and by utilizing a power-efficient receive-side carrier recovery scheme based on a dual-loop optical phased-locked loop (OPLL). The proposed technology will enable energy efficient coherent optical transceivers that will allow dramatic scaling in datacenter traffic capacity to support the unprecedented growth in networked devices driven by emerging applications such as connected automobiles, for example. This proposal’s research goal is to develop a coherent optical interconnect architecture with novel high-bandwidth quadrature modulators with thin-film LiNbO3 (TF-LN) Mach-Zehnder modulators (MZMs) and quadrature demodulators with graphene photodetectors. Co-design of energy-efficient CMOS transmitters with dynamic voltage frequency scaling (DVFS) with efficient switching regulators and energy-efficient CMOS receivers with DVFS, adaptive bandwidth front-ends, and auto-tuned quadrature demodulators will be fabricated to accomplish this goal. In addition, an optical phase-locked loop (OPLL) based carrier recovery scheme with wide-range electronic voltage-controlled oscillator (VCO) tuning will be developed. Applying the proposed technology will revolutionize the future of both datacenter and high-performance computing systems due to its ability to offer low-latency interconnects without error coding. This project will involve an interdisciplinary team of 2 Texas A&M University (TAMU) students and 3-4 National Chung Hsing University (NCHU) students. Two sets of prototypes will be implemented using two advanced CMOS processes, a silicon photonic process, and custom-fabricated thin-film TF-LN integrated circuits. Project outreach activities include exchange and visiting activities where TAMU and NCHU students work together face-to-face on-site during critical IC design phases and also participate in joint workshops, interactions with high school teachers via the Enrichment Experiences in Engineering (E3) program and introducing basic research concepts to PK-12 students through the Spark! Program. Project results will be broadly disseminated by inclusion in the syllabus and website of a new graduate course entitled "Coherent Optical Systems", the development of online modules for academia and industry, and through publication in national and international journals and conferences.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.

为了支持机器学习、人工智能、传感器系统和5G/6 G工作负载的进步，有必要大幅改善网络中心和高性能计算系统的互连带宽密度和能效。然而，由于极端的波特率，即每秒信号的变化，在传统的强度调制直接检测（IMDD）光链路中缩放数据速率存在基本的限制。相干光互连提供了一种潜在的解决方案，因为它们调制光载波的振幅和相位，并利用双偏振（DP）操作来允许每波长带宽密度的急剧增加。虽然相干光链路具有频谱效率，但关键挑战包括有限的硅光子调制器带宽、由于光子器件和前端电路的独立设计而产生的高功率收发器、对光子器件制造变化的敏感性以及通常在复杂数字信号处理器（DSP）块中执行的高功率接收器侧光载波恢复。该提案通过共同设计高带宽光子器件和先进节点CMOS前端来解决这些重要问题，这些前端可以适应光器件性能的变化，并通过利用基于双环光锁相环（OPLL）的功率高效的接收侧载波恢复方案。拟议的技术将实现节能相干光收发器，从而允许数据中心流量容量的大幅扩展，以支持由新兴应用（例如联网汽车）驱动的联网设备的前所未有的增长。该提案的研究目标是开发一种相干光互连架构，该架构具有新型高带宽正交调制器（具有薄膜LiNbO 3（TF-LN）Mach-Zehnder调制器（MZM））和正交解调器（具有石墨烯光电探测器）。协同设计具有动态电压频率缩放（DVFS）功能的高能效CMOS发射机和具有DVFS功能的高能效CMOS接收机、自适应带宽前端和自动调谐正交解调器将实现这一目标。此外，一个光学锁相环（OPLL）为基础的载波恢复方案与宽范围的电子压控振荡器（VCO）调谐将被开发。应用所提出的技术将彻底改变数据中心和高性能计算系统的未来，因为它能够提供无错误编码的低延迟互连。这个项目将涉及一个跨学科的团队2德州农工大学（TAMU）的学生和3-4国立中兴大学（NCHU）的学生。两套原型将使用两种先进的CMOS工艺，硅光子工艺和定制制造的薄膜TF-LN集成电路来实现。项目外展活动包括交流和访问活动，TAMU和NCHU学生在关键的IC设计阶段面对面地现场工作，并参加联合研讨会，通过工程丰富经验（E3）计划与高中教师互动，并通过Spark向PK-12学生介绍基本研究概念！程序.项目成果将通过在教学大纲和网站中纳入题为“相干光学系统”的新研究生课程、为学术界和工业界开发在线模块以及通过在国家和国际期刊和会议上发表来广泛传播。该奖项反映了NSF的法定使命，并通过使用基金会的知识价值和更广泛的影响审查标准进行评估，被认为值得支持。