Smallest Integrated Networks for Highest Data Density

数据的指数级增长继续迫使世界各地发生重大的社会变革。从科学模拟到消费电子产品，数据处理给我们带来了娱乐，但更重要的是拯救了生命，影响了经济增长。然而，世界上许多地区的基础设施薄弱，或者无法负担现代计算系统的高昂成本。事实上，为了满足对数据处理和存储的高需求，具有数千个处理器的复杂系统通过电气开关相互连接，并行工作，内在地限制了它们在性能方面的长期可扩展性。随着光子集成的最新进展，光技术可以取代电互连。拟议的计划解决了光子学向高数据密度方向的系统集成。预期的研究成果将在现代计算机系统中带来实用的可扩展性和性能增强。此外，随着性能的提高，价格下降使得这些现代计算系统更容易被世界各地访问，包括加拿大偏远地区，那里的能源不易获得，以处理采矿业或支持北部社区所需的大量数据。*LiboIron-Ladouceur博士的建议是研究基于光互连的基于CMOS的异质计算系统中最高片上带宽密度的基本限制。她建立了一种研究方法，基于她最近的两项创新，这两项创新是通过她的第一个NSERC发现计划开发的低功率光子互连。在第一项发明中，她实验证明了一种能够同时支持调制光信号的高带宽电纳米级波导，这导致了数据通信中潜在的重要范式转变，其中金属传输线可以用于电气上的控制信号，以及光学上的大型数据传输。在她的第二项发明中，LiboIron-Ladouceur博士证明了在单个芯片上集成互连网络体系结构的可行性，将体系结构的规模从m2缩小到mm2，缩小了一千倍，从而提高了带宽密度。该研究计划的目标是开发在处理器之间的互连中具有增强的带宽密度的计算机体系结构。该多学科研究项目共为27名学生提供了良好的培养平台，其中本科生10人，硕士9人，博士后5人，博士后3人。该预算利用了NSERC CREATE计划提供的光子集成方面的高级培训机会以及麦吉尔大学的贡献。因此，请拨的数额占拟议支出的45%。**

The exponential increase in data continues to force important societal changes across the world. From scientific simulations to consumer electronics, data processing has entertained us, but more importantly saved lives and influenced economical growth. However, many parts of the world have weak infrastructures, or are unable to afford the high cost of modern computing systems. Indeed, to meet the high demand for data processing and storage, complex systems with thousands of processors are working in parallel interconnected through electrical switches, inherently limiting their long term scalability in terms of performance. With recent advancements in photonic integration, the electrical interconnectivity can be replaced with optical technologies. The proposed program addresses the system integration of photonics towards high data density. The expected research outcome will lead to practical scalability with performance enhancement in modern computer systems. Furthermore, price deflation which comes with the increase in performance makes these modern computing systems more accessible to all parts of the world, including remote areas of the Canadian landscape where energy is not easily available to process large amount of data required such as in the mining industry or in supporting the northern communities.******Dr. Liboiron-Ladouceur's proposal is on the investigation of the fundamental limits of the highest on-chip bandwidth density in optically interconnected CMOS-based heterogeneous computing systems. She builds a research methodology upon two of her recent innovations developed through her first NSERC Discovery program on low-power photonic interconnects. In the first invention, she has experimentally demonstrated a high-bandwidth electrical nanoscale waveguide capable of simultaneously supporting modulated optical signals leading to a potentially important paradigm shift in data communication where metal transmission lines can be used electrically for control signals and optically for large data transfer. In her second invention, Dr. Liboiron-Ladouceur has demonstrated the feasibility of integrated interconnection network architectures on a single chip downsizing architecture by a factor of a thousand from m2 to mm2 leading to bandwidth density enhancement. The objective of the research program is to develop computer architectures with enhanced bandwidth density in the interconnectivity between processors. The multi-disciplinary research program offers and excellent training platform for a total of 27 students, which includes 10 undergraduates, 9 master's, 5 doctoral, and 3 postdoctoral students. The budget leverages advanced training opportunities in photonic integration offered by an NSERC CREATE program along with contributions from McGill University. Thus, the amount requested represents 45% of the proposed expenditure. **