Collaborative Research: DESC: Type I: Transforming Integrated Electro-Photonic Fabrics for Light Speed Communication and Computation
openNSF
Current electronic systems that power today's intelligent digital solutions including smartphones, wearables, smart vehicles, and smart home appliances are facing significant limitations in scaling up to meet the performance demands of emerging smart applications. The goal of realizing smart cities, smart agriculture, and smart healthcare is jeopardized due to the bottlenecks facing traditional electronic platforms at the heart of these emerging applications. Silicon photonics represents an innovative paradigm that can overcome these bottlenecks, by supporting light-speed and high throughput communication and computation. However, the design of silicon photonics based smart computing platforms is still in its infancy, with many open challenges. One of the most significant of these challenges is to design integrated electro-photonic platforms that are truly sustainable while also simultaneously minimizing manufacturing costs, energy footprint, and fault susceptibility. This project will take the first steps towards realizing efficient electro-photonic computing platforms that far surpass traditional electronic platforms in terms of performance, energy-efficiency, and reliability. In doing so, this project will help establish a new ecosystem of development of emerging digital intelligence platforms, for multiple application use cases that span datacenters, scientific computing, and edge systems, where photonics is making inroads. This outcome will help reduce costs and overheads of designing future smart computing platforms, which will benefit technological innovation across U.S. consumer, industrial, and defense sectors. This will also help the U.S. outcompete global competitors to emerge as the pioneer in innovative computing design for such emerging technologies. By exposing students to diverse aspects of electro-photonic circuit design, yield analysis, lifetime evaluation, modeling and trade-offs, and computing architecture design, the proposed research will contribute towards educating an agile high-tech workforce that will maintain continued U.S. leadership in technological innovation, especially in emerging post-Moore technologies.
This project will employ the principles of heterogeneity, reconfigurability, and recycling to transform performance, functional flexibility, and resource utilization of integrated electro-photonic platforms. The overarching goal will be to extend operational lifespan and co-reduce the embodied and use-phase costs of emerging heterogeneous integrated electro-photonic-based 2.5D computing platforms. To that end, this project will make three synergistic contributions: 1) Analyze variations and aging-aware footprints, by developing the first modeling framework for integrated electro-photonic platforms, quantifying sustainability costs of enabling variation resilience, and characterizing impacts of aging in these platforms, 2) Prolong carbon-minimum lifespan and co-optimize embodied and operational metrics, by developing techniques to prolong integrated electro-photonic platform lifespan with design time and runtime adaptive techniques, devising cross-layer techniques to reduce operational overheads, managing failure risk with runtime mechanisms for graceful degradation, and co-reducing use-phase and embodied sustainability metric costs, and 3) Introduce reuse and hardware polymorphism to reduce costs for multifunctional computing, communication, and storage components in integrated electro-photonic platforms.
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.