FuSe-TG: Regaining the Edge

This project brings together a diverse community of experts from academia and industry to explore technologies for designing the edge computers of the future. These computers will be capable of untethered operation for weeks or months while providing advanced features such as artificial intelligence, seamless interfacing with humans, multiple sensing modalities, high-speed communication with the cloud, and robust security. Realizing this vision calls for breakthroughs in energy efficiency, security and privacy technologies, and the integration of computing, communication, and sensing. Current design practices are unlikely to yield these breakthroughs, as their siloed paradigm that coerces materials scientists, device engineers, computer architects, and algorithm designers to work in relative isolation from each other and across rigid interfaces has reached its limits. The research team will leverage its wide-ranging expertise to explore the co-design of novel materials, devices, architectures, and algorithms, relying on fluid interfaces and cross-level simulators to uncover potentially orders of magnitude in energy efficiency improvements that enable next-generation application, security, sensing, and communication features for edge computers. It will also explore workforce development programs at the master’s level that yield industry-bound graduates with co-design expertise. The research team will come together around a series of quarterly workshops on energy-efficient devices and architectures, security technologies for privacy protection, hardware-software co-design for adaptive operation, and distributed processing for high-bandwidth communication and sensing. In collaboration with application-domain experts from industry, these four themes will be explored in the context of two specific use-inspired cases that occupy different parts of the edge-computer design spectrum: distributed wearables and deeply embedded implants for health, and distributed autonomous agents (e.g., vehicles, drones). In addition to exposing the potential for significant synergies across the design abstraction layers, the workshops and ensuing interactions are expected to yield a curated data and software framework that consists of existing data sets, interfaces, tools, and simulators and includes examples of design activity that exercises the framework within each abstraction layer as well as across abstraction layers. 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.