Edge devices such as smartphones/tablets, Internet-of-Things and connected vehicles will continue to proliferate in our environments. They possess sensing and computing resources and are able to gather and process various data to produce information and insights for novel applications in many domains, such as smart vehicles/buildings/campuses/transportation. However, many such applications require peripheral modules (especially sensors/radios) customizable in both types and quantities, a capability gravely lacking in existing edge devices. This project will design, create and evaluate a novel hardware and software platform where heterogeneous peripheral modules and modularized FPGA/software computation components can be easily composed electrically and computationally like interlocking Lego pieces, to create various customized edge sensing and computing devices required in these "smart" applications.
The PIs will develop Field-Programmable Gate Array (FPGA)-based peripheral controllers to connect diverse sensors/radios, and accelerators for common data/signal processing and security algorithms; an edge tailored stream processing system that can compose reusable hardware/software computation components to create applications; and pilot applications in smart vehicles/buildings/campuses to test and validate the platform. The easy customization in sensor/radio types and quantities, and flexible composability in FPGA/software computation are urgently needed but unavailable from both commodity or research prototypes. Researchers and educators in edge computing, mobile sensing, connected vehicles and Internet-of-Things communities can all benefit immensely from this platform, using it to create customized devices and compose novel applications. Collaborations with industry partners are established for both pilot studies and potential technology transfer.