xGFabric: Coupling Sensor Networks and HPC Facilities with Advanced Wireless Networks for Near-Real-Time Simulation of Digital Agriculture

Our objective is to design and evaluate an end-to-end, multi-scale, and adaptive distributed system that couples leadership class resources with scientific instruments, sensors, and actuators through a tiered network of supporting computational and storage resources. Our plan is to develop the xGFabric comprising the software systems necessary to sustain a resilient and performant computing capability that uses 5G and 6G technologies to couple computing, storage, sensing, and actuation at multiple resource scales to enable the next generation of scientific applications. Enabling this requires designing new abstractions, programming systems, and middleware that unify and make portable compute and data resources across resource types and scales. Our second research objective is to study the xGFabric comprehensively and end-to-end rather than as an integration of separate technologies; each developed for a different purpose. The key to this approach is integrated support for HPC resources and distributed workflow management, which are critical to their efficient use. We plan to use, as science drivers, applications from advanced wireless digital agriculture. The research results and prototypes will leverage agricultural field testbeds at the University of Nebraska and the University of California for testing and validation. Our previous systems research for 5G and 6G and the Internet of Things and systems for multi-messenger astronomy demonstrate that such objectives are feasible. In this work, we will develop a sensor virtualization capability for integrating with 5G and 6G network slicing. We will also pursue a full-stack, distributed, multiscale platform for implementing the xGFabric and make these artifacts available as open-source prototypes.