Cici: ucss: innovations for securing HPC cyberinfrastructure: compression-enabled confidentiality and verifiable provenance

The National Science Foundation Office of Advanced Cyberinfrastructure awarded the University of Utah $599,999 on August 1, 2026, under the Computer and Information Science and Engineering program (CFDA 47.070) to develop practical security mechanisms for high-performance scientific computing workflows.

The project develops a data-centric software library addressing confidentiality and integrity risks as large-scale simulations in weather prediction, energy systems, materials science, and biomedical modeling move floating-point data across shared and cloud-adjacent infrastructure. The library integrates three capabilities: application-specific compression using autoencoder-based methods to reduce communication and storage costs; compression combined with standards-based authenticated encryption to provide strong confidentiality within performance constraints; and proof-carrying artifacts—scientific data objects carrying compact, cryptographically verifiable receipts that enforce approved transformations such as reblocking, recompression, and format conversion. For sensitive data, these receipts use zero-knowledge proofs, allowing downstream users to verify integrity and provenance without accessing the underlying scientific state. The resulting tools are evaluated on representative simulation applications and released as open-source software to support adoption across scientific cyberinfrastructure.

Work is performed in Salt Lake City, Utah. The project runs through July 31, 2029. University of Utah's Office of Sponsored Projects Division is the recipient.

Abstract

This project develops practical security mechanisms for high-performance scientific computing workflows. Large-scale simulations in areas such as weather prediction, energy systems, materials science, and biomedical modeling generate and move large volumes of floating-point data across compute nodes, storage systems, and workflow services. As these workflows increasingly use shared and cloud-adjacent infrastructure, data may pass through networks and intermediary services that are not uniformly trusted. This creates two key risks: sensitive simulation data may be exposed, and intermediate or final artifacts may be modified or misrepresented before they are used for scientific or operational decisions. The project addresses these challenges through a data-centric software library with three integrated capabilities. First, IT develops application-specific compression, including autoencoder-based methods, to reduce communication and storage costs for simulation data. Second, IT combines compression with standards-based authenticated encryption, making strong confidentiality practical within the performance constraints of modern computing systems. Third, IT introduces proof-carrying artifacts: scientific data objects that carry compact, cryptographically verifiable receipts enforcing that approved transformations, such as reblocking, recompression, and format conversion, are correctly applied. When data are sensitive, these receipts can use zero-knowledge proofs so that downstream users can verify integrity and provenance without learning the underlying scientific state. The resulting tools strengthen trust in scientific cyberinfrastructure by enabling secure, auditable, and reproducible workflows without sacrificing performance. They are evaluated on representative simulation applications and released as open-source software, supporting adoption across scientific computing communities and training students at the intersection of high-performance computing and cybersecurity. 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.- subawards are planned for this award.