ANNOUNCEMENT

NOBEL PRIZE 2026

This year's laureates, their fields, and the discoveries that earned them the prize.Live nowSee the winners
Moonbase
← Back to Awards
NSFDirectorate for Computer and Information Science and Engineering

CICI: UCSS: FILOLI: Auditable and Side-Channel-Resilient Confidential Computing Infrastructure

Active
Institution
University of Southern California, CA
Principal investigator
Mengyuan Li
Funding
$600K
Duration
2026–2029
Open portfolio

Moonbase score

Still being scored

Abstract

Collaborative science depends on trusted shared computing infrastructure, where institutions pool sensitive data such as medical records and proprietary models for joint analysis. Confidential computing offers a hardware-assisted foundation for creating trusted execution environments where such sensitive computations can run without exposing data to cloud operators or collaborating parties. However, today's confidential computing platforms still face two major gaps: privileged system software and shared hardware can leak secrets through side channels, and most systems provide limited evidence that computations were executed with adequate privacy, integrity, and accountability. This project develops FILOLI, an auditable and side-channel-resilient confidential computing infrastructure for collaborative science. FILOLI integrates CPU-based confidential virtual machines with graphics processing unit trusted execution environments to support secure accelerated workloads. Its first thrust confines the privileged host runtime using a hardware root of trust, Trusted Platform Module measurements, and runtime code admission, ensuring that the hypervisor and host services remain in a known-good state while minimizing unnecessary interfaces and shared-resource exposure. Its second thrust enforces secure execution inside confidential virtual machines through a monitor that mediates sensitive operations and produces cryptographically sealed, append-only audit records. Remote attestation links these records to genuine hardware so collaborators and auditors can verify platform integrity and the integrity and completeness of the logs. The project releases open-source prototypes, benchmarks, and toolkits, integrates research artifacts into education and training activities, and helps strengthen the national cybersecurity workforce. 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.

Research team

  • Mengyuan LiPI
  • Murali AnnavaramCo-PI
  • Ruishan LiuCo-PI

LOADING MOONBASE SCORE

Directorate for Computer and Information Science and Engineering01002627DB NSF RESEARCH & RELATED ACTIVITCybersecurity Innovationlinkssourcegraphicsassistedthrusttrustedremainwithoutconfinessecrets

Questions for University of Southern California

Answered by the team behind this research.

Same institution, same topic, scored first.