What the study found
The study found that an AI-driven hybrid architecture for multi-cloud storage can use telemetry-guided adaptive fragmentation to improve predictive reliability and support reconstruction-resistant storage. It also found that cloud-only access is not enough to rebuild the data without the local vault fragments and the encryption key.
Why the authors say this matters
The authors conclude that telemetry-driven adaptive fragmentation helps establish a resilient zero-trust framework for secure multi-cloud storage. They present this as a way to address concerns about confidentiality and recoverability after provider-side breaches.
What the researchers tested
The researchers introduced an AI-driven hybrid architecture that predicts fragment sizes from real-time telemetry, including bandwidth, latency, memory availability, and disk I/O. They first used synthetic telemetry to test feasibility and scalability, then used hybrid telemetry combining real Microsoft system traces and Cisco network metrics to test generalization under realistic variability.
What worked and what didn't
Across the evaluations, XGBoost and Random Forest achieved the highest predictive accuracy. Neural Network and Linear Regression models showed moderate performance. Security validation showed that partial-access and cloud-only attack scenarios could not reconstruct the data without the local vault fragments and the encryption key.
What to keep in mind
The abstract does not describe detailed limitations beyond the evaluation setup. The reported findings are based on synthetic telemetry and a hybrid dataset with real system and network traces, so the summary here is limited to those stated conditions.
Key points
- The architecture uses telemetry-guided fragmentation to predict fragment sizes from bandwidth, latency, memory availability, and disk I/O.
- All payloads are compressed, encrypted with AES-128, and dispersed across independent cloud providers.
- Two encrypted fragments are retained in a VeraCrypt-protected local vault to prevent cloud-only reconstruction.
- XGBoost and Random Forest achieved the highest predictive accuracy in the evaluations.
- Partial-access and cloud-only attack scenarios could not reconstruct the data without the local vault fragments and the encryption key.
Disclosure
- Research title:
- Telemetry-guided multi-cloud storage resists reconstruction
- Authors:
- Munir Ahmed, Jiann-Shiun Yuan
- Institutions:
- University of Central Florida
- Publication date:
- 2026-01-27
- OpenAlex record:
- View
Get the weekly research newsletter
Stay current with peer-reviewed research without reading academic papers — one filtered digest, every Friday.