Accès ouvert · CC BY

RansomShield: Threat-score-driven adaptive backup with multi-modal behavioural ransomware detection

Article scientifique 2026 Sciences mathématiques et statistiques Anglais

Résumé

Abstract Ransomware defences usually treat detection and backup as separate systems: detectors raise alerts while backups run on fixed schedules, regardless of the current threat. This paper presents RansomShield, a proof-of-concept architecture in which a continuous threat score from a multi-modal detector directly governs backup frequency, snapshot immutability and isolation. The detector combines four signals accounting for a random forest on memory-forensic features, a rule-based API-call sequence scorer, a storage I/O anomaly analyser, and a file-type-aware Shannon entropy monitor. Their outputs are fused by confidence-weighted averaging with exponential smoothing. A four-level controller with hysteresis maps the smoothed score to a backup policy. The components were built using CIC-MalMem-2022, RanSMAP, NapierOne and the Mendeley Ransomware/Benignware System Calls dataset. On CIC-MalMem-2022, the memory detector achieved a cross-validated F1 score of 0.9999 ± 0.0002 for binary classification and 76.07% accuracy for 16-class classification. In simulated attacks modelled on seven ransomware families, all seven were detected after a mean of 13.4 file-system events and escalated to immutable, air-gapped snapshots at 30-second intervals, while a benign scenario produced a 5.6% event-level false-positive rate.

Citer ce document

OLOJA, S. V., Famutimi, R., OLOJA, O. A., BALOGUN, J. T., & SAMSON, E. A. (2026). RansomShield: Threat-score-driven adaptive backup with multi-modal behavioural ransomware detection. Research Square. https://doi.org/10.21203/rs.3.rs-11337939/v1

Exporter : BibTeX · RIS (Zotero, Mendeley, EndNote)

Accès au document

Texte intégral archivé sur SPHAERO, en accès ouvert : lisible gratuitement en ligne après connexion. Le document reste aussi disponible à sa source. Se connecter

Voir l'article sur le site de la revue

Licence et provenance

Licence : CC BY

Notice moissonnée depuis OpenAlex le 09/11/2026. Le document reste hébergé par sa source.
Voir le document à la source →

Statistiques

Consultations (hors robots) : 6

Téléchargements (hors robots) : 0

Méthode de comptage