A method and system for detecting and warning of cryptographic protocol anomalies

By combining multimodal feature extraction and interpretable artificial intelligence models, interpretable early warning decisions are generated. The rule base and model are optimized under the federated collaborative nodes, which solves the problems of insufficient interpretability and data privacy protection in existing encryption threat detection and achieves efficient encryption protocol anomaly detection.

CN122137619APending Publication Date: 2026-06-02SHANGHAI UNI SENTRY INTELLIGENT TECH CO LTD

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
SHANGHAI UNI SENTRY INTELLIGENT TECH CO LTD
Filing Date
2026-03-04
Publication Date
2026-06-02

Smart Images

  • Figure CN122137619A_ABST
    Figure CN122137619A_ABST
Patent Text Reader

Abstract

This invention provides a method and system for detecting and warning of anomalies in encryption protocols, belonging to the field of network security technology. The method includes: extracting a multimodal feature set from the target encryption protocol communication stream; inputting it in parallel into an adaptive rule engine and an interpretable artificial intelligence model to obtain rule-triggered events and confidence weights, a first anomaly score, and an attribution explanation report, respectively; calculating a second anomaly score based on the above outputs using an improved scoring algorithm; fusing the first and second anomaly scores to obtain a comprehensive anomaly score and generating a warning decision; and uploading the warning information to a federated collaborative node to drive the collaborative evolution of the global rule base and the model. This invention, employing the aforementioned method and system for detecting and warning of anomalies in encryption protocols, solves the collaborative problems of poor interpretability, lagging knowledge updates, and data silos in encryption threat detection through deep collaboration between an interpretable artificial intelligence model and a rule engine, achieving accurate, reliable, and adaptive anomaly warnings.
Need to check novelty before this filing date? Find Prior Art