A real-time monitoring and blocking method and system for abnormal behavior of an internet of things terminal
By generating behavior interval graphs, conflict intensity tables, and scheduling conflict groups, combined with isolated forest sorting, dynamic capture of thread behavior and isolation of abnormal threads are achieved, solving the problem of thread behavior identification and tracing in existing technologies, and improving the real-time performance and effectiveness of security state management.
Patent Information
- Application Number
- CN202511441085.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2025-10-10
- Publication Date
- 2026-07-24
- Estimated Expiration
- 2045-10-10
AI Technical Summary
Existing technologies struggle to effectively identify abnormal thread behavior in multi-threaded resource contention scenarios. They lack detailed analysis of thread-level resource access sequences and dynamic interaction trajectories, cannot identify high-conflict paths in real time, and lack dynamic context analysis capabilities. As a result, abnormal thread behavior is hidden in frequent interactions and cannot be captured and traced in a timely manner.
By extracting thread execution cycles and resource access sequences, a behavior interval graph is generated, a conflict intensity table is constructed, high-intensity thread pairs are filtered, time period overlap rate is calculated, scheduling conflict groups are generated, and isolated forest sorting is adopted in combination with control command frequency and resource weight value. Permissions are suspended and resource call entry points are blocked to generate thread blocking sets and establish a conflict topology graph.
It achieves dynamic capture of thread behavior and efficient identification of abnormal pressure points, can isolate threads with abnormal behavior, statically freeze high-conflict paths, and realize the source tracing and policy adjustment of safety status under multi-threaded resource contention.