基于Wi-Fi感知的异常行为自适应双模检测方法
By employing a dual-mode detection method combining Wi-Fi sensing and millimeter-wave radar, along with federated learning and selective reporting, the issues of continuous reporting, personalized adaptation, and privacy protection in home anomaly detection are resolved, achieving high accuracy and low false alarm rate in home anomaly detection.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- SHENZHEN CUDY TECH CO LTD
- Filing Date
- 2026-06-17
- Publication Date
- 2026-07-17
AI Technical Summary
Existing home-based abnormal behavior detection solutions suffer from problems such as continuous reporting issues, lack of ability to distinguish between abnormal and normal behavior, inability to adapt to individual differences, outdated models, and insufficient privacy protection.
An adaptive dual-mode detection method based on Wi-Fi sensing is adopted. The sensing nodes collect channel state information in real time, and a baseline model is constructed by combining federated learning. The deviation degree is calculated by using the Mahalanobis distance algorithm, and the millimeter-wave radar is woken up for secondary confirmation. Privacy leakage and false alarms are avoided by using a selective reporting mechanism and KL divergence update model.
It improves the accuracy and privacy protection of abnormal behavior detection, reduces the false alarm rate, adapts to individual differences, solves the model drift problem, and reduces interference from indiscriminate reporting.
Smart Images
Figure CN122420818A_ABST