A method of data security aggregation and collaborative anomaly detection for IoT devices based on federated learning

Abstract

The invention discloses an IOT device data security aggregation and collaborative anomaly detection method integrating federal learning, which relates to the IOT data security and artificial intelligence technology field.Aiming at the problems of privacy disclosure risk, poor adaptability of heterogeneous devices and low accuracy of anomaly detection in existing IOT data aggregation, this method realizes the collaborative optimization of data security and detection efficiency through the integration of hierarchical federal learning architecture design, multidimensional security aggregation mechanism and collaborative anomaly detection strategy.Specifically, it includes: building a three-level federal learning architecture of "edge node - regional aggregation node - cloud center node" to realize hierarchical access and task collaboration of heterogeneous IOT devices;In the stage of data transmission and aggregation, the dual security mechanism of homomorphic encryption and differential privacy fusion is adopted to ensure the privacy security in the process of data transmission and the reliability of aggregation results.