Abstract: To solve the problem of slow data anomaly detection in the Internet of Things, based on the edge computing model, this research designed to add spatial correlation detection algorithm to the anomaly detection system to build the edge nodes in the network, calculate the Pearson correlation coefficient between nodes, and then establish a similarity matrix to measure the degree of correlation and evaluate the degree of anomaly of each sub sequence. Set a correlation threshold z for filtering and improving detection accuracy. In addition, this study also integrates CNN models into SDN architecture, enabling the system to process and analyze network traffic in real-time, enhancing the system's real-time streaming data analysis capabilities. At the same time, it enhances its ability to resist interference, ensuring efficient anomaly detection performance even in complex and changing network environments. After testing, the accuracy of anomaly detection in IoT networks in this study was 98.53%, and the response time to resist network attacks was 5 s. Greatly improve the efficiency of anomaly detection and ensure the security performance of network architecture.