A practical and efficient anomaly detection scheme in network data security for edge devices

As the issue of network security anomaly in edge devices becomes increasingly severe, traditional detection methods struggle to balance spatial feature mining and temporal dependency capture. Therefore, we propose a spatiotemporal feature weighted fusion model based on multi scale feature (STWMS) in network security detection. STWMS consists of three core modules, namely spatial feature learning module, multi scale feature fusion module and temporal feature learning module. This architecture integrates optimized convolutional neural network (CNN) and bidirectional long short-term memory network (BiLSTM), aiming to deeply explore the spatial and temporal features of network data for edge devices. With its efficient feature extraction capability, STWMS can accurately mine hidden abnormal patterns in edge device network data to achieve accurate detection while reducing redundant calculations. Finally, experimental results display that STWMS exhibits excellent performance in spatiotemporal feature extraction, which can significantly enhance the security detection of network for edge devices.