A Lightweight Deep Learning-Based Method for Anomalous Traffic Detection in Wireless Communication Networks

Aiming at the problem that the application of deep learning method in abnormal traffic detection of wireless communication network is limited due to its high complexity, a new method based on lightweight deep learning for abnormal traffic detection of wireless communication networks is proposed. According to the actual situation of wireless communication network traffic, the collection strategy is dynamically adjusted through the adaptive traffic collection method to ensure the effective collection of key traffic data. After the collected network traffic data are converted into a two-dimensional gray-scale image, the image is input into the lightweight convolutional neural network MobileNetV3 model. This model significantly reduces the amount of parameters and computation of the model through deeply separable convolution, effectively reducing the computational complexity and dealing with the problem of low fusion of different channel features when detecting wireless communication network traffic with a large amount of noise and redundant information, quickly extracts important features useful for traffic anomaly detection. At the same time, based on the extracted features, the model classifies the traffic data through the Softmax layer to detect abnormal traffic in wireless communication networks. Experimental results show that this method not only improves the efficiency and accuracy of detection, but also significantly enhances the stability and adaptability of wireless communication networks in the face of complex environments. It can be effectively applied to the field of wireless communication network anomaly detection.