SecureNet: Innovating Wireless Network Security Assessment with CNN-LSTM Fusion, Self-Attention, and Feedforward Network

With the rapid advancement of information technology, wireless networks have become integral to daily life and work, particularly with the widespread deployment of 5G technology. This expansion, while reducing IT infrastructure costs, introduces significant security challenges. These challenges are exacerbated by the large volume, variety, and high dimensionality of network data, impacting the efficiency and accuracy of network security situation assessment (NSSA). To address these issues, we propose CLAFNet, a novel model integrating Convolutional Neural Network (CNN), Long Short-Term Memory (LSTM), self-attention, and Feedforward Neural Network (FNN). CLAFNet leverages the strengths of convolutional and recurrent neural networks, along with self-attention mechanisms, to identify dependencies among multi-dimensional inputs and capture complex long-term relationships. Furthermore, it uses a feedforward network to mitigate nonlinear transformation losses. We employ CLAFNet to develop an innovative NSSA method for real-time wireless network security evaluation. Our prototype implementa-tion and performance evaluation demonstrate that this method significantly outperforms existing NSSA approaches in terms of efficiency and accuracy.