Application of NAWL-ILSTM Algorithm in Network Security Situation Awareness Prediction

The high-speed advancement of the Internet has increased the risk of network attacks and brought considerable challenges to network security. The existing network security measures, such as firewalls and virus-killing technologies, are insufficient to prevent network attacks effectively. Therefore, it is necessary to establish a network security situational awareness prediction model. According to the improved Adaptive Moment Estimation algorithm, this paper optimizes the online update mechanism of the Long Short-term Memory network, updates the parameters in real-time, and improves the model’s accuracy. The Look-ahead algorithm is introduced to lift the network model’s convergence rate, reduce the training cost, and deduct the prediction error. The established model was verified through experiments. The original Long short-term memory network, Support Vector Machines algorithm and Radial Basis Function were used as the comparison models. Comparative experiments have shown that the error between the predicted values of the designed model and the actual values is minimal. Compared with the support vector machine and Radial basis function, the average absolute percentage errors are 0.0198, 0.0523, and 0.0225, respectively; The Standard errors are 0.0126, 0.0326, and 0.0157. Network security situational awareness prediction accuracy is as high as 95.343%. Therefore, the proposed optimized model has particular perception and prediction capabilities for network attacks, and its development potential and reference value are worth exploring. © (2024), (International Journal of Network Security). All rights reserved.