Design of Computer Network Security Defense System Based on Big Data

The traditional computer network (CN) security defense (SD) system is mainly based on feature matching, rule and signature detection, which is difficult to adapt to new and unknown network threats and attack technologies in time. The effectiveness and adaptability of cybersecurity can be improved by leveraging big data technologies to enable more accurate threat detection and analysis to reduce false positives and missed alarms. The paper collected a large amount of network data, preprocessed the data, converted the data into time series data, built a Long Short-Term Memory (LSTM) model, used mean square error to feedback and adjust the network parameters of the LSTM model, and outputted the network threat prediction results. The network SD system combined with big data technology and LSTM was compared with the traditional feature-matching detection system and Support Vector Machine (SVM)-based detection system. The threat detection accuracy of the system combined with the LSTM model was 98.8%, and the average response time was 5.1 s. The combination of big data technology and LSTM can effectively improve the performance of network SD.