Detecting Anomalies in Network Traffic Using Fuzzy Logic with A Comparative Analysis Against Deep Learning Techniques

Abstract

Network traffic anomaly detection plays a crucial role in cybersecurity by identifying suspicious activities that may indicate cyberattacks, malware infections, or unauthorized access. Traditional rule based methods often struggle with evolving attack patterns, necessitating more adaptive and intelligent approaches. This study explores the effectiveness of Fuzzy Logic and Deep Learning models (LSTM, Autoencoder, CNN) for detecting anomalies in network traffic. Fuzzy Logic offers an interpretable rule based framework for handling uncertainty, while deep learning models leverage data-driven learning for improved anomaly detection accuracy. Using publicly available datasets such as NSL-KDD and CICIDS 2017, we evaluate these methods based on key metrics such as accuracy, precision, recall, F1-score. The results indicate that while Fuzzy Logic provides reasonable accuracy (85.2%), deep learning models—particularly CNN (94.1%) and LSTM (92.4%)—demonstrate superior performance. CNN outperforms other models due to its ability to recognize spatial patterns in network traffic, while LSTM effectively captures sequential dependencies. These findings highlight the trade-off between interpretability and accuracy, suggesting that deep learning models are more effective for real-time and large-scale anomaly detection, whereas Fuzzy Logic remains a viable option where transparency is prioritized.