Enhanced Anomaly-Based Detection of the Distributed Denial of Service Attack using Modular Memetic Behavioral Analysis

Abstract

The Internet’s popularity has proven to be an effective mode for data dissemination, and also advance the proliferation of adversaries whose exploits network for personal gain via unauthorized access that compromises a user device. Adversaries have achieved such feats via socially-engineered, subterfuge schemes – some of which deny users of network resources. These distributed denial of service (DDoS) attacks are carefully crafted to impact a large magnitude with the capability to wreak havoc at high levels of network infrastructures. This study posits a deep learning approach to distinguish between benign exchange of data and malicious attacks from data traffic. With benchmark ensemble such as XGBoost, Random Forest and Decision Tree – the results shows our proposed ensemble yields F1 of 0.9945, and outperforms XGBoost, RF and DT (with F1 of 0.9925, 0.9881 and 0.9805 respectively); And with an Accuracy of 0.9984 to outperform XGBoost, RF and DT (with 0.9981, 0.9964 and 0.9815 respectively). The proposed ensemble incorrectly classified only 283-instances with 13,418 correctly classified test instances with a 99.84% accuracy. Result shows our use of the deep learning memetic model effectively differentiate between genuine and malicious packets via anomaly-based detection. Keywords: Memetic Algorithm, Random Forest, XGBoost, feature selection, imbalanced dataset Aims Research Journal Reference Format: Sulu, G.A., Akazue, M.I., Edje, A.E., (2024): Enhancing computer network intrusion detection using the network behaviour analysis technique: a pilot study. Advances in Multidisciplinary and Scientific Research Journal Vol. 10. No. 2. Pp 125-142 www.isteams.net/aimsjournal. dx.doi.org/10.22624/AIMS/V10N2P11