ABSTRACT
Artificial intelligence helps to design better solutions to detect possible attacks. Intrusion detection systems using machine learning classifiers easily justify the trust factor in terms of accuracy. This chapter provides a comparative study of machine learning techniques to investigate the trust of IoT-based aerial ad hoc networks. Aerial vehicles move in three dimensions. Due to the highly mobile topological design of the UAV network, it is vulnerable. Intruders take advantage of the vulnerabilities and deploy cyber attacks to unbalance the overall network. The five machine learning techniques—XGBoost, naïve Bayes, complement naïve Bayes, ADA boost, and GaussianNB—are simulated using the UNSW-NB 15 dataset. For a detailed comparison, metrics like accuracy, precision, recall, F1-score, and support are used. Moreover, theoretical analysis of IDS using machine learning is also incorporated. Simulation results indicate that ADA boost performs better in comparison with the other techniques.
