Improving Intrusion Detection in UAV Communication Using an LSTM-SMOTE Classification Method

Abstract

Unmanned Aerial Vehicles (UAVs) proliferate quickly and play a significant part in crucial tasks, so it is important to protect the security and integrity of UAV communication channels. Intrusion Detection Systems (IDSs) are required to protect the UAV communication infrastructure from unauthorized access and harmful actions. In this paper, we examine a new approach for enhancing intrusion detection in UAV communication channels by utilizing the Long Short-Term Memory network (LSTM) combined with the Synthetic Minority Oversampling Technique (SMOTE) algorithm, and this integration is the binary classification method (LSTM-SMOTE). We successfully achieved 99.83% detection accuracy by using the proposed approach and the Canadian Institute for Cybersecurity Intrusion Detection Evaluation Dataset 2017 (CICIDS2017) dataset. We demonstrated the efficiency of LSTM-SMOTE in defending UAV communication channels against possible attacks and bolstering the overall security posture through the use of a realworld scenario.