Optimized Quantum Equivariant CNN for Real-Time Intrusion Detection in Drone Networks

Unmanned Aerial Vehicles (UAVs) are increasingly used in real-time surveillance, logistics, and monitoring of infrastructures. Their dependency on open wireless networks, however, leaves them highly susceptible to cyber-attacks in the form of spoofing, jamming, and denial-of-service attacks. Conventional intrusion detection systems (IDS) are ineffective in high-dimensional, real-time UAV data due to low sensitivity and adaptability. Our framework introduces multiple enhancements in both temporal and structural analysis of UAV data when compared to conventional approaches like ConvLSTM or FFCNN. This calls for a dynamic, deep learning-based solution specifically designed for UAV network environments. This paper introduces Clouded Leopard Optimized Quantum Equivariant Convolutional Neural Network-Enhanced Intrusion Detection for Real-Time Drone Network Security (QE-CNN-CLO). Each component made to address a particular intrusion detection challenge in drone networks. The approach starts with RN-Cluster-based SMOTE (RNC-SMOTE) for preprocessing to solve UAV network data class imbalance and noise. This is followed by Return-Aligned Decision Transformer (Re-ADT) for feature engineering through temporally structured, context-aware embeddings generation. The embeddings generated are input into a Quantum Equivariant Convolutional Neural Network (QE-CNN) for intrusion detection, using symmetry-aware feature extraction. Finally, Clouded Leopard Optimization (CLO) optimizes QE-CNN parameters to provide optimal classification performance. Experimental results show unparalleled effectiveness with 99.26% accuracy, 99.10% F1-score, 99.88% precision, and 99.37% recall, outperforming conventional IDS baselines significantly. Under dynamic aerial environments, these components cooperate to provide real-time, adaptive intrusion detection. In conclusion, QE-CNN-CLO offers a robust, real-time, and adaptive intrusion detection system for UAV networks with guaranteed cybersecurity resilience improvement in dynamic aerial environments. This makes QE-CNN-CLO particularly suitable for latency-suitable UAV applications.