Intrusion Detection System Based on Quantum Generative Adversarial Network

Intrusion Detection Systems (IDS) are crucial for ensuring network security in increasingly complex digital en vironments. Among IDS techniques, anomaly detection is effective in identifying unknown threats. However, classical machine learning methods face significant limitations, such as struggles with high-dimensional data and performance constraints in handling imbalanced datasets. Generative Adversarial Networks (GANs) offer a promising alternative by enhancing data generation and feature extraction, but their classical implemen tations are computationally intensive and limited in exploring complex data distributions. Quantum GANs (QGANs) overcome these challenges by leveraging quantum computing’s advantages. By utilizing a hybrid QGAN architecture with a quantum generator and a classical discriminator, the model effectively learns the distribution of real data, enabling it to generate samples that closely resemble genuine data patterns. This capability enhances its performance in anomaly detection. The proposed QGAN use a variational quantum circuit (VQC) for the generator and a neural network for the discriminator. Evaluated on NSL-KDD dataset, the QGAN attains an accuracy of 0.937 and an F1-score of 0.9384, providing a robust, scalable solution for next-generation IDS.