Quantum Audio Steganalysis Based on Quantum Fourier Transform and Deutsch-Jozsa Algorithm

In recent years, researchers have considered quantum steganography and its various methods with the development and progress of research in computation theory and quantum signals processing. The destructive use of quantum steganography methods to establish illegal covert communications is rising, so it is essential to introduce ways to detect hidden data in a quantum medium. Accordingly, this paper presents a frequency-based universal audio steganalysis approach to detecting quantum steganography. First, based on the quantum Fourier transform, the characteristic of the quantum spectrum centroid (QSC) was computed, and its circuit network was implemented to extract feature vectors. The proposed method classifies quantum audio signals using a quantum machine learning approach called a quantum ensemble of quantum classifiers. This approach was implemented within the framework of the Deutsch-Jozsa algorithm, which uses the superposition property to create an ensemble of classifiers evaluated in parallel, significantly increasing the computational speed. The accuracy weight of the classifiers is adjusted based on the classifiers' performance in training data classification; finally, the measurement of the first n qubits of the Deutsch-Jozsa algorithm predicts whether the quantum audio signals belong to the stego or clean class. The idea stems from the classic ensemble methods that try to build more robust models by combining different classifiers. The results show that the proposed frequency domain steganalysis method with 95% accuracy performs better than the previous methods in the time domain.