Research on Digital Signal Modulation Recognition Using Ant Colony Algorithm with Ensemble Machine Learning

Nowadays, the digital signal modulation recognition in non-cooperative situations is significant research area. The existing recognition methods are complex to meet the requirement of its performance which leads to reduce the recognition rate. This research proposes an Ant Colony Optimization (ACO) with Ensemble Machine Learning (ML) approach for digital signal modulation recognition. The ensemble ML includes Artificial Neural Network (ANN), Support Vector Machine (SVM), Random Forest (RF) and Extreme Gradient Boosting (XGBoost). Ensemble models provide solutions for most sophisticated and noisy signal environments since different learners are able to respond to the changes in SNR and modulation type. This approach leads to higher recognition accuracy, particularly in complex situations which makes ensemble ML is reliable and effective for digital signal modulation recognition. The ACO has an ability of generating a diverse number of solutions within a short period due to probabilistic decision making which ensures getting near-optimal solutions for feature selection and classification in modulation recognition. Also, Continuous Wavelet Transform (CWT) is used to extract features and provide multiple resolutions allowing one to detect high-frequency components and low frequency components. The ensemble ML achieved overall recognition rate of 0.96, 0.95 and 0.99 for Binary Phase Shift Keying (BPSK), Quadrature Phase Shift Keying (QPSK) and 8 Phase Shift Keying (8PSK) respectively.