Abstract: Spread Spectrum techniques (SS) attract the attention they are widely used in wireless communications and radar. Direct sequence spread spectrum and frequency hopped spread spectrum are the two most used techniques. In this research, the clustering of these signals is performed by feature-based model. Features are extracted by Gray Level Co-occurrence Matrix (GLCM), gray level run-length matrix, cumulants, moments, PCA, KPCA and Fast-ICA features. Clustering by GLCM features gets the best result which is one of the common textures features extraction techniques. The selection of relevant features is the big challenge. Therefore, the main contribution is to optimize the SS identification based on clustering techniques by decreasing the number of features without accuracy degradation which is based on filters, sequential forward selection and binary metaheuristic search strategies such as Binary Particle Swarm Optimization (BPSO), Genetic Algorithm (GA), bat feature selection (BBA) and hybrid Whale Optimization Algorithm with Simulated Annealing and Tournament mechanism (WOASAT). BPSO as a wrapper method is proposed to the optimization as it outperforms the other techniques in terms of accuracy and selected features with k-means, k-medoids or HAC.
Haidy S. Fouad, Hend A. Elsayed and Shawkat K. Girgis, 2020. RETRACTED ARTICLE: Optimizing Classification of Spread Spectrum Signals Based on Features Extraction. Asian Journal of Information Technology, 19: 1-11.