Blind Source Signal Separation Based on Meta-heuristics Methods

Year 2024, Volume: 14 Issue: 3, 1456 - 1470, 15.09.2024

Eyup Gedikli , Emin Tuğcu

https://doi.org/10.31466/kfbd.1474735

Abstract

The blind source separation problem is the process of identifying unknown source signals from at least two mixed signals. Source signals are important for accurate diagnosis in the medical field, wireless communication, and analysis of radar, image and sound data. Independent component analysis (ICA) is often used for the problem of blind source separation. In independent component analysis, entropy and correlation compatibility are checked using advanced statistical and algebraic methods. FastICA, one of the most widely used independent component analysis algorithms for signal separation, iteration-based searches for non-Gaussianity and negentropy maximum fitness criteria. In this study, meta-heuristic algorithms (MHA), which are also iteration-based methods, were used to optimize the fitness function. The fitness function is used to generate the separation matrix for mixed signal to control convergence. In this study, the generation of the separation matrix is proposed based on the Gram-Schmidt process, which orthogonalizes the vectors. Experiments were performed using FastICA and meta-heuristic algorithms such as the firefly algorithm and the particle swarm optimization algorithm. Mixed signals are generated by mixing the signals from three sources and adding noise. In the experiments carried out by generating signals with different frequencies, more successful correlation coefficients and root mean square error results were obtained with the proposed method than the traditional FastICA algorithm.

Keywords

Firefly algorithm, Independent component analysis, FastICA, Gram Schmidt Process, Blind source separation, Particle swarm optimization

Meta-Sezgisel Yöntemlere Dayalı Kör Kaynak Sinyal Ayırma

Abstract

Kör kaynak ayırma problemi, en az iki karışmış sinyalin bilinmeyen kaynak sinyallerini belirleme işlemidir. Kaynak sinyaller, tıbbi alanda doğru teşhisin yapılmasında, kablosuz haberleşmede, radar, görüntü, ses verilerinin analizi için önemlidir. Kör kaynak ayırma probleminde yaygın olarak bağımsız bileşen analizi kullanılır. Bağımsız bileşen analizinde, ileri istatistiksel ve cebirsel yöntemler kullanılarak entropi ve korelasyon uyumluluğuna bakılır. Sinyalleri ayırmak için en yaygın kullanılan bağımsız bileşen analizi (Independent Component Analysis, ICA) algoritmalarından FastICA, Gauss dağılımı olmama ve negentropinin maksimum uygunluk kriterlerini iterasyon tabanlı olarak araştırır. Bu çalışmada, benzer şekilde iterasyon tabanlı yöntemler olan meta-sezgisel algoritmalar (MSA), uygunluk fonksiyonunu optimize etmek için kullanılmıştır. Uygunluk fonksiyonu, karışık sinyal ayırma matris üretimi ve yakınsamayı kontrol etmek için kullanılır. Bu çalışmada, vektörleri ortogonalleştiren Gram Schmidt sürecine dayalı ayırma matris üretimi önerilmiştir. Deneyler, FastICA ile meta-sezgisel (MS) algoritmalardan ateş böceği algoritması ve parçacık sürü optimizasyonu algoritmasıyla yapılmıştır. Üç kaynaktan üretilen sinyallerin karıştırılıp gürültü eklenmesi ile karışık sinyaller oluşturulmuştur. Sinyallerin farklı frekanslarda üretilerek gerçekleştirilen deneylerde, önerilen yöntem ile geleneksel FastICA algoritmasından daha başarılı korelasyon katsayısı ve kök ortalama kare hata sonuçları elde edilmiştir.

Keywords

Ateş böceği algoritması, Bağımsız bileşen analizi, FastICA, Gram Schmidt süreci, Kör kaynak ayırma, Parçacık sürü optimizasyonu

Ethical Statement

Yapılan çalışmada araştırma ve yayın etiğine uyulmuştur.

References

• Abbas, N., and Kabudian, J., (2017). Speech Scrambling based on Independent Component Analysis and Particle Swarm Optimization, The International Arab Journal of Information Technology (IAJIT) , 14(4), 109–115.
• Baysal, B., and Efe, M.Ö., (2023). A comparative study of blind source separation methods. Turkish Journal of Electrical Engineering and Computer Sciences, 31(7). https://doi.org/10.55730/1300-0632.4047
• Bonyadi, M.R., and Michalewicz, Z., (2017). Particle swarm optimization for single objective continuous space problems: a review. Evolutionary Computation. 25(1),1–54. doi:10.1162/EVCO_r_00180
• Comon, P., and Jutten, C. (2010). Handbook of Blind Source Separation: Independent Component Analysis and Applications. Oxford, UK: Academic Press.
• Feng, F., and Kowalski, M., (2018). Revisiting sparse ICA from a synthesis point of view: Blind Source Separation for over and underdetermined mixtures. Signal Processing, 152, 165–177. https://doi.org/10.1016/j.sigpro.2018.05.017
• Ghahramani, H., Barari, M., and Bastani, M.H. (2014). Maritime radar target detection in presence of strong sea clutter based on blind source separation. IETE Journal of Research, 60(5), 331–344. https://doi.org/10.1080/03772063.2014.961573
• Hyvärinen, A., and Oja, E., (2000). Independent component analysis: algorithms and applications. Neural Networks, 13 4–5, 411-430. https://doi.org/10.1016/S0893-6080(00)00026-5
• Jiang, L., Li, L., and Zhao, G.Q. (2015, January). Pulse-compression radar signal sorting using the blind source separation algrithms. International Conference on Estimation, Detection and Information Fusion (ICEDIF) (pp.268-271). Harbin. doi: 10.1109/ICEDIF.2015.7280204
• Jin-Wang, H., Jiu-Chao, F., and Shan-Xiang, L. (2014). Blind source separation of chaotic signals in wireless sensor networks. Acta Physica Sinica, 63(5): 050502. https://doi.org/10.7498/aps.63.050502
• Kumar, M., and Jayanthi, V.E., (2020). Blind source separation using kurtosis, negentropy and maximum likelihood functions. International Journal of Speech Technology, 23, 13–21. https://doi.org/10.1007/s10772-019-09664-z
• Li, H., Li, Z., and Li, H., (2016). A Blind Source Separation Algorithm Based on Dynamic Niching Particle Swarm Optimization. MATEC Web of Conferences 61, 03008. DOI: 10.1051/matecconf/20166103008
• Luo, D., Sun, H., and Wen, X., (2012). Research and Application of Blind Signal Separation Algorithm to the Aircraft Engine Vibration Signal and Fault Diagnosis Based on Fast ICA. J. Converg. Inf. Technol. 7(10), 248–254. http://dx.doi.org/10.4156/jcit.vol7.issue10.29
• Metsomaa, J., Sarvas, J., and Ilmoniemi, R.J. (2014). Multi-trial evoked EEG and independent component analysis. Journal of Neuroscience Methods, 228, 15–26. https://doi.org/10.1016/j.jneumeth.2014.02.019
• Pati, R., Pujari, A.K., Gahan, P., and Kumar, V., (2021). Independent Component Analysis: A Review, with Emphasis on Commonly used Algorithms and Contrast Function. Computación y Sistemas, 25(1). doi: 10.13053/CyS-25-1-3449
• Prakash, K., and Hepzibha Rani, D., (2015, January). Blind Source Separation for Speech Music and Speech Mixtures. International Journal of Computer Applications. 110, 40-43. DOI=10.5120/19372-1087
• URL-1: Moore, B., PCA and ICA Package, MATLAB Central File Exchange. (https://www.mathworks.com/matlabcentral/fileexchange/38300-pca-and-ica-package), (Erişim tarihi 23 Nisan 2024).
• Wang, R., (2021). Blind Source Separation Based on Adaptive Artificial Bee Colony Optimization and Kurtosis. Circuits System and Signal Processing, 40, 3338–3354. https://doi.org/10.1007/s00034-020-01621-5
• Yang, X. S. (2008). Nature-Inspired Metaheuristic Algorithms. United Kingdom: University of Cambridge, Luniver Press. ISBN 978-1-905986-10-1.
• Zi J., Lv D., Liu J., Huang X., Yao W., Gao M., Xi R., and Zhang Y. (2022). Improved Swarm Intelligent Blind Source Separation Based on Signal Cross-Correlation, Sensors, 22(1),118. https://doi.org/10.3390/s22010118