Effect of Finite Register Length on Bacterial Foraging Optimization based ICA and Constrained Genetic Algorithm based ICA Algorithm

D. P. Acharya, G. Panda, Y. V. S. Lakshmi
2008 2008 International Conference on Signal Processing, Communications and Networking  
Independent Component Analysis (ICA) specifications [11]. Such an analysis with fast ICA technique separates mixed signals blindly without any and algebraic ICA algorithms has been carried out in information of the mixing system. Bacterial Foraging [12] Optimization based ICA (BFOICA) and Constrained [ 12 Genetic Algorithm based ICA (CGAICA) are two recently Bacterial Foraging based Independent Component developed derivative free evolutionary computational ICA Analysis (BFOICA) [13] and
more » ... A) [13] and Constrained Genetic techniques. In BFOICA the foraging behavior of E.coli Algorithm based Independent Component Analysis bacteria present in our intestine is mimicked for (CGAICA) [14] are two recently developed derivative evaluation of independent components (IC) where as in free evolutionary computational ICA techniques. Both CGAICA Genetic Algorithm is used for IC estimation in a constrained manner. The present work evaluates the error BFO and GA being population search based performance of BFOICA and CGAICA algorithm for its optimization techniques, they have several fixed-point implementation. Simulation study is carried on commonalities. However, BFOICA is reported to have both fixed and floating point ICA algorithms. It is faster convergence as compared to CGAICA [13]. observed that the word length greatly influences the . separation performance. A comparison of fixed-point Therefree it S qute movatng to study the effect of error performance of both the algorithms is also carried finite register length implementation of both the out in this work. algorithms. The present work focuses on the fixed-point
doi:10.1109/icscn.2008.4447197 fatcat:tbs4pkwsinearlplmhblrsc2ge