Please use this identifier to cite or link to this item:
Title: Efficient Blind Source Separation for Next Generation Wireless Networks
Authors: Zahooruddin
Keywords: Applied Sciences
Engineering & allied operations
Other branches of engineering
Chemical engineering
Issue Date: 2016
Publisher: COMSATS Institute of Information Technology Islamabad- Pakistan
Abstract: Efficient Blind Source Separation for Next Generation Wireless Networks Independent component analysis (ICA) is a signal processing technique for separating statistically independent and non-Gaussian mixed source signals. It has its applications in different areas e.g., wireless communication, speech and biomedical signal processing, vibration analysis, and machinery fault diagnosis. In wireless co well as in quasi static wireless channels, even for smaller data block lengths. Simulation results show that the proposed transceiver system improves the un-mixing performance of the batch ICA algorithms in highly time varying channels. Then, we propose a generalized framework called the modified Infomax algorithm that improves the separation performance of the batch ICA algorithms for reduced lengths of the transmitted data blocks in wireless MIMO systems. Finally, we propose a hardware design of the mixing model for the ICA algorithms. The proposed model represents unity mixing, scaled mixing and ill-conditioned mixing. The proposed model may serve as a test bench for evaluating the performance of the ICA algorithms. Simulation results obtained are such that the unity mixing provides excellent performance of the batch ICA algorithms, while the scaled mixing provides good performance and the ill-conditioned mixing gives worse performance of the algorithms
Appears in Collections:PhD Thesis of All Public / Private Sector Universities / DAIs.

Files in This Item:
File Description SizeFormat 
Zahooruddin_Electrical_Engineering_2017_HSR_CIIT_27.04.2017.pdfComplete Thesis1.58 MBAdobe PDFView/Open

Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.