Independent Component Analysis Algorithms in Wireless Communication Systems

Independent Component Analysis Algorithms in Wireless Communication Systems

Sargam Parmar (Ganpat University, India) and Bhuvan Unhelkar (MethodScience.com & University of Western SydneyMethodScience.com & University of Western Sydney, Australia)
DOI: 10.4018/978-1-60566-156-8.ch043
OnDemand PDF Download:
$30.00
List Price: $37.50

Abstract

In commercial cellular networks, like the systems based on direct sequence code division multiple access (DSCDMA), many types of interferences can appear, starting from multi-user interference inside each sector in a cell to interoperator interference. Also unintentional jamming can be present due to co-existing systems at the same band, whereas intentional jamming arises mainly in military applications. Independent Component Analysis (ICA) use as an advanced pre-processing tool for blind suppression of interfering signals in direct sequence spread spectrum communication systems utilizing antenna arrays. The role of ICA is to provide an interference-mitigated signal to the conventional detection. Several ICA algorithms exist for performing Blind Source Separation (BSS). ICA has been used to extract interference signals, but very less literature is available on the performance, that is, how does it behave in communication environment? This needs an evaluation of its performance in communication environment. This chapter evaluates the performance of some major ICA algorithms like Bell and Sejnowski’s infomax algorithm, Cardoso’s Joint Approximate Diagonalization of Eigen matrices (JADE), Pearson-ICA, and Comon’s algorithm in a communication blind source separation problem. Independent signals representing Sub-Gaussian, Super-Gaussian, and mix users, are generated and then mixed linearly to simulate communication signals. Separation performance of ICA algorithms is measured by performance index.
Chapter Preview
Top

Ica Algorithms

Consider the classical ICA model with instantaneous mixing

x = As + n(1)

where the sources s = [s1,s2,…, sn]T are mutually independent random variables and Anxn is an unknown invertible mixing matrix and noise n = [n1,n2,…, nn]T . The goal is to find only from observations, x, a matrix W such that the output

y = Wx(2)

is an estimate of the possible scaled and permutated source vectors.

Several algorithms exits for blind source separation. This chapter describes the performance of some major ICA algorithms. This section presents a brief description of the respective approaches of the compared ICA algorithms.

Key Terms in this Chapter

A: Mixing matrix

PCA: Principal Component Analysis

W: Demixing matrix

s: Source signal vector

MAI: Multiple Access Interference

TDMA: Time Division Multiple Access

p: Number of observation

CDMA: Code Division Multiple Access

pdf: Probability density function

n: Noise vector

xi: ith sensor output

q: Number of sources

ICA: Independent Component Analysis

ni: ith noise signal

BSS: Blind Source Separation

x: Sensor signal vector

ISI: Inter-Symbol Interference

si: ith source signal

y: Separator output vector

Complete Chapter List

Search this Book:
Reset