Doctoral Dissertations

Abstract

"Acoustic echo cancellation (AEC) is a well studied problem. The underlying assumption in most echo cancellation solutions is that the echo path following the reference signal is completely linear. However, in many handheld devices, the echo path following the reference signal is nonlinear. The reason for this nonlinearity in the echo path is the use of smaller and cheaper loudspeakers. In order to cut the manufacturing cost, device manufactures use cheaper loudspeakers such that they satisfy the carrier specifications. Such loudspeaker can be easily over driven in to their nonlinear region and thus add nonlinearities to the downlink path. This brings about the need for a nonlinear echo canceler to maintain the required echo return loss enhancement (ERLE). Software-based solutions have been proposed to solve the nonlinear echo cancellation problem. The computational complexity of these solutions is prohibitively high for practical implementation. This dissertation analyzes the sources of nonlinearities in smartphones and proposes a simple and elegant hardware modification to significantly reduce nonlinear echo. Thorough analysis and intensive testing results show that up to 6 dB of improvement in ERLE in a real device is possible using the proposed technique.

Canceling the network echo requires the identification of network impulse response. This is a sparse system identification problem. Existing solutions do not completely exploit the sparsity of the network impulse response. An iterative method has been proposed to improve the convergence of network echo cancellation algorithms with special focus on exploiting sparsity. Zero attractor and gear shifting methods are used to further improve convergence and reduce the computational complexity of the proposed algorithms. Simulation results show faster convergence of the proposed algorithms. The computational complexity comparison of the proposed algorithm is comparable or lower than that of the existing algorithms"--Abstract, page iii.

Advisor(s)

Grant, Steven L.

Committee Member(s)

Sedigh, Sahra
Moss, Randy Hays, 1953-
Zheng, Y. Rosa
Hurson, A. R.

Department(s)

Electrical and Computer Engineering

Degree Name

Ph. D. in Electrical Engineering

Sponsor(s)

Research in Motion (Firm)

Publisher

Missouri University of Science and Technology

Publication Date

Spring 2013

Pagination

x, 85 pages

Note about bibliography

Includes bibliographical references (pages 79-84).

Rights

© 2013 Pratik Vijay Shah, All rights reserved.

Document Type

Dissertation - Open Access

File Type

text

Language

English

Subject Headings

Echo suppression (Telecommunication) -- Computer simulation
Smartphones -- Design
Signal processing -- Mathematical models

Thesis Number

T 10342

Print OCLC #

860983160

Electronic OCLC #

909374855

Download

Share

 
COinS