Please use this identifier to cite or link to this item:
Title: Signal processing for acoustic arrays and for single-carrier block-based transmission
Authors: Song, Yang
Degree: Ph.D.
Issue Date: 2013
Abstract: The dissertation has six contributions towards space-time signal processing for acoustic sensor arrays which are summarized below. (1) Closed-form direction finding using collocated but orthogonally oriented higher-order acoustic sensors. This work introduces new closed-form formulas to estimate an incident source's azimuth-elevation angle-of-arrival (AOA), for various combinations of higher-order directional acoustic sensors, that are orthogonally oriented in a collocated triad. (2) Azimuth-elevation direction finding using a microphone and three orthogonal velocity sensors as a non-collocated subarray. An acoustic vector-sensor consists of three identical but orthogonally oriented acoustic particle-velocity sensors, plus a pressure sensor - all spatially collocated in a point-like geometry. This collocation constriction is relaxed in this work, to realize a spatially distributed acoustic vector-sensor, allowing its four component-sensors to be separately located. (3) Acoustic direction finding using a spatially spread tri-axial velocity sensor. This work shows how a triad of orthogonally oriented uni-axial velocity sensors may be spatially separated, yet facilitates direction finding of incident emitters via closed-form subspace-based parameter estimation algorithms, while extending the triads spatial aperture in three-dimensional space to enhance the resolution of the azimuth/elevation direction of arrival estimates. (4) "Blind" calibration of an array of acoustic vector-sensors suffering gain errors / mis-location/ mis-orientation. A series of direction-finding algorithms have recently been advanced, deploying a multi-array network (MAN) of acoustic-vector-sensors, each consisting of three collocated but diversely oriented uni-axial particle-velocity sensors, plus an optional pressure sensor. All these algorithms presume the particle-velocity sensors and the pressure sensors of ideal gain/phase responses, correct orientations, and (in some algorithms) precise locations. Such perfection is seldom (if ever) achieved in realworld systems or in field deployment. Indeed, these non-idealities need to be calibrated, often blindly without any training signal from any prior known arrival-angle. Towards this end, this work will advance a new "blind" calibration algorithm (a.k.a. "auto-calibration", "self-calibration",or "unaided calibration") that is computationally orders-of-magnitude more efficient than maximum-likelihood estimation. These advantages are achieved hereby exploiting the acoustic vector-sensor's quintessential characters, to interplay between two complementary approaches of direction-finding: (1) customary interferometry between vector-sensors, and (2) "acoustic particle-velocity-field normalization" DOA-estimation within each individual vector-sensor. (5) A lower bound of direction-of-arrival estimation for an acoustic vector sensor subject to sensor breakdown. In an acoustic vector-sensor, any particular velocity sensor must either function or fail, over the entire time-window when measurements are collected. This work derives an approximate lower bound for the error-variance of direction-finding using a single acoustic vector sensor subject to random breakdown in its sensors. (6) Three dimensional localization of a near-field emitter of unknown spectrum using an acoustic vector sensor. The work develops a parameter estimation algorithm to estimate a near field wide-band emitters azimuth elevation direction of arrival plus radial distance, based on data collected by one acoustic vector sensor. This new algorithm needs no prior knowledge of the incident sources spectrum.
The dissertation has two contributions towards signal processing for single-carrier block-based transmission/reception which are summarized below. (1) A precoder/two-stage equalizer for block-based single-carrier transmission with an insufficient guard-interval. To reduce the cyclic-prefix overhead in block-based cyclically prefixed single-carrier modulation, herein proposed is a zeros-inserting precoder that can reduce the net overhead in symbols. This precoder allows the formation adata-group that contains only interference and noise but not the desired signal, thereby facilitating a sub-sequent "signal to interference-plus-noise" (SINR) maximizer. Also proposed is an accompanying two-stage linear equalizer, that may be pre-computed off-line. This equalizer's first stage is a linear minimum mean-square-error (LMMSE) based linear frequency-domain equalizer(FDE); the equalizer's second stage is its second stage is a SINR-maximizer in the time-domain. Monte Carlo simulations show the proposed scheme's capability to shorten the cyclic prefix (CP) more than the inserted number of zero-energy symbol-periods, thereby reducing the transmission's net overhead. (2) "Blind" reception-beamforming to null unknown interference for block-based single-carrier transmission with an insufficient guard interval. This work proposes a "blind" beamformer to spatially pass the signal-of-interest, but to spatially null any co-channel interference, any adjacent-channel interference, any out-of-system interference, and/or any spatio-temporally correlated additive noises. This proposed scheme's transmission uses zero-paddingin an insufficient guard interval, shorter than the temporally spreading channel’s order. This proposed scheme's receiver first undergoes frequency-domain equalization, to "clear" the data of the signal-of-interests energy during the nominally zero-padded guard interval, to facilitate subsequent "signal-to-interference-and-noise ratio" maximization in the spatial dimension to realize the aforementioned "blind" beamformer.
Subjects: Detectors.
Signal processing.
Sound-waves -- Transmission.
Hong Kong Polytechnic University -- Dissertations
Pages: 123 p. : ill. ; 30 cm.
Appears in Collections:Thesis

Show full item record

Page views

Last Week
Last month
Citations as of May 28, 2023

Google ScholarTM


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