Browse

Estimation of Time-Varying Sparse Underwater Acoustic Communication Channel Parameters
시변 스파스 수중 음향 통신 채널 매개변수 추정

Cited 0 time in Web of Science Cited 0 time in Scopus
Authors
변성훈
Advisor
성우제
Major
공과대학 조선해양공학과
Issue Date
2012-08
Publisher
서울대학교 대학원
Keywords
Channel estimationarray signal processingmatching pursuitgreedy algorithmsparse estimationtime-varying multipath channel
Description
학위논문 (박사)-- 서울대학교 대학원 : 조선해양공학과, 2012. 8. 성우제.
Abstract
This dissertation addresses the problem of estimating the channel parameters of time-varying sparse underwater acoustic communication channels. A new method to estimate the channel parameters including arrival time delay, incidence angle, Doppler frequency, and complex amplitude of impinging wave components is presented.
The new method exploits the sparse structure of the wideband underwater acoustic communication channel and is based on the matching pursuit which iteratively identifies multipath components by projecting the target signal on the columns of dictionary which are hypothesized by the channel parameters. Because of the large dimension of the parameter space, the size of dictionary can be prohibitively large especially when the parameter range is oversampled for effective sparse approximation.
In order to prevent the dictionary from being too large, the parameter estimation is achieved in two stages which are the identification and the iterative estimation stages. In the identification stage, the initial parameter values are identified using a pre-computed dictionary of low coherence. In the next estimation stage, a coherent and redundant dictionary of the oversampled parameter range is constructed from the identified parameter values, and the channel parameters are estimated by projecting the residual signal onto the redundant dictionary. To reduce memory requirement and computational complexity caused by using the redundant dictionary, a space-alternating estimation scheme is introduced to separate the parameter search space. The space-alternating scheme limits the size of the redundant dictionary within practical extent and accordingly reduces the computational burden of the matrix-vector product required in the iteration.
The performance of the new method is evaluated via Monte Carlo simulation and real channel measurement data analysis.
The Monte Carlo simulation evaluates the resolution performance by resolving two paths of small parameter differences, and its result shows that the new method successfully decomposes multipath components whose parameter differences are merely a subfraction of the resolution limit of the classical correlation-based method.
It is also applied to the experimental data obtained in the large scale water tank which is capable of making a surface gravity wave with designated wave parameter. The channel parameters under the time-varying regular surface wave condition is analytically derived from a simple reflection constraint, and the channel parameters by the new channel estimation method are compared with those analytic solution showing that the estimation results are consistent with theoretical expectation.
Finally, it is applied to the real channel data of shallow water which were acquired at various transmitter-receiver ranges. The performance of the estimated channel parameters is evaluated indirectly via comparison of the channel characteristic functions which are the delay-Doppler-spread function, the angle-Doppler-spread function, and the power delay and angle profiles. The comparison result shows that the estimated channel parameters coincide well with the channel characteristic functions obtained by the matched filter and accordingly proves that the presented method gives consistent estimation result for the estimation of real channel parameters.
Language
English
URI
https://hdl.handle.net/10371/119994
Files in This Item:
Appears in Collections:
College of Engineering/Engineering Practice School (공과대학/대학원)Dept. of Naval Architecture and Ocean Engineering (조선해양공학과)Theses (Ph.D. / Sc.D._조선해양공학과)
  • mendeley

Items in S-Space are protected by copyright, with all rights reserved, unless otherwise indicated.

Browse