Advisor(s)
Abstract(s)
This paper presents experimental results of coherent communications comparing the following methods of underwater channel identification applied to a time-reversal processor: pulse compression, L1-norm regularization and channel physical modeling. The first method is the Passive Time Reversal in its conventional form: a low complexity technique to mitigate inter-symbol interference due to multipath propagation; the second method is an estimator of sparse channels, inspired by the theory of compressed sensing; and the third method is the proposed environmental-based approach that generates channel replicas through inverse numerical modelling using the physical properties of the underwater media between the emitter and the receiver. Data from a quadrature-phase-shift-keyed low resolution image transmitted on May 27, 2011 during the Underwater Acoustic Network 2011 experiment conducted off the coast of Throndheim (Norway), are processed using these three methods and their mean square error performance is compared. The results show that the proposed environmental-based approach outperforms the other two channel identification methods by one to four dB over the duration of the transmitted image packet.
Description
Keywords
Source localization