|
Yuanning Yu
Advisor: Athina P. Petropulu, Ph.D.
Abstract:
Blind identification of a Linear Time Invariant (LTI) Multiple-Input
Multiple-Output (MIMO) system is of great importance in many applications,
such as speech processing, multi-access communication, multi-sensor
sonar/radar systems, and biomedical applications. The objective
of blind identification for a MIMO system is to identify an unknown
system, driven by Ni unobservable inputs, based on the No system
outputs.
We first present a novel blind approach for the identification
of a over-determined (No greater than or equal to Ni ) MIMO
system driven by white, mutually independent unobservable inputs.
Samples of the system frequency response are obtained based on
Parallel Factorization (PARAFAC) of three- or four-way tensors
constructed respectively based on third- or fourth-order cross-spectra
of the system outputs. We show that the information available
in the higher-order spectra allows for the system response to
be identified up to a constant scaling and permutation ambiguities
and a linear phase ambiguity. Important features of the proposed
approaches are that they do not require channel length information,
need no phase unwrapping, and unlike the majority of existing
methods, need no pre-whitening of the system outputs.
While several methods have been proposed to blindly identify
over-determined convolutive MIMO systems, very scarce results
exist for under-determined (No < Ni
) case, all of which refer to systems that either have some special
structure, or special No, Ni
values. We propose a novel approach for blind identification of
a general size under-determined convolutive MIMO systems. As long
as min(No, Ni)
greater than or equal to 2, we can always find the appropriate
order of statistics that guarantees identifiability of the system
response within trivial ambiguities. We provide the description
of the class of identifiable MIMO systems for a certain order
of statistics K, and an algorithm to reach the solution.
Finally we propose a novel approach for blind identification
and symbol recovery of a distributed antenna system with multiple
carrier-frequency offsets (CFO), arising due to mismatch between
the oscillators of transmitters and receivers. The received baseband
signal is over-sampled, and its polyphase components are used
to formulate a virtual MIMO problem. By applying blind MIMO system
estimation techniques, the system response is estimated and used
to subsequently decouple the users and transform the multiple
CFOs estimation problem into a set of independent single CFO estimation
problems.
Friday, July 6th, 2007 at 9 a.m.
Bossone 303
|
