Blind Identifiability Analysis in a MIMO LTI System with Inputs from a Finite-Alphabet Set
| Authors: | Kwon Soonman, KERI, Korea, Republic of
Fuhrmann D. R., Washington University, United States
Kim Seog-Joo, KERI, Korea, Republic of
Lee Jongmoo, KERI, Korea, Republic of |
|---|
| Topic: | 1.1 Modelling, Identification & Signal Processing |
|---|
| Session: | Identification of Multivariable Systems |
|---|
| Keywords: | MIMO, Blind identification, Finite-Alphabet, Ambiguity |
|---|
Abstract
This paper considers a blind separation problem in amultiple-input-multiple-output (MIMO) linear time-invariant (LTI)system where the inputs to the system come from a finite-alphabetset. A discrete-time matrix equation model is used to describe theinput-output relation of the system in order to fully exploit theadvantages of modern digital signal processing techniques. Atfirst, ambiguity problem is investigated. Then, based on theresults of the investigation, a new identifiability condition isproposed for the case of a widely-used alphabet set. Thiscondition is compared to an existing condition in terms of theprobability of satisfaction of the condition for an arbitrarysignal set. A probability bound such that an arbitrary inputsignals satisfies the identifiability condition is derived. AlsoMonte-Carlo simulation is performed to demonstrate the validity ofthe
