Second and Higher-Order Statistics based Multiple-Input-Multiple-Output System
Blind Identification Matlab Code Readme file

Communications and Signal Processing Laboratory
ECE Department, Drexel University
Philadelphia, PA 19104, USA
February, 2002
http://www.ece.drexel.edu/CSPL
----------------------------------------------------------------------------------

---- Introduction ----

This package contains the MATLAB code for the algorithm proposed in the following
papers. The file "mimo_main.m" is the implementation of the algorithm in the papers.
The goal of this Matlab script is to identify a MIMO system with white inputs based
on the second and higher-order statistics of the outputs only. This code will plot 
the blind estimation results using figures.

This packages also includes some supporting functions. 
"rec_frommag_complex.m" is system reconstruction methods from amplitude response; 
"true_r.m" computes the true-correlation of the system outputs for comparison purpose only; 
"g_cc3_matrix.m" is an efficient MATLAB code to compute the third-order cross cumulants of 
  three signals;
"g_cc4_3D.m" is an efficient MATLAB code to compute the fourth-order cross cumulants of 
  four signals, this code runs even faster than the correpsonding C code in PC/Windows NT 
  platform.
"joint_diag.m" is the MATLAB code written by Dr. Jean-Francois Cardoso for the implementation 
  of his "Joint Diagonalization" method, this code is proved to be able to improve the
  performace the system impulse response estimation greatly; 
"hosmatrix.m" is the MATLAB code to generate a matrix consists of only 1 and 0 for the
  phase estimation method proposed in [1].
"interpolate_func.m" is the MATLAB code to interpolate a periodic function, like the FFT.
  It takes the advantage of the fact that FFT is periodic. It is used to interpolate the
  estimated V(w) at certatin frequencies where the power spectrum matries have high
  condition numbers.
"shadow_plot.m" can plot the Monte-Carlo simulations results in a clear way where both the
  mean and standard deviation are shown on the plot.
"cum3equalizer.m" is the implementation of Tugnait's deconvolution algorithm based on
  higher-order statistics for single-input single-output case.

[1] Binning Chen and Athina P. Petropulu, "Frequency Domain Blind MIMO System Identification 
    Based On Second- And Higher-Order Statistics," IEEE Transactions on Signal Processing, 
    vol. 49(8), pp. 1677-1688, August 2001.
[2] Binning Chen, Athina P. Petropulu, Lieven De Lathauwer and Bart De Moor, "Blind MIMO 
    System Identification Based on Cross-Polyspectra," 2000 European Signal Processing 
    Conference - EUSIPCO'2000, September 2000, Tampere, Finland.
[3] Binning Chen and Athina P. Petropulu, "Multiple-Input-Multiple-Output Blind System 
    Identification Based on Cross-Polyspectra," IEEE International Conference on Acoustics, 
    Speech and Signal Processing - ICASSP'2000, June 2000, Istanbul, Turkey.
[4] Binning Chen and Athina P. Petropulu, "Blind MIMO System Identification Based on Cross-
    Polyspectra," 34th Annual Conf. on Information Sciences and Systems, CISS'2000, Princeton 
    University, March 2000.

----------------------------------------------------------------------------------

Last Updated: Sunday, February 10, 2002