steps:
1, given channel response samples X with dimension 1*N: estimate the power coefficient matrix \Sigma_{i,j} via equations (19)-(25); (the code for this estimation is very trivial.)
2, given \Sigma_{i,j}, using matlab "function lsqcurvefit" to find the approximate Green Matrix, G, in the least-squares sense. (The lsqcurvefit function is suitable for the matrix computation and fitting. You can find some example by typing "help lsqcurvefit" in matlab command window.)
3, given the approximate Green Matrix, G, find its inverse by function inv(G)
4, using the elements of the inverse matrix of G, i.e.,"inv(G)", to calculate the EC function via equation (8)