SMI (subspace model identification) toolbox

Does anyone know or have any reference for the SMI (subspace model identification) MatLab toolbox? I have read and understood the examples in it but I need something more for what I have to do.

I have understood how to build SISO discrete state space models by dividing data in batches. Now I would like to build SIMO (or MIMO) discrete state space models by concatenating the results of the SISO cases. What I thought was something like this

 [S1 R1] = dordpi(u1,Y1,10); 
 [S2 R2] = dordpi(u1,Y2,10,R1);
 [S3 R3] = dordpi(u1,Y3,10,R2);
 [Ae Ce] = domdpi(R3,3);
 [Be,De,R1bd] = dac2bd(Ae,Ce,u1,Y1);
 [Be De R2bd] = dac2bd(Ae,Ce,u1,Y2,R1bd);
 [Be De R3bd] = dac2bd(Ae,Ce,u1,Y3,R2bd);
 x0 = dinit(Ae,Be,Ce,De,u1,Y_est);         // initial state estimation
 Ye = dlsim(Ae,Be,Ce,De,u2,x0);            //simulating the model

Y1, Y2, Y3 are matrices whose columns are zeros except in the one of the channel considered so that the total matrix of measurements for the estimation part is

 Y_est = [Y1(:,1) Y2(:,2) Y3(:,3)];
u1 is the input vector for the estimation part, u2 is the input vector for the evaluation part.

[Fernando Ortolano]

Where I find the SMI tollbox?