how to implement moving window

Hi, I have a question regarding moving window implementation. I have this equation. how to implement/coding it in Matlab where m = window size; k=epoch ?

C = 1/m * (summation(i=1 to m) of V(k-i)*V(k-i)')

where V is the residual between observation and measurement size 3x1

Any advise? Thank you.

[Jan]

What did you try so far?

[dab483]

Hi, Actually i am implementing extended kalman filter. i am extending it to estimate QekfA(:,:,t) where have moving window as shown below in AAA. i have no idea how i should continue this..Need help to continue this.

 for t=2:T  

mu_ekfPred(:,t) = feval('ffunasp',mu_ekf(:,t-1),t);

    Ax_ekf = ....

    PPred_ekf(:,:,t) = QekfA(:,:,t-1) + Ax_ekf*P_ekf(:,:,t-1)*Ax_ekf';

    Cy_ekf = [0 0 1 0];

    yPred_ekf(:,t) = Cy_ekf*mu_ekfPred(:,t);

    M_ekf = Rekf + Cy_ekf*PPred_ekf(:,:,t)*Cy_ekf'; 
    K_ekf = PPred_ekf(:,:,t)*Cy_ekf'*(M_ekf^(-1));          
    mu_ekf(:,t) = mu_ekfPred(:,t) + K_ekf*(y(:,t)-yPred_ekf(:,t));
    P_ekf(:,:,t) =(eye(4)-K_ekf*Cy_ekf)*PPred_ekf(:,:,t)......

    QekfA(:,:,t)=QekfA(:,:,t-1)*sqrt(AAA)
end

where AAA = 1/m * (summation(i=1 to m) of V(t-i)*V(t-i)') - Rekf

where V(t-i) = residual = y(:,t)-yPred_ekf(:,t)