In this paper, we derive a new subspace algorithm to consistently iden
tify stochastic state space models from given output data without form
ing the covariance matrix and using only semi-infinite block Hankel ma
trices. The algorithm is based on the concept of principal angles and
directions. We describe how they can be calculated with QR and Quotien
t Singular Value Decomposition. We also provide an interpretation of t
he principal directions as states of a non-steady state Kalman filter
bank.