Closed-form blind mimo channel estimation for OSTBCS: Resolving ambiguities in rotatable codes

In this paper, the problem of blind subspace-based channel estimation in multiple-input multiple-output (MIMO) systems under orthogonal space-time block coded (OSTBC) transmission is investigated. We introduce a virtual snapshot model in which the redundancies in the OSTBC are exploited to augment the received data. We show that the vector of true channel parameters is scaled version of the normalized principal eigenvector of the associated augmented data covariance matrix, which in the case of rotatable OSTBCs is not unique. We propose a simple weighting of the different virtual snapshots in the computation of a modified covariance matrix and derive general conditions that guarantee uniqueness of the channel estimates from the principal eigenvector of that matrix. Further, we prove that the blind estimation schemes of [8] and [9] can be viewed as a particular examples satisfying these uniqueness conditions. In previous works, the uniqueness of these schemes has only been concluded from simulation results but it has not been proven analytically before.