Efficient blind receiver design for orthogonal space-time block codes

We consider stochastic blind maximum-likelihood detection of orthogonal space-time block codes (OSTBCs) over a quasi-static flat multiple-input multiple-output (MIMO) Rayleigh fading channel. A general decision rule for stochastic blind maximum-likelihood OSTBC detection is derived. This rule is simplified using OSTBC linear dispersion matrices to realize a blind detector, which is implemented by semi-definite relaxation or sphere decoding. For the latter, the modifications necessary for both unitary and non-unitary constellations are developed. Two totally blind detectors using dual constellations or a superimposed training scheme are proposed. As a side product, two conditions for a rotatable OSTBC are also derived. A decision-directed, minimum mean-square-error (MMSE) channel estimator is developed. We also derive the Cramer-Rao bound (CRB) for channel estimation and discuss the optimal power allocation. Extensive simulation results are used to compare the different detectors in terms of complexity and performance.