Adaptive Subspace Iterative Information Measure Directed Synchronization in MIMO Channel

The paper addresses the problem of iterative soft-information measure supported synchronization. We use a subspace based model for the arbitrarily time-variant MIMO channel. The channel subspace can have arbitrary dimension allowing to model also a fast time-variant channel. The equivalence between expectation-maximization based iterative synchronizer and weighted least squares problem is demonstrated to hold for general MIMO system. We introduce the idea of adaptive subspace control during the iterative synchronization iterations using the order recursive solver. This allows to use a coarse subspace (low dimension) at initial iterations while fine subspace (higher dimension) for later iterations. We show that the asymptotic properties of the estimate (MSE) are not affected. The reduction of the dimensionality of the estimator at the beginning of procedure lowers substantially the complexity