Blind estimation of MIMO channels with an upper bound for channel orders

Many known second-order statistics (SOS)-based blind algorithms for MIMO channel estimation, including the subspace (SS) and linear prediction (LP) method, are sensitive to channel-order overestimations. To overcome this problem, an algorithm is proposed in [IEEE Trans. Signal processing 50(6) (2002) 1449–1458] for the SIMO system only, and then its simple generalization to the MIMO system is presented in [Signal Processing 84(2) (2004) 435–439]. In this paper, improvements and refinements on the algorithm in [Signal Processing 84(2) (2004) 435–439] are given, which makes the method robust to noise and round-off error. The method can give estimations of all channel impulse responses subject to a matrix ambiguity when only an upper bound for all MIMO channel orders is known. Simulations show that the algorithm is effective and robust.