Frequency sampling scheme for effective MIMO-OFDM channel estimation with SALSA and LS algorithms

In this paper we consider the channel estimation problem in MIMO-OFDM system, when the base station has a hybrid architecture of antenna array. Considering uplink CE, overhead reduction for pilot resources in time/frequency OFDM grid was studied. Utilizing properties of channel decomposition via sum of Kronecker product factors and applying Tucker format decomposition for every decomposition term, SALSA (Sequential Alternating Least Squares Approximation) algorithm was applied for partial channel tensor reconstruction on regularly measured sub-carrier frequencies. LS (Least Square) algorithm was used to reconstruct the whole channel tensor after partial channel tensor estimation at comb frequencies. Additionally, we investigated schemes where sub-carriers set is different for different OFDM symbols. It was shown that significant reduction of pilot sub-carriers can be achieved while keeping good accuracy of CE if positions of regular sampling patterns with large period are shifted in frequency during multiple consecutive symbols. Simulations show the effectiveness of the proposed method of channel estimation by SALSA and LS algorithms in comparison with fixed set of sub-carriers.