MIMO-OFDM pilot symbol design for sparse channel estimation

Channel equalization is a crucial part of the OFDM communications protocols, which in turn requires channel estimation. In this paper, we consider the problem of orthogonal pilot design in MIMO-OFDM systems for sparse channel estimation. The pilot design in MIMO scenarios compared to the conventional SISO case has the additional constraint that the capability of recovery should be uniformly provided for all single channels. For instance, perfect estimation of a channel at the cost of another one is not permitted. This requirement is even more significant in the emerging Massive-MIMO systems. Our pilot design is based on the compressed sensing technique of minimizing the coherence measure of the Fourier submatrix associated with the pilot subcarriers. However, we desire to minimize the coherence value of such matrices simultaneously for all channels. Here, we optimize over both the pilot locations and values. As finding the global minimizer is a combinatorial problem, we resort to a greedy method. Although there is no guarantee for the optimality of the achieved patterns, our simulation results confirm their suitability in practice.