Pilot design for compressed sensing based OFDM sparse channel estimation

In order to improve channel estimation performance, the pilot design problem in orthogonal frequency division multiplexing(OFDM) is investigated from the perspective of compressed sensing(CS).Since the reconstruction performance of the sampling matrix cannot be accurately measured by the existing methods, the pilot designed by the existing methods has poor channel estimation performance.Therefore, the cubic sum criterion which computes the cubic summation of entries of correlation matrix is proposed to measure the reconstruction performance of sampling matrix.Besides, for the pilot design of OFDM channel estimation which is a discrete combinatorial optimization problem, a novel pilot search method named grouped substitution with concurrent full trees is also proposed to search optimal pilot.At each iteration of the proposed algorithm, the pilot pattern set is divided into groups.Then, the pilot patterns are successively updated by obtained pilot sets.The proposed method enlarges the search space and avoids getting in local optimum in searching pilot pattern.The simulation results show that, the proposed evaluation method can accurately evaluate the reconstruction performance of the sampling matrix in comparison to the existing evaluation methods and compared with mutual coherence criterion, the proposed criterion can gain 3 dB improvement in mean square error.Furthermore, the proposed pilot search method has faster convergence speed and the best searching performance.