Optimal training design and placement for MIMO orthogonal frequency division multiplexing channel estimation

Abstract

This paper deals with optimal training design and placement over multiple orthogonal frequency division multiplexing (OFDM) symbols for the least squares (LS) channel estimation in multiple-input multipleoutput (MIMO) OFDM systems. First, the optimal pilot sequences over multiple OFDM symbols are derived by co-cyclic Jacket matrices based on the minimum mean square error (MSE) of the LS channel estimation. Then, an enhanced channel estimation method using sliding window is proposed to improve further the performance for the optimal pilot sequences in fast-varying channels. Simulation results show that the enhanced method can efficiently improve the performances for the optimal pilot sequences over two and four OFDM symbols, especially in fast-varying channels.