Analysis of Two-Step Subspace-Based Channel Estimation Method for OFDM Systems

Blind subspace-based channel estimation (SS-CE) methods have been widely discussed in orthogonal frequency division multiplexing (OFDM) systems since the spectral efficiency can be improved by reducing the number of pilot or training symbols. However, the signal matrix, constructed from transmitted signals, applied in SS-CE must be full rank to make SS-CE method successfully estimate the channel impulse response (CIR). Our two-step signal matrix construction (TSSMC) method applied in SS-CE has been proposed in VTC-Spring 2015 and only the performance under static channels was shown. In this paper, the effects of splitting factor G for proposed TSSMC method are analyzed and the suggestion to choose good value of splitting factor G is also presented. Additionally, the probability of full rank analysis with different construction parameters, necessary condition analysis, and computational complexity analysis are discussed. Finally, simulation results conclude that the SS-CE with proposed TSSMC method outperforms related SS-CE in terms of bit error rate (BER) performance even under time-varying channel environments, where the number of available OFDM symbols is few under these channels.