Efficient subpace based channel estimation for time variant OFDM systems

Subspace based time domain channel estimation can separate signal subspace and noise subspace using channel statistic information to reduce noise with low complexity. In this paper, an improved threshold is deduced based on MSE criterion, which average the successive (orthogonal frequency division multiplexing) OFDM symbols in time domain efficiently reducing the addictive white Gaussian noise (AWON) and Inter-carrier Interference (ICI). Using the statistic feature of Gaussian noise, the Gaussian probability threshold (GPT) algorithm is proposed. To be adaptive with fading signals, robust Time OFDM symbol average adaptive threshold (TAAT) algorithm is proposed. The two algorithm proposed in this paper can achieve high resolution of signals without channel information. Also it is not subject to frequency leakage. Simulations under VA30 channel shows the algorithm can almost remove the entire noise path and achieve 2.5dB over the discrete Fourier transform(DFT)based method.