Improving time variant channel estimation for 3GPP LTE-downlink

Accurate estimation of the doubly selective mobile radio channels is essential to achieve satisfactory performance in wideband wireless communication systems like 3GPP LTE. Pilot Assisted Channel Estimation (PACE) techniques like Linear Minimum Mean Squared Error (LMMSE), Least Square (LS) etc. are used to estimate the channel states at a certain time-frequency grid specified by the pilot symbols. A model like Basis Expansion Model (BEM) is used to track the doubly selective channel in time between the pilot symbols. BEM performance is sensitive to the PACE results. Imperfections in the PACE results are inevitable due to Inter Carrier Interference (ICI) at high speeds, changing signal to noise ratio (SNR) of the system and mismatch in the parameter values used in PACE calculations. We are able to reduce the effect of such imperfections by shaping the noise at the high frequencies in extended Fixed CE-BEM. Simulations show a decrease in MSE by a factor of ≈2 when the LMMSE results, obtained at 0dB SNR, are used to track the doubly selective channel at 350 km/h.