Skip to Main Content
Channel estimation based on comb-type pilot pattern has a good performance in frequency-selective channels with the help of MMSE algorithm. But MMSE is too complex for practical implementation and its performance deteriorates rapidly in time-varying channels due to extremely Doppler effects. In this paper, we present an optimized Kalman filter algorithm, which performs iteration operations within separate symbols. Optimized pilot matrices have been posted, where pilots in the special sub-carrier were inserted into rectangle pilot matrices. The modified Kalman filter estimation in the special sub-carrier refines estimation in other pilot sub-carriers. We also propose a low-pass-filter interpolation and FFT transform in time domain and frequency domain respectively. This three-step solution offers a better performance than that of MMSE algorithm. Compared with conventional Kalman estimator the proposed estimator has higher spectral efficiency, while its performance is comparable.