Threshold-based ML Channel Estimation for OFDM System in Sparse Wireless Channel

A joint maximum likelihood (ML) estimator, computing both channel impulse response (CIR) and noise variance , is proposed . Then, an adaptive threshold, defined as a linear function of the square root of noise variance, is introduced into this estimator. It can effectively filter channel noise over those weaker paths of the estimated CIR such that the entire performance of channel estimator can be further improved. This new ML channel estimator with threshold is called as the improved ML (IML) channel estimator. The simulated results in high and medium frequency channels show that the IML estimator obtains 1.5-2 dB SNR improvement over traditional ML for realizing the same bit error ratio (BER, <0.1), and achieves approximately the same BER performance as linear minimum mean square error (LMMSE) by using the lowest computational amount. This makes it very attractive for OFDM system in sparse wireless channel.