Channel estimation and equalization based on implicit training in OFDM systems

Abstract

Implicit training (IT) based channel estimation exploits the first order statistics in the received data, induced by superimposing periodic training sequences with good correlation properties, along with information symbols. Hence the need for additional time slots for training the equalizer is avoided. In this paper we investigate on the applicability of this technique to orthogonal frequency division multiplexing (OFDM) systems over a frequency selective fading channel. Based on the estimate, a zero forcing solution is used to equalize the channel. The effect of the deterministic mean of the data on the channel estimate is considered. By exploiting the periodicity of the training sequences in the frequency domain, improved estimates of the channel coefficients are obtained. The performance of the estimator is presented in terms of the mean square estimation error (MSEE) and uncoded bit error rate (BER). All the improvements come at the cost of a loss in bandwidth which makes this scheme similar to the comb type pilot based channel estimation scheme for OFDM.