Iterative B-spline estimator using superimposed training in doubly-selective fading channels

An iterative channel estimator combining B-spline approximation and superimposed training is proposed for estimating doubly-selective fading channels in a turbo receiver. An MMSE approach based on dividing the transmitted data block into sub-blocks is applied to calculate spline coefficients. The iterative channel estimation greatly improves the detection performance by feeding the output of the data decoder back to the estimator. Systems with Rake receiver and MMSE equalizer, using either convolutional or turbo code, are considered. The mean square error (MSE) and bit error rate (BER) performance of the iterative scheme is investigated by simulation. The simulation results show that the proposed scheme can provide accurate estimation of doubly-selective channels, and the BER performance close to the case of the perfect channel knowledge.