Semiblind Channel Estimation and Symbol Detection for Block Transmission Using Superimposed Training

In this paper, we proposed a semi blind channel estimation and symbol detection approach based on superimposed training (ST) sequence. At the transmitter, a periodic training sequence is arithmetically added to the information sequence in a lower power. A complex Gaussian random Rayleigh frequency-selective channel is used for simulation. At the receiver, we exploit a two-step semi blind iteration approach. In the first step, only the first-order statistics of the received signal is used to initially estimate the channel. Then in the second step, a kernel weighted least squares superimposed training (KWLSST) iteration approach with a maximum-likelihood sequence estimation (MLSE) equalizer is used to iteratively estimate the single-input single-output (SISO) channel and detect the information symbols sequentially. Simulation results present the normalized channel mean squares error (NCMSE) and symbol error rate (SER) of both the KWLSST approach and the data-dependent superimposed training (DDST) approach in QPSK modulation with MLSE equalizer. The simulation results show that the KWLSST approach outperforms DDST approach, but at an expense of higher computational complexity.