Channel estimation and equalization for GSM with multiple antennas

We present a complete space-time demodulation scheme for the Global System for Mobile Communications (GSM) assuming two antennas at the receiver. The scheme includes multipath channel estimation using a blind method (Cross-Relation (CR) with Maximum In-band Energy (MIE) method), two semi-blind methods (combination of the training sequence with either the CR method alone or the CR with MIE method), and a method based on the training sequence alone. For the equalization we apply a zero-forcing equalizer, which equalizes only over the band of the Gaussian Minimum Shift Keying (GMSK) signal. The demodulation of the equalized GMSK signal is done by an implementation of the Viterbi-Algorithm employing decision metrics based on the amplitude of the equalized signal. The simulation results show that the semi-blind scheme which is based on the maximum in-band energy method is superior to the other three methods and therefore outperforms the approach of solely exploiting the training sequence. The success of these methods shows that it's not necessary to apply the linear approximation of the GMSK signal in the case of multiple antennas to effect blind and semi-blind channel identification. Thus, these multiple antenna based algorithms are applicable to non-linear modulation schemes as well as standard linear modulation schemes.