Semi-blind channel identification for individual data bursts in GSM wireless systems

We investigate the application of semi-blind equalization principle in prac tical wireless cellular systems. Specifically, we focus on the popular glob al system for mobile (GSM) cellular receiver in an effort to improve system efficiency without compromising performance. We begin by briefly introduci ng a linear quadratic amplitude modulation (QAM) approximation for the Gaus sian minimum shift keying (GMSK) signal used in GSM, which can be modeled a s a single input two output discrete linear system model after sampling and de-rotation (Ding and Li, IEEE J. Selected Areas Commun. 16 (October 1998) 1493-1505). We develop two different semi-blind algorithms based on the li near system model by combining both second-order statistical (SOS) informat ion and information from training. These semiblind methods can overcome som e serious limitations of SOS blind algorithms. We also analyze the semi-bli nd identifiability conditions. Finally, we compare simulation results for G SM receivers using both pure training and semiblind channel identification methods. (C) 2000 Elsevier Science B.V. Ail rights reserved.