Training sequence design for adaptive equalization of multi-user systems
Giuseppe Caire, Urbashi Mitra · 2002
Preliminary work on training sequence design for block adaptive channel estimation methods for multi-user direct-sequence/code-division multiple-access (DS/CDMA) systems is provided. The focus is on linear minimum-mean squared error estimation methods. First, a joint channel estimation scheme is considered for hybrid quasi-synchronous CDMA/time-division multiple-access systems. For the CDMA/TDMA hybrid optimal training sequences are designed. The second scenario considers a completely asynchronous classical DS/CDMA system. Bayesian channel estimation is considered for a single user of interest. Based on estimated interference statistics, an optimal training sequence is designed. Numerical results are provided which indicate that significant improvements in performance can be achieved when employing optimized training sequences.