Training-sequence-based determination of optimum sampling time in unequalized TDMA mobile radio systems

In time-dispersive mobile radio channels, determination of the proper sampling instants of the receiver is one of the most difficult problems. A promising way is to oversample the received signal while a training sequence is transmitted, and to use this information to extract the optimum sampling instant; this approach has been termed "adaptive sampling." We analyze the influence of various metrics for the evaluation of the training sequence and find that even very simple implementations can achieve near-optimum performance. The non-idealities of the training sequence used in the (DECT) system have no significant impact on the performance if proper evaluation methods are used.