Far-end crosstalk identification method based on channel training sequences

Far-end crosstalk identification results to improved frequency spectrum utilization in asymmetric digital subscriber line (ADSL) systems. The accurate determination of the crosstalk coupling function is a demanding task that is affected by various system impairments. In this paper, we present a new real-time method that exploits the exchange of signaling information between the modems of a new activated ADSL line in order to determine the crosstalk function between this line and any other active line in the binder. Each remote modem subtracts the decoded data from the received signal, decreases the noise effect of the previously identified crosstalkers, is synchronized to the disturber's timing and then applies a least-squares estimator for identifying the new crosstalk coupling function. The method is used along with a maximum margin bit-loading algorithm in order to provide the best possible estimation results. The estimated crosstalk coupling functions, the transfer function of each line and the noise measurements can be exploited by a centralized bit-loading algorithm for maximizing the total system performance in terms of maximum achievable data rate and/or bit error rate.