In frequency-division duplex transceivers, the duplexer connects the transmitter and receiver to the common antenna. Due to non-idealities in the receiver front-end, a transmitter leakage signal is present in the receive path, where it overlies the wanted receive signal. To cancel this receiver interference, the main estimation task is the estimation of the transmitter-to-receiver leakage path. This is mainly determined by the duplexer stop-band response. Since the duplexer is time variant we must use an adaptive filter for the estimation task. The performance of adaptive filters is highly dependent on the used input signal. If the input signal is highly correlated the condition number and thereby the eigenvalue spread of the estimated autocorrelation matrix is high. This leads to a slow convergence of the Least-Mean-Square (LMS) adaptive filter. A possible way to improve the performance is the use of Transform-Domain (TD) or Subband-Adaptive-Filters (SAF). In this thesis different input signals and implemented algorithms have been tested and compared to each other with regard to estimation performance, convergence and computational efforts.