Nlms-type System Identification of Miso Systems with Shifted Perfect Sequences

This paper addresses the fundamental problem of multi-channel system identification given the multiple input and single output signals of a MISO system. The presented approach is based on the normalized least mean square (NLMS) algorithm in combination with a set of special excitation signals, constructed from so called perfect sequences (PSEQs). The new excitation strategy opens up the possibility to uniquely identify the true impulse responses of multiple channels in a simple and efficient way from a single-time recording. The method can be applied to any number of channels or system lengths. Due to its fast tracking property, this new approach allows the real-time acquisition of time-variant impulse responses. These can be used, e.g., in simulations for the evaluation of stereophonic acoustic echo cancellation algorithms under real-world conditions or for the calibration of multi-channel loudspeaker systems. Furthermore, the method allows an identification of linear channels of any kind, e.g., radio or acoustic, and can easily be extended to MIMO (e.g. wireless) transmission.