Adaptive recursive LK transversal algorithms for data echo cancellation applications

A family of adaptive transversal algorithms for data echo cancellation applications in full-duplex data transmissions is described. This family of algorithms (RLK algorithms) is a generalization of the classical recursive least square (RLS) algorithm, where K=1. The convergence behavior of the RLK algorithms is analyzed, and it displays a strong dependence on the noise statistics. For certain noise statistics, the RLK algorithms (K>1) will yield an improvement over the RLS algorithm (lower excess mean square error for the same amount of input data). In a computer simulation, the RLK algorithms (K>1) are applied to data echo cancellation, and an improvement in cancellation performance is achieved