Using orthogonal least squares identification for adaptive nonlinear filtering of GSM signals

J.-P. Costa, T. Pitarque, E. Thierry
1997 IEEE International Conference on Acoustics, Speech, and Signal Processing  
The miniaturization of GSM handsets creates nonlinear acoustical echoes between the microphone and the loudspeaker when signal level is high. Nonlinear adaptive ltering can tackle this problem but the computational complexity has to be reduced by restricting the number of coecients introduced by nonlinear models. This paper compares performances of dierent nonlinear models. In a rst training stage we use the OLS (Orthogonal Least Squares) identication method to nd models using the fewest
more » ... g the fewest coecients along with a good tting accuracy. In a second ltering stage these parsimonious models are used to adaptively lter the GSM signals.
doi:10.1109/icassp.1997.599537 dblp:conf/icassp/CostaPT97 fatcat:5zuqcyydzjbv7iowbwqwgwo2xa