This paper proposes a novel de-noising algorithm using a smoothing window with adaptively varying length.After the signal to be processed is segmented according to its properties in time domain and wavelet transform domain and interpolated into a higher sampling rate

a median and linear filter is employed to estimate the noise and then the estimate is removed out from the signal.The algorithm is particularly suitable for de-noising of segment-wise continuous or segment-wise stationary signals.Computer simulation results suggest that the algorithm has very good performance for de-noising.The algorithm is applied to anti-jamming in a spread spectrum communication system

and computer simulation shows that its performance is much better than that of setting-pulse-to-zero method in anti-pulse-jamming.