부산대학교 도서관

A decision-feedback equalizer based on zeros-order statistics is proposed in this paper to improve the performance and convenience of the method based on fraction low-order statistics which needs to identify characteristic exponent alpha. The new method uses an iterative factor based on zeros-order statistics to renew weight coefficients of adaptive equalizing filter. The coefficient factor can restrain effectively impulsive noise and need no the identification of characteristic exponent alpha. Transform functions involved are similar to logarithm, sigmoid, arctangent, etc. Simulations on decision-feedback equalizer applied in underwater acoustic channel show that the new arithmetic has better equalizing capability in impulsive noise surroundings.