Abstract:  As a potentially effective method to mitigate inter symbol interference caused by multi-path, channel
equalization of underwater acoustic communication has attracted considerable attention. Among existing algorithms that can
be found in the literature,the classic least mean square(LMS)and various variants of it arc of particular interest for practical
implementation due to their low computational complexity. However, as the variable step size as well as the parallel filter
bank structure can improve the performance of LMS type algorithms under time varying environment,there is a lack of in-
vestigation on their adaptability to the dynamic order of underwater acoustic channels, In this paper,a new dual-parameter,
adjustable method is presented which embeds the variable step size and filter length into the parallel filter bank LMS algo-
rithm for equalizztion of time varying underwater acoustic channel.The mechanism of dual parameter(step size and filter
length)adjustment ensures that the proposed algorithm has better tolerance upon the time variations caused by either specific  coefficients or the order of the channel response. Both numerical simulations and real data experiments show that the per- formance of the new method outperforms the classic methods.