The wideband underwater acoustic multipath channels are usually modeled as a multi-scale multi-lag(MSML)channel.Different paths possess different Doppler scales which brings great challenge to channel parameter estimation.In this paper

we propose a novel algorithm for parameter estimation of MSML channels

which is called modified particle swarm optimization(MPSO)algorithm.This algorithm creates a multi-path list to record positions and fitness values of particles which are local best(

lbests

)

and uses these

lbests

to update particles' velocities at each iteration.At the end of the algorithm

the Doppler scale

the time delay and the amplitude can be estimated simultaneously for each path according to the multi-path list.Simulation results show that

the MPSO algorithm ou

tperforms both the matching pursuit(MP)algorithm and the fractional Fourier transform(FrFT)algorithm in estimation accuracy.