Application of an improved LS channel estimation algorithm to sparse multipath underwater acoustic channel

In order to eliminate the effect brought from the noise in underwater orthogonal frequency division multiplexing (OFDM) system, an improved least-square (LS) channel estimation method is proposed for sparse multi-path channel estimation. The conventional discrete Fourier transform (DFT)-based interpolation channel estimation method is improved by using threshold detecting channel estimation, which can decide significant channel taps and erase the effect of noise by comparing the time domain channel estimation value with the predetermined threshold. The proposed method increases estimation precision effectively. Simulation results demonstrate that the DFT-based interpolation is practical in sparse multipath underwater acoustic channel, and the bit error rate (BER) becomes better when the circular prefix approximate actual channel length gradually. Moreover, the threshold coefficient regions of this method for two modulation modes are ascertained. The results show that BER of the proposed LS channel estimation is better than that of the conventional DFT-based channel estimations by about 2 dB, therefore the problem of using circular prefix to substitute actual channel length, which exists in the conventional DFT-based channel estimation method, is solved.