Abstract: The improvement of the strategic position of the Arctic region has promoted the rapid development of various applications in the polar region, and the under-ice acoustic communication becomes one of the hot issues. In order to study the performance of single-carrier acoustic communication under ice, an under-ice test was conducted in the Yellow Sea ice area. The multi-channel decision feedback equalization (M-DFE) technique with embedded phaselocked loop is mainly concerned with the ability to resist multipath delays in under-ice acoustic communication, and the equalization performances of the improved proportional normalized least mean square (IPNLMS) algorithm and the recursive least squares (RLS) algorithm are studied and compared. The experimental results show that both IPNLMS and RLS algorithms can overcome the impact of channel multipath effects. RLS algorithm has a better equalization effect at the expense of computational complexity, and M-DFE technique can effectively eliminate intersymbol interference in the under-ice environment.