Abstract: Cable tunnels are completely satellite-rejected, and the electromagnetic channel state inside is complex. Therefore, traditional positioning methods are facing problems, such as declining positioning accuracy and increasing positioning variance. A high-precision personnel positioning in cable tunnels is proposed on the basis of channel model analysis and distance measurement by using the near ultrasound method. First, multiple sets of near ultrasound base stations are established in the tunnel, and their deployment is optimized through the finite element analysis method. Second, a statistical parameter model is established at the signal receiving side for dynamic changing of channel states due to target movement to estimate signal parameters quickly. Meanwhile, the received signal is also reconstructed in the fractional Fourier transform domain, and the moving speed of the target is estimated through the channel frequency shift state. Finally, the arrival time and moving speed estimated by the channel model are applied to track person position based on the Kalman filter. Simultaneously, the fusion arrival time estimation result is fed back to the channel parameter estimation of the next stage to correct the actual reference signal. Actual experimental results in the underground cable tunnel verify the effectiveness of the proposed method, which outperforms the traditional UWB positioning method.