Abstract: The traditional Kalman filter algorithm requires that the noise model conforms to the Gauss distribution. In UWB indoor positioning, the observation noise is not only white noise, but also colored noise. Particle filter can deal with the problem of colored noise. The accuracy of particle filter for target tracking is improved by adding adaptive adjustment of likelihood distribution. The advantages and differences of adaptive adjustment of likelihood distribution particle filter and extended Kalman filter in UWB under white noise and colored noise are also studied. The experimental results show that when the observation noise is white, the extended Kalman filter and particle filter can achieve better pedestrian location and tracking; when the observation noise is colored, the adaptive particle filter is better than particle filter and extended Kalman filter.

Key words: colored noise, Kalman filter, particle filter, likelihood distribution adaptive, UWB