Abstract: Focusing on the instability of the actual signal in wireless sensor networks (WSNs), the simply algorithm improvement cannot meet the requirement of localization precision. An improved self-adaptive log-normal shadow model was proposed in distance measuring. Kalman filter and Markov process were adopted in coordinate estimation. The state equations were established and combined with the states of the measured coordinates to estimate the optimal values of coordinates. Finally, these two improved processes were introduced in the existing triangle localization algorithm. The experimental results show that the improved self-adaptive log-normal shadow model enhances the adaptability and accuracy of the distance-measuring model, and Kalman filter and Markov process reduce localization errors.