Fast and precise indoor pedestrian positioning is essential for location-based services. Traditional inertial-based methods suffer from drift in long-term tracking, while visual navigation lacks accuracy in challenging environments. This paper proposes a robust LiDAR-inertial odometry (LIO) approach for accurate long-term pedestrian trajectory estimation. Our method incorporates pedestrian motion constraints and loop closure correction using small-size solid-state LiDAR and low-cost MEMS sensors. By leveraging foot-mounted MIMU with zero-velocity gait, we reduce the computational cost of loop closure detection during LiDAR point cloud keyframe construction. We also introduce a two-stage loop closure detection algorithm based on position and point cloud descriptor similarity to correct cumulative errors in the LIO. Experimental results demonstrate competitive positioning accuracy, with an error of 1.0 m observed during a pedestrian walking of 500 m.