 This paper proposes an autonomous multi-floor localization framework based on smartphones and pedestrian network matching, MLSNM. It uses a robust data and model-driven pedestrian trajectory estimator to accurately locate users in complex and large-scale urban environments. Additionally, it combines the estimated trajectories with the pedestrian network matching algorithm to obtain more precise location and floor observations. Experiments show that the proposed MLSNM can achieve accurate and autonomous multi-floor positioning performance in complex and large-scale urban buildings. This article was authored by Cha Yongshu, Wenxin Teng, Yi Zheng, and others.