(For the above problem, this paper aims to use the network (geometric) topology information to propose a novel IoT localization model that uses partial information to describe the overall distribution density information: Robust Partially Persistent Canonical Correlation Analysis Localization Model LE-RLPCCA. Compared with the experimental results of the existing similar typical methods in the real environment)