Abstract: In multi-sensor networks，range-based localization is a common means of target localization；However，because of some interference such as the system’s equipment and environmental factors，it will have a range of data errors which affect localization accuracy.At present，there is no better way to eliminate error；Certain specific algorithms can only be applied to minimize the error of localization accuracy.This paper presents a measurement error based on the statistical characteristics of obedience probability algorithm，first according to all the location data，fitting-out the initial estimates point，then take the initial estimates point as the center position of the selected regional rectangular grid，each a small grid on behalf of the center of the small grid，and through the perception of all the nodes 3-3 portfolio Statistics small mesh weights，weights to the largest point as a target point.Through the MATLAB software programming and simulation results show that the algorithm can effectively improve positioning accuracy.