With increasing demand for indoor positioning, a lowcost inertial sensor positioning method based on PDR algorithm is favored. This paper proposed a sf relationship model on the basis of PDR algorithm . This model means step size may vary adaptively with step frequency changes. Because of PDR algorithm will inevitably produce accumulative error, this paper uses the absolute locations of WiFi signal access point to eliminate the accumulative error, while taking advantage of two adjacent WiFi signal access points to dynamically adjust the step size avoiding the influence of pedometer cumulative error. Experimental results show that this method can effectively improve indoor location accuracy.