These days, most of the smartphones come with integrated MEMS (Microelectromechanicalsystems) sensors like accelerometer, magnetometer, gyros etc. It has opened the ways to use them for location based applications. The big advantage of these sensor based navigation system is that they are environment independent in contrast to other existing positioning technologies. A lot of work is already done on fixed position sensor based systems where sensors are either attached to foot or belt. This work is focussed on developing a smartphone based pedestrian dead reckoning (PDR) system. The big issue with smartphone’s sensor based PDR system is that the position of mobile is not deterministic in contrast to fixed position inertial measuring unit. In this work, three different modes of smartphone placement (Idle, Handheld and Listening) are investigated and accelerometer and magnetic sensors are used for step detection and heading determination respectively. Various step detection and stride length estimation methods are implemented and a comparison is given at the end.