Motion detection based fine grained place extraction on mobile cellular phone

Location-based services (LBS) are proliferating in the field of mobile and pervasive computing. Normally, locations are expressed as coordinates or landmarks, which are geography related and hardly to be understood by average user. Recent researches pay much attention to extracting personally meaningful places where one works, lives or plays, using mobile phone as the platform. With the advantages of universal availability and low energy consumption, Global System for Mobile Communication (GSM) based method is more attractive than using GPS or Wi-Fi. However, because of large coverage and overlap of cells, existing methods can only extract coarse grained places and have a limit capability to discriminate the places closed with each other. This paper proposes a motion detection based fine grained place extraction method to discover meaningful places for individual user using GSM information. The proposed method can detect the user´s motion states through analyzing the fluctuation of GSM received signal strength (RSS) and divide the cell trace into moving segments and stationary segments. Each stationary segment is represented as a time-weighted cell ID vector, which can confine the places to a small area. Experimental results indicate that the proposed method has strong capability to discriminate places closed with each other and, compared with existing method, extract finer grained places.