Research On Radio Map Construction Algorithm Based On Smart Phone Sensors And Landmarks

With the popularization of location based service, indoor localization has been attracting more and more attentions in research field. Although the WLAN based localization performs better in location estimation, equipment cost and extra preparation than other techn iques, the huge consumption in building radio map remains a problem. Therefore, it is vital to find a radio map construction method which could both improve the building efficiency and ensure the expected localazation accuracy.In this paper, we propose d a fast radio map construction method based on smart phone equipped sensors a nd indoor landmarks, which construct the database by analyzing daily activit ies of the normal users. The key point of our method is to recon the traces of user by combing the data c ollected from sensors and the landmarks. Then the radio map will be established by adding the measured RSS(Received Signal Strength).As for the problem that tranditional PDR(Pedestrain Dead Reconing) performs poor when user changes his phone placement, we propose a step and heading estimation algorithm considering the mode changing situation. The algorithm first recognizes the walking mode of the user, and then estimates the steps respective ly. The heading of the user is estimated by combiming the changed attitude angle and landmarks. Results showed that our algorithm has a better performance in step and heading estimation in the mode changing situation, and no need to learn the initial user posit ion and phone attitude.In trandit ianal algorithms, landmarks are used to correct the user’s location. However, they are only useful when users have passed them. To solve the problem, we utilize these landmarks to estimate the user traces and to build radio map instead of simple location correction, which could reach a better localization result whether the user passes those landmarks. Radio map will be established by combining the RSS measurements after the traces are estiamted. In this paper, landmarks are classified into corner landmarks and magnetic la ndmarks, and are recognized by different algorithms respectively.The proposed algorithm has considered the defects in both WLAN localizition and PDR algorithms. The experimental results showed that our algorithm has improved the radio map building efficie ncy a lot regardless of the participation of the professionals,as well as to reach a satisfying localiztion accuracy.