Design And Implementation Of A Multi-source Indoor Localization System

Indoor positioning technology recently has been more and more popular with the rise of location-based services(LBS). In supermarkets, shopping malls, residential exhibition and other complex indoor environments, indoor positioning technology allows users to know their location, to better define the route; in the mining industry, the fire scene and other occasions, indoor positioning technology can timely determining the location of trapped persons, which provide useful information to carry out search and rescue work. In many other areas, indoor positioning technology also has a wide range of applications. However, despite its low cost and easy deployment, traditional single source WiFi fingerprint based indoor positioning technology has a very coarse localization accuracy. So this paper aims at designing a multi-source indoor localization system.Under this circumstance, this paper propose a multi-source indoor location techniques. The multi-source for this algorithm is reflected in its usages of sensor data. It utilizes sensor data of sound, WiFi, accelerator. And it isn't just the simple combination of this sensor information. It uses these sensor information proposing a new indoor localization structure. This new indoor localization scheme smartly utilizes the location information of the surrounding stationary peers to assist localization. This innovative approach first filter out those devices which has a high probability to remain static. These static peers will be employed as beacons. After pinning down the location of these beacons, an acoustic method will be applied to estimate the distances between the beacons and target. Finally, trilateration will be utilized to localize the target. Compared with fingerprint method, our proposed algorithm also doesn't need to place any anchors in the place of interest but it can achieve a far more better localization result.Based on the previous theory, we give an implementation of our proposed system. It is a client/server based architecture. The server runs on the Linux platform which is written in JAVA. The client side is on Android platform and we have already developed an application prototype.We validate our proposed scheme in our office environment. Our scheme has an average localization error of three meters which is obviously superior over traditional Wi Fi fingerprinting method.