System Architecture
Our system follows a mobile-based network-assisted architecture and has therefore two highly desirable properties:
- Low communication overhead, because the unknown user location is estimated on the mobile device and we avoid uploading the observed RSS fingerprint to a positioning server on the network side.
- Respects user privacy because the location is estimated directly by the user and not by the positioning server.
In a typical positioning scenario, a user enters an indoor environment, featuring WiFi APs. Then, the user’s smartphone obtains the Signal Strength radiomap and algorithm-specific parameters from the local distribution server in a single communication round. Subsequently, the client positions itself independently thereafter using only local knowledge and without revealing its personal state.
Our Android positioning platform consists of three components:
- The RSS Logger is an application developed around the Android RSS API for scanning and recording a user-defined number of samples in specific locations at predefined intervals. The collected data is stored locally in a log file and the user can contribute them to the system. In this way the RSS data are collected collaboratively by crowd sourcing, which greatly facilitates the construction and updating of the radiomap.
- The Distribution Server is mainly responsible for the construction and distribution of the RSS radiomap to the clients. Another interesting feature in our server implementation is a methodology for fine tuning algorithm-specific parameters, such as the number of nearest neighbours K used in the well-known K-Nearest Neighbour (KNN) algorithm.
- The Find Me application is our positioning client that runs on Android smartphones and connects to the server in order to download the radiomap and algorithm-specific parameters. The user may select any of the available and the client is then able to position itself independently.