| Abstract [eng] |
The main goal of this work is to create a prototype of indoor localization system, which would be able to determine the floor of a building, and location on the floor. In this work methods which are used for indoor localization were analyzed. Also localization schemes using range-based and range-free methods were analyzed and in detail explained in this work. The selected base method for our prototype of indoor localization system is one of range-free methods - WiFi fingerprint method. Other goals of this work are analysis of floor determination algorithms and analysis of localization on the floor algorithms. Best results giving algorithms were adapted to work with our system prototype. Furthermore, the implemented system uses, only MIF infrastructure (existing wireless network in the building, and virtualized server in the MIF cloud system) and smartphone. The created prototype of indoor localization system is implemented to work in the building of MIF faculty (Didlaukio g. 47), on 2nd, 3rd and 4th floors. According to our analysis of algorithms, which are being used in our implemented localization system prototype, they can estimate correct location with high accuracy, but as we have seen in our prototype system, the accuracy highly suffers from the recency of the reference data. It means that we get best results when reference data is newly collected. And after that as time passes by, and our reference isn't renewed the accuracy of our system's localization decreases. |
