Post-processing of Fingerprint Localization using Kalman Filter and Map-matching Techniques

Most of the available positioning technologies have limitations in either accuracy of absolute position (cellular localization), accumulated error (dead reckoning techniques) or availability of the signal (GPS). This paper describes the features of post-processing algorithms performed to improve the positioning accuracy achieved with fingerprint localization within a cellular network. Position estimates got from the so called neural network (NN) localization are further processed through a Kalman filtering-based tracking algorithm and thereafter, the processed position is matched to the road according to the map-matching technique applied. Results show that an accurate positioning within a cellular network can be achieved and could serve as complementary to the existing GPS.