Research On Indoor Localization Algorithms Based On Multi-Data Fusion

Positioning is one of the key technologies of Location based Service(LBS). Positioning accuracy, power, robustness and some other factors will have a great impact on positiong system. In modern city, more and more human activities take place in indoor screens, conventional satellite based positioning has been unable to meet the needs of indoor LBS applications. A professional indoor positioning technology is critical. On one hand, proposed “Internet+” strategy promotes the development of indoor LBS application; on the other hand, the Big Data makes the relationship between data mining technology and indoor LBS applications more closely.Indoor positioning technology is still immature, a high-accuracy, low-power and low-cost indoor positioning system has become a hot research target in both academia and industry. This thesis studies indoor positioning technology, and the research target of this thesis is to propose a high-accuracy and low-cost indoor positioning system based on multi-data fusion. The studies of this thesis are as follows:(1) Indoor positioning system based on the Data Fusion of Wi-Fi and Radio Frequency Identification(RFID)In order to improve the performance of indoor location system and eliminate the blind areas of location, we propose a novel indoor localization mechanism, which realizes an effective data fusion of Wi-Fi and RFID signals via on-demand deployment of Wi-Fi access points and RFID tags. The experimental results show that KILA achieves a better accuracy of positioning than typical Kalman-filter-based localization algorithm, about 13% to 28% accuracy of positioning improvement in the indoor environments with the [35dB, 65dB] indoor noises.(2) Indoor positioning model based on inertial measurement unit(IMU)This thesis proposes a pedestrian positioning model using calibrated data from the IMU of a smart mobile terminal and an indoor tracking algorithm namely Autocorrect Particle Swarm Optimization(APSO) respectly. The experimental results show that APSO achieves a better accuracy of positioning than Particle Swarm Optimization(PSO), about 150% accuracy of positioning improvement. And this improvement is benefit from the APSO which can correct positioning logic error in an indoor positiong process.(3) Indoor positioning based on multi-data using Linear Chain Conditional Random Field(LCCRF)At first, this thesis builds a novel continuous magnetic fingerprint map based on the characteristic that magnetic flux density is different everywhere in an indoor scene. Secondly, this thesis uses LCCRF model to fuse multi-data including magnetic fingerprint map, BLE fingerprint map, IMU and so on. Comparing with traditional indoor positioning methods, the method based LCCRF model achieves a better positioning accuracy and robustness.