Research On Landmark Recognition For Indoor Localization Based On Signal Analysis

In recent years, the requirement of various mobile services in daily life has grown rapidly along with the development of mobile network. The location service is becoming a daily routine and many research efforts have been devoted into indoor location technology development. As the basic for indoor map construction and location, more researches should be conducted on the indoor location feature recognition.A study of the difficulty and feasibility of the landmark recognition has been conducted in this paper by analyzing the signal data collected by mobile device in real scenarios. A novel algorithm is proposed in this paper. Meanwhile, the improvement and validation of the state-of-art recognition algorithms is also presented in this paper to improve the precision of indoor landmark detection and to increase the available landmark classes.After a literature review of the state-of-art indoor location technology, the indoor context perception is chosen to be the research direction of this paper. Large quantity of signal data are collected using WiFi scanner, accelerator, gyroscope, barometric sensor and other sensors embedded in mobile devices to analyze the signal fluctuation features. Different recognition algorithms are then designed according to different landmarks, such as stair recognition utilizing the combination of walking features on stairs and atmospheric pressure fluctuation as well as door corner recognition based on undirected graph. The proposed novel algorithms can increase the precision of stair corner and door corner recognition effectively.Moreover, an efficient solution to distinguish floors is also proposed in this paper. Large quantity of data are collected by collecting the pressure difference between each floor and the first floor. The feature pressure value of each floor is then concluded from the data classification for the purpose of floor classification of indoor landmarks using the pressure feature attribute. The system can also contribute to floor location as a part of indoor location.At last, an Android based indoor landmarkcollection and analysis system has been developed utilizing the proposed algorithms. A series of experiments in several buildings with various types are conducted and the system has demonstrated satisfied results.