The Algorithm Based On Filtering And Data Fusion On Indoor Location Technology

With the rapid development of communication network technology, prehensive coverage of positioning technology service has becoming essential to social development. Positioning technology has brought a lot of convenience to the community, and even promotes a new round of sociopolitical transformation. GPS (Global Positioning System) provides important technical support application functionality in the path planning, vehicle navigation, real-time rescue, etc. But, GPS is more used in outdoor environments. For complex indoor environments, GPS technology has a lot of limitations. Therefore, in order to achieve a full range of localization, indoor positioning technology has rapidly developed and widely applied.Indoor positioning technology commonly includes ultrasonic, frared, Bluetooth, ZigBee, UWB, RFID, and so on. Different positioning technologies are applied to different positioning range because of positioning accuracy, the primary reason is that the positioning algorithm. The positioning algorithm commonly divided into two types:One is based on distance measurement, which calculates the coordinates of the target nodes with nodes distance and angle; another is based on non-distance measurement, which is mainly determine the target location according connectivity and network topology. Of the two the latter often requires the deployment of a network between the sites to be relatively dense in order to avoid larger errors. Thus, distance-based positioning algorithm usually is used for the target tracking system.Distance-based positioning algorithm includes:TOF, TOA, TDOA, AOA and RSSI, based on the signal arrival time; AOA, based on the signal arrival angle, and RSSI, based on the signal arrival intensity. In this paper, the advantages and disadvantages of these algorithms were analyzed. To achieve higher location accuracy, the fusion location algorithm was proposed. The main work and innovation points of the paper as follow:1、There was only targeting on a static in ranging localization algorithm described above. In order to achieve real-time tracking during exercise, according to the kinematic principle, dynamic movement model equations were established in this paper, and the motion parameters were estimated effectively. As the process noise and measurement noise were the main factor to cause ranging error, Kalman filter was used to effectively filter the data, and particle filter was used for noise reduction processing for nonlinear systems. 2、On the basis of the existing motion model, considering the impact of the acceleration on a moving object, Interacting multiple model (IMM) equation was established. The motorized and non-motorized model equation model was effectively combined and between the different state transition probability matrix model was joined by Markov (Markvo) chain. Finally, the Kalman filter was used for filtering and tracking.3、In order to order to improve the positioning accuracy, a new location algorithm was proposed in this paper. RSSI and TOF were fused to improve the measurement accuracy in case of different distances, measurement was needed to adapt to complex environments. The experimental results were compared and analyzed in detail, and showed that the method based on neural networks was effective and the final positioning result was excellent, and could result in a significant improvement for indoor location application.