Multi-sensor Fusion Localization Technology Research And Implementation

With the rapid development of wireless communication technology, wireless location technology has played an important role in various fields. In recent years, mobile internet spawned a large number of location based services, making wireless location has become a hotspot. In order to improve wireless location accuracy and robustness, the author studied and improved some kinds of wireless location schemas, and a multi-sensor fusion location technology has been proposed to take advantage of a variety of sensor data, and thus build a more stable and reliable wireless location positioning platform. The main contents of this paper is divided into the following three parts:Firstly, a indoor location methods based on SVM classification and regression has been proposed.The method divides the indoor area into a number of small areas, and builds the nonlinear relationship between RSSI and location in each area based on the improved triangulation principle without knowing the reference nodes’ locations, and then use SVM powerful classification and Regression ability to location the target.Secondly, achieved pedestrian dead reckoning on handheld devices. By analyzing the model of human walking, the relationship between acceleration change and step counting and step length calculation was cleared. And in order to reduce the impact of interference data on step counting and step length calculation, some new schemas such as sensor acceleration projection and decomposition in three demensional coordinates, data smoothing using SMA algorithm and related parameter dynamically set was proposed. Finally, kalman filter was used to further improve the step length calculation algorithm performance and ensurethe positioning system can be realized on a common handheld devices.Finally, based on multi-sensor fusion technology,a multi-sensor fusion location schema was achieved. By modeling the movement of people and using kalman filter to fusing Wi Fi location result and pedestrian dead reckoning result, thus builded a more effective than single-sensor positioning positioning platform.