| Summary: | 碩士 === 國立臺灣海洋大學 === 通訊與導航工程系 === 97 === The Global Positioning System (GPS) is one of the most popular positioning systems at present. However, the GPS does not perform well in indoor environments and metropolitan city areas. Development of wireless regional network technology for indoor positioning applications has become very popular in recent years. The tools of wireless location technology include WiFi, ZigBee, RFID, bluetooth and ultra broadband, etc. This paper discusses the tracking precision for\ a maneuvering object in a sensor network. Traditional extended Kalman filter (EKF) may bring large errors due to modeling errors and filter limitation. Selection of more advanced nonlinear filters is possible for performance improvement. The unscented Kalman filter (UKF) employs a set of sigma points through deterministic sampling, such that the linearization process is not necessary, and therefore the error caused by linearization as in the traditional EKF can be avoided. The particles filter (PF) exhibits superior performance as compared to EKF and UKF in state estimation for the nonlinear, non-Gaussian system. Nevertheless, the degeneracy of particles and cumulation of estimation errors in the PF are difficult to overcome. To handle the problem of heavy-tailed probability distribution, one of the strategies is to incorporate the EKF into the PF as the proposal distribution, leading to the extended Kalman particle filter (EPF). The simulation results show that the extended kalman particle filter(EPF) is more effective and it’s precision is higher than Extended Kalman Filter (EKF), Unscented Kalman Filter (UKF) and Particle Filter (PF) in a sensor system .
|
|---|
