| Short-range location has caused widespread concern with the requirement of precise positioning, especially for indoor location. While for indoor location, the transmission characteristics of the wireless signal are susceptible to the obstacles, and lower location error is needed (generally less than1m), GPS which is widely used for outdoor location is not suitable for indoor location with weak signal and low accuracy. The wireless sensor network (WSN) and radio frequency identification (RFID) may be good choices with the characteristics of compact size, low cost, non-contact, non line of sight, and wide coverage.For the static target localization problem in WSN, we elaborately present two algorithms (least square and Kalman filter) under the measurements of distance, angle, or both of distance and angle, respectively. Theoretical analysis is made by deriving the expectation and covariance of the estimation error. With the analysis of LS-distance model,3-sensor case is discussed to obtain the optimal placement by minimizing the trace of the error covariance. Some simulation tests were aslo made to compare the two algorithms and the results demonstrate that EKF with distance measurements performs better and is more robust than LS when the measurement noise is related to the distance between sensor and the target. At last,4normal sensor placements (square, circle, triangle, and hexagon) and it is founded that the square placement performs best.Furthermore, a new method for moving robot localization based on RFID is proposed. Using the tags which just get into or out of the detection region of the reader, the binary measurements are transformed to distance measurements between the changing tags and the reader. Then extended Kalman filer is used to estimate the location of the robot based on the dynamic model and the distance model. The simulation results show that the position of the robot can be tracked and the algorithm does not need too much computation. |
PDF Full Text Request