| Wireless sensor networks need to transmit their collected information to users. Only data with location information is meaningful, so the user must first know the location information of each node. This paper presents a discrete probability estimation localization algorithm, and the application of a refinement algorithm (Extended Kalman Filter (EKF)) in the node positioning is studied. Eventually the localization and refinement algorithm is test in a practical experiment through the woods.Known the phenomenon of normal distribution, we tried some new data acquisition and processing methods. By analyzing the probability distribution mechanism, we search for alternative rectangle area, and use the grid method, presenting a discrete probability estimation localization algorithm. The algorithm is easy to understand and could be easily applied, while trying to minimize the calculate pressure of the nodes. It also overcomes the defect of the continuous localization algorithm, and corrects its deficiencies. Experiments show that the algorithm has better positioning results than the triangular method.Distance measurement causes large error in RSSI-based localization technique in WSNs. In order to improve the precision of localization, an extended kalman filter algorithm is introduced to restain the calculation error. Two application approaches of EKF, which can be used for WSNs, are compared with each other in this paper. At the same time, all kinds of topological conditions that may occur in the process of localization and how to improve the localization accuracy are investigated with convergence probability and relative error as indicators. Applicable situations of EKF is then analyzed via simulation, and factors that may impact on the performance of EKF is discussed. Based on the above findings, we set the initial state error covariance matrix and other parameters to apply the EKF refinement algorithm in practical. The experiments obtained satisfactory results. |
PDF Full Text Request