| Summary: | 碩士 === 國立清華大學 === 通訊工程研究所 === 94 === Abstract
Accuracy and energy efficiency are two major requirements for the development of a positioning system in wireless sensor networks. Some distance-based positioning approaches have been proposed to be implemented in a decentralized manner for energy saving, leaving out the need of transmitting all data to a central server for processing. These algorithms are usually formulated as a least square problem which assumes measurements from all sensors are accurate equally. As a matter of fact, the accuracy of distance estimation from each sensor varies according to the actual distance between the target and the sensor. Thus, some inaccurate distance estimation may cause serious error in location estimation for these methods.
In this thesis, we propose a decentralized positioning scheme based on weighted interpolation. We formulate the positioning problem as a weighted least square problem and distribute different weights among sensors according to the reliability of the distance estimation from each sensor. An iterative scheme is then derived from weighted interpolation such that the proposed positioning method can be implemented in a decentralized manner；Each sensor adjusts the estimate based on local data and then passes the modified result to its neighbor. In addition, an adaptive step size is used for the iterative procedure to have fast convergence. Moreover, a token ring-like distribution scheme is proposed for the decentralized positioning algorithms, which provides energy efficient design principle for practical implementation. Computer simulation results show that, as compared to a previous iterative decentralized positioning method based on the projection onto convex sets, the proposed scheme has faster convergence speed and achieves better positioning accuracy.
|
|---|
