Abstract
II
Abstract
In the applications of wireless sensor network, such as environmental science, traffic
management, health care, anti-terror military, space exploration, and disaster monitoring, etc, the
exact location of the information happened must to be known as a guarantee. Therefore, the
localization technology of node in wireless sensor network plays an import and indispensable role.
In the actual environment, it often encounters the problem of irregular network topology, as well as
complex broadcasting environment and uneven node deployment. It required higher performance
on the localization algorithm. This paper carrying on the research analysis based on the existing
localization algorithms, considering the shortcomings of the existing methods, studying the
positioning algorithm performance under ideal environment and the non-ideal environment.
Through the theoretical research, this paper puts forward the corresponding improved algorithms
for meeting the requirement of applicability. The main works of this paper are as follows:
1. Research on the classical multidimensional scaling localization algorithm MDS-MAP(C)
and the mobile anchor node localization algorithm based on convex programming, and analyze the
basic principle and positioning performance of the original algorithm.
2. Due to the RSS-based multidimensional scaling algorithm MDS-MAP(C) using the
shortest path distance to replace the Euclidean distance, and using SVD decomposition to localize
the node, the accuracy of localization can be improved. Aiming at the shortcomings of the MDS-
MAP(C), this paper proposes an improved MDS-DMC localization algorithm based on the shortest
path distance matrix correction. The improved algorithm using heuristic search to reduce the error
between the shortest path distance and Euclidean distance, and using the smacof algorithm instead
of SVD decomposition in order to improve the positioning accuracy and the adaptability of
irregular network. The experimental results show that the improved algorithm has higher
positioning accuracy than the original algorithm and can better adapt to different network model.
3. In order to enhance the node localization accuracy in wireless sensor networks, this article
had studied related references, and proposed an improved convex programming localization
algorithm of mobile anchor node. The algorithm made some improvements on the original
algorithm: to reduce the computational complexity of solving optimization problems, positive
semi-definite relaxation method was utilized for enlarging the feasible region of solving problems;
to improve the accuracy of solving optimization problems, local gradient descent method was used
