科研之友
微信
新浪微博
Facebook
分享链接
摘要
Integer ambiguity decorrelation has long been an important problem for global navigation satellite system high-precision relative positioning applications for its effect of deducing search number of candidate integer ambiguities. It is difficult to design a decorrelation method with fine performance. Among those decorrelation methods based on LDLT decomposition, direct-ordering method (DOM) is the one that follows the principle of ordering diagonal elements of variance-covariance matrix before decorrelation. In the present study, the authors propose a diagonal element precomputing and an ordering method (DEPOM) based on LDLT. DEPOM orders diagonal elements of the matrix according to values after LDLT decomposition, in contrast to DOM according to values before LDLT decomposition. Thus, DEPOM is closer to the aim of arranging the larger diagonal element to the larger row before decomposition. The nodus is to determine elements in decomposed L and D matrices, which constitutes an improved LDLT decomposition method. The above study show that DEPOM has better decorrelation degree and a higher success rate than DOM from the numerical simulation tests. The performance of DEPOM is equivalent to the united ambiguity decorrelation method. However, it is different. DEPOM is a supplement to the integer least-squares theory. The improved LDLT decomposition is also a new form in the LDLT decomposition family.
- 出版日期2013-1
- 单位北京大学
