基于MEDLL的分级搜索抗多径算法

以多径估计延迟锁定环路(MEDLL)为代表的参量式基带抗多径算法以其能够同时估计直达信号和多径信号各参数而受到广泛关注,然而该类算法估计精度的提高是以计算复杂度的增加为代价的。为了在保证估计精度的同时降低资源消耗并加快搜索速度,提出了一种MEDLL最大似然(ML)估计搜索方法,将粗搜索后得到的参数估计值作为下一级搜索的先验信息,缩小搜索范围后减小搜索步进,完成逐级精细的参数估计。推导了原始MEDLL及树状分级结构的MEDLL(T-MEDLL)算法计算复杂度表达式;其中在存在一路多径信号的情况下,对于0.01 chip的多径延迟分辨率,复杂度可以降低为传统盲搜索的30%~50%。进一步基于自研的数字中频(IF)GPS信号模拟器和软件GPS接收机平台对不同多径情况下盲搜索法与分级搜索算法的参数估计精度及码伪距多径误差等性能进行了对比,验证了更低计算复杂度的T-MEDLL可以达到与传统MEDLL算法同样的跟踪精度。 

Multipath mitigation algorithm with hierarchical search based on MEDLL

Multipath mitigation techniques using parametric baseband processing, represented by multipath estimating delay locked loop (MEDLL), have attracted widespread attention by estimating the parameters of direct path and multipath signals simultaneously. The improvement of the estimation accuracy for such techniques, however, is at the cost of the increase of complexity. In order to reduce the resource consumption and speed up the search while maintaining the accuracy, this paper proposes a hierarchical search strategy for the maximum likelihood (ML) estimation of MEDLL. The rough estimates obtained by the former stage of search units serve as the priori information for the next-stage search with a narrower range and a smaller step to achieve refined estimates step by step. The expressions of computational complexities of the original MEDLL and the proposed tree MEDLL (T-MEDLL) with different number of stages are derived analytically, which shows that the computational burden of our approach can be reduced to 30%-50% of that of the traditional blind search with the same delay resolution of 0.01 chip in the presence of a single multipath. In addition, a digital intermediate frequency (IF) GPS signal simulator and a software GPS receiver are developed to test the parameter estimation accuracy and the pseudorange multipath errors of the hierarchical search and the blind search under different multipath scenarios. The results demonstrate that the T-MEDLL can maintain the same performance as MEDLL with significantly reduced implementation complexity.