室内定位技术可提供准确的位置信息的服务,因而在多种领域中得到广泛应用.在Wi-Fi网络环境下,相关系数法和KNN算法是基于指纹数据库的2种常用定位算法,但两者的定位精度都十分有限,不能满足室内精确定位的要求.为此提出一种最邻近相关系数算法,该算法结合了均值的相关系数与KNN算法,在一定的室内环境下共同发挥了两者的定位优势,并且较好解决了斯皮尔曼等级相关系数和皮尔逊相关系数的异值点问题.在实际物理空间建立起指纹数据库,选择多个测试点对新算法进行了性能测试,测试结果表明,文中提出的最邻近相关系数法在定位精度上有所提升,分别比相关系数法和KNN法的定位精度提升了38.86%和23.35%,同时在数据运算量上并没有增加,可以在室内定位中广泛运用.%Indoor localization can render a service of accurate position information, thus there is a widespread use in many fields. Correlation coefficient method and KNN( K?Nearest Neighbor) are two kinds of common localization algorithms based on fingerprinting database in Wi?Fi environment. But the positional accuracy of them is very finite and can′t meet the requirement of accurate indoor localization. In this paper, we propose a K?Correlation Coefficient algorithm which combines the mean value correlation coefficient and KNN. Their localization advantages can be de?veloped in a certain indoor environment by K?Correlation Coefficient, and the point of different values between Spearman Rank Correlation Coefficient and Pearson Correlation Coefficient is also solved satisfactorily. A finger?printing database is established in the physical space. We choose multiple test points to detect the novel algorithm and the result shows localization accuracy of K?Correlation Coefficient is improved, 38. 86% improved comparing with correlation coefficient method and 23.35%with KNN. Meanwhile the computing workload isn′t increased and it can be used widely in indoor localization.
展开▼
