Abstract:

Highly accurate and reliable indoor wireless positioning services have been widely used.In order to obtain good positioning accuracy, the design of positioning algorithms needs to be matched with wireless positioning facilities.fiber to the room (FTTR) is an indoor access network solution based on IEEE 802.11 ax, a new generation of wireless local area network (WLAN) standard.Compared with the existing Wi-Fi networks, FTTR has a much larger available band width.However, FTTR WLAN also lacks of a public valid data set to support localization functions, which makes the localization research based on FTTR scenarios face huge obstacles.In order to solve the above problems, firstly, a frequency response-based FTTR scene dataset generation method was proposed, which uses the existing Wi-Fi localization dataset to generate the frequency response matrix within the available band width of FTTR.Then, the parallel path principal component analysis (PCA) method was used to generate the classification matrix.And the generated dataset was trained using a fully connected neural network to improve the accuracy.The experimental results on the real measurement dataset show that the proposed localization algorithm can achieve a localization accuracy of less than 1 m, which is not only more accurate than the traditional location estimation algorithm, but also basically meets the fine-grained localization requirements for practical applications.

Key words: FTTR, dataset synthesis, principal component analysis