A Research On Radio Fingerprints Of Multipath And Its Indoor Location Algorithm

There are two critial problem about the Autonomous robot of navigation and inspection : the indoor position and the closed-loop detection,whose diggiculties are the position accuracy and the accumulated error of the closed ring detection in large scale environment.The proposed novel algorithm based on the radio fingerprint position improve the accuracy of indoor location and the closed-ring detection based on radio fingerprint,and inhibit the accumulated errors.A novel single site position feature descriptors and the corresponding identification algorithm are proposed.the signal subspace method is used to extract the position information contained in matrix of transient space,with the Low-dimensional position feature descriptor computed by projection matrix and the noise of reserved signal and model error restrained.Introducing the Glassman projection scale function to construct the cost function,the optimal position fingerprint is obtained by the principal component analysis of projection matrix.In the computating of location identification,similar measurement is conducted through Cosine Distance,and the optimal position is estimilaited through the Global maximum similarity measurement strategy,so that the robustness of fingerprint position and the Positioning accuracy are both improved a lot.In order to solve the problems of positioning accuracy and accumulated error,introduce the radio fingerprint position to the closed-loop detection of SLAM,search the optimal location by setting Path sliding window and Dual-resolution map,Reduce the fuzzy effect of location and relocate rapidly.As a result,the closed-loop detection efficiency of large scale indoor environment map has improved.From the simulation experiment based on indoor wireless data,the results show that the Identification precision of position fingerprint and identification alrithm are smaller than 1m,the distribution of accumulated error is better than SSP,KPCA and other method,during the radio closedloop detection of random and regular path,the accuracy reach the standard of closed path,and the error bound is less than 1m in various environment.