Key Technologies Of Acoustic Indoor Localization Systems For Smart Mobiles

The indoor location-based service(ILBS)for smart mobiles has a promising future in many real applications.And it is of great significance to developing a high-precision indoor positioning system for our daily life.Even though the market has a huge demand for commercial applications,the commercial-off-the-shelf positioning technology with high performance has not yet proposed.Our government and major technology giants have also invested heavily in this area,aiming to pro-mote the development of indoor localization technologies.Acoustic indoor positioning technolo-gies have the advantages of being fully compatible with commercial off-the-shelf smart mobiles,higher positioning accuracy and low-cost infrastructures,and,thus,have attracted researchers’ at-tention.However,the performance of acoustic based systems are seriously impaired by the heavy indoor multipath propagation and non-line of sight phenomenon.Many key technologies are still needed to be tackled from basic theory.This thesis is aiming to realize the high-precision localiza-tion and navigation for smart mobiles in the complex real indoor environments,and pave the way for the real applications of acoustic indoor localization systems.According to the characteristics of indoor scenarios and the acoustic based systems,the key technologies are thoroughly studied from basic theory based on a low-cost experimental,and new methods with low computation complexity are proposed for indoor localization and navigation.The main content of this thesis is summarized as follows:1 A novel TOA estimator based on a "iCleaning" process is proposed for low-cost acoustic chirp ranging,by detecting and extracting the first path signal in FrFD and estimating the time delay based on maximum-detection MF estimator.And a detection method based on a slack threshold and a strict threshold are proposed in FrFD to dynamically detecting the first path component.Six termination criteria are proposed to improve the performance of TOA estimation.And the method of filters design and parameter selection are also introduced.The simulation and experimental results show that the proposed method can realize high accurate TOA estimation compared with optimal threshold based MF estimator,and the ranging accuracy in 30 m is better than 30 cm.2.A new acoustic NLOS identification approach based on acoustic channel characteristics is proposed for smart mobiles indoor localization in the real world.An efficient approach to esti-mate relative channel gain-delay method is proposed to represent the difference of acoustic channel between LOS and NLOS scenarios.Then,9 novel features are proposed based on these character-istics.The performance of classifiers based on supervised learning,semi-supervised learning,and unsupervised learning method are evaluated,and the optimal classifiers are given under accuracy criterion.The experiments in the real scenario shows that the accuracy of the proposed approaches is better than 87.1%,and has a higher accuracy than the other method.Meanwhile,our method is the first one to realize acoustic NLOS identification based on channel characteristics.3.A novel NLOS positioning method based TOA and FOA for moving targets is proposed by using the range the relative speed information which contained in the Doppler Effects.The"IDLSE-TDFS" and "IDLLSE-TDFS" algorithm are proposed based on numerical search and pseudo-linearization method,respectively,and the needed LOS measurements number is reduced to 2 successfully.The information supplement strategy is introduced for single LOS measurements scenario.To further improve the performance,"KCIDLSE-TDFS" and "KCIDLLSE-TDFS" are proposed by using kinematic constrain method.The simulation and experimental results show that the proposed 4 kinds of algorithm have obvious advantages over the other algorithm.Based on low-cost experimental platform,the mean accuracy of KCIDLSE-TDFS algorithm and KCIDLLSE-TDFS algorithm is 0.18 m and 0.68 m,respectively.4.A novel indoor map construction method based on mobile crowd sensing,"iMapDo",is proposed to realize indoor map dynamic construction,which includes the accessible area,furniture area and structure area.A posterior probability model and a template model based map construc-tion methods are proposed,respectively.A reliability factor calculation method for the participant information is proposed to select the participants in mobile crowd sensing process.The numerical simulation and experimental results show that the proposed method can realize the rapid construc-tion of the accessible area,furniture area and structure area.The information of the constructed map is more complete and the construction speed is faster,while the other MCS based low-cost mapper can only realize indoor binary map construction.