| Location information can be used to provide personalized services to enhance user experience,promote the development of the Internet of Things industry and the construction of smart cities.With the widespread application of the Global Positioning System(GPS)and Bei Dou Navigation System(BDS),the outdoor positioning problem has been basically solved,and the outdoor location information services(Location Based Services,LBS)industry is booming.However,humans spend more than 80% of their time in indoor environments.Location services,social networks,health assistance,smart cities,emergency rescue,Internet of Things,and precision strikes all require indoor positioning.Due to the impact of building occlusion and multipath effects,the indoor positioning performance of GPS or BDS cannot meet people's application requirements for indoor LBS.Therefore,indoor positioning has become the most urgent problem to be solved in the promotion of LBS applications.Practical indoor positioning solutions require to meet the needs of accuracy,coverage,reliability,cost,power consumption,scalability,and response time.In recent years,researchers at home and abroad have proposed indoor localization technologies and application systems such as Bluetooth,infrared,RFID,WLAN,ultra-wideband,and ultrasound.However,due to factors such as human activity interference,signal multipath propagation,and infrastructure dependence,different Indoor localization technology has certain application limitations according to its performance.To date,there is still no universal indoor localization technology that can provide a global,all-weather localization service that dominates the indoor,allowing the LBS industry to complete indoor and outdoor coverage.How to perform scene recognition and improve the localization performance of interested targets in a complex indoor environment has become a common focus of various intelligent interaction and communication system industries.This thesis studies the situation of indoor localization,especially the development of indoor acoustic localization,and pins down the key technology research plan of Acoustic Room Geometry Reconstruction(ARGR)and Indoor Acoustic Localization(IAL).Through multi-channel and single-channel acoustic signal acquisition and analysis,related research on ARGR,indoor sound source localization and indoor acoustic fingerprint localization are carried out to reveal the influence of ARGR on IAL performance.The significance of investigating the key technologies of ARGR and IAL is not only to meet the multi-scale requirements of different LBS users for location accuracy,but also to meet the privacy protection requirements of LBS users for their position.This thesis introduces the background theory that the above research needs to involve,clarifies the basic principles of ARGR and the working mechanism of the IAL system,as well as analyzes and summarizes the influence of ARGR on IAL in terms of location accuracy and efficiency.The main contents of this thesis are summarized as follows:1.The development of IAL technology is reviewed firstly,especially for indoor sound source localization and indoor acoustic fingerprint localization.The development of ARGR is outlined from the aspects of multi-channel and single-channel acoustic signal acquisition and analysis,and the influence of ARGR on IAL performance is summarized.2.The basic theory of ARGR is introduced,the indoor sound field theoretical model is analyzed,and the basic formulas of the three-dimensional wave equation are deduced,as well as the process of using room wall acoustic resonance characteristics to realize ARGR is presented mathematically.Several problems of selecting parameters in practical applications of wave acoustics based ARGR are summarized;The basic methods of ARGR based on geometrical acoustics are investigated from multi-channel(microphone array)and singlechannel(smartphone)acoustic signal acquisition and analysis respectively for providing indoor acoustic field environment information to IAL systems.Finally,the basic theory of IAL is introduced,and several typical IAL methods are investigated,including time delay estimation,beamforming and machine learning.The important influences of the indoor sound field environment on the IAL system are concluded,theoretical and methodological support for the subsequent research work of indoor sound source localization and indoor acoustic fingerprint localization are also provided.3.An Acoustic Room Geometry Reconstruction method based on three-dimensional microphone array perception is proposed,which makes full use of the advantages of array topology,reduces the possibility of misjudgment of the first-order echo peaks,simplifies the ARGR method based on multi-channel acoustic signal acquisition and analysis,and then improves reconstruction accuracy and efficiency,as well as guarantees the feasibility of a steered response power with phase transform(SRP-PHAT)beamforming algorithm using Stochastic Region Contraction(SRC);On the basis of ARGR,a Delaunay Triangulation based Search Volume Reconstruction algorithm,called DTSVR,is proposed,which reasonably and effectively reduces the search volume of indoor moving sound source(a walking speaker),while ensuring the position estimation accuracy of the SRP-PHAT-SRC,to efficiently realize the real-time positioning and tracking of a speaker.4.A continuous self-localization method for indoor pedestrians using a sensor-rich smartphone is proposed,which combines pedestrian trajectory estimation(Pedestrian Dead Reckoning,PDR)based on smartphone inertial sensor data with indoor sound source localization based on time delay estimation algorithm,and an acoustic constraint update algorithm which reduces the PDR cumulative errors by using the geometric relationship between the sound source and its first-order image sources is provided to ensure that the update eliminates the PDR iteration error,and realizes accurate,effective,and continuous self-positioning of indoor pedestrian without user collaboration.5.An indoor LBS user location method based on Auditory Scene Analysis(ASA)is proposed.This method uses the psychoacoustic feature spectrum of the indoor background sound as the acoustic feature to construct the multi-dimensional acoustic fingerprint,and combines the room geometry information and the clustered results of the subareas in the room to outline the indoor sound energy distribution status(room color acoustic image),so that which it can not only provide,on the one hand,more detailed location information and spatial layout for indoor small subarea localization,on the other hand,does not but also avoid to expose the indoor scene layout and the image details of objects,and so as to have solved the contradiction of privacy protection of traditional indoor maps;on this basis,the related theory and algorithm of machine learning are adopted to build and update the "locationfingerprint" mapping relationship database offline,matching the minimum variance online to quickly and accurately locate the user's location,and realize the indoor small subarea level localization without infrastructure dependence and user collaboration. |
PDF Full Text Request