Research On Key Technologies Of Acoustic Indoor Positioning For Smart Mobile Devices

With the rapid development of urbanization,the area of human indoor activities is gradually increasing,and the demand for indoor positioning is also increasing,such as smart medical care,public safety,commercial promotion,and so on.Because satellite signals cannot penetrate indoors,indoor location services can only rely on other positioning technologies such as WIFI,Bluetooth,UWB,RFID,PDR,geomagnetism,computer vision,audio,etc.Among them,the acoustic indoor positioning technology is expected to provide users with high-precision indoor positioning services of decimeter level through low-cost hardware facilities due to its good compatibility with smart devices.In recent years,it has received widespread attention from domestic and foreign researchers.In practical applications,the strong multipath,non-line-of-sight(NLOS),and Doppler effects brought by complex indoor scenes are key factors that affect the performance of the acoustic indoor positioning system(AIPS).In addition,the influence of factors such as the geometric conditions between the target and the node,the acoustic transmission scheme,the clock deviation of the node,etc.cannot be ignored.Based on the self-developed low-cost audio indoor positioning hardware platform,and facing many complex factors in practical applications,the dissertation focuses on the research of audio detection algorithms,positioning algorithms,and exception handling and reliability judgment methods,aiming to provide solutions for realizing high-precision and robust real-time audio positioning.Specific research work and results include:(1)To realize the estimation of time difference of arrival(TDoA)and node relative velocity(NRV)in complex indoor scenes,a novel audio arrival time(AAT)detection algorithm including coarse search and fine search steps is proposed.The coarse search uses cosine theorem and system synchronization timing to achieve adaptive extraction of audio data segment(ADS)to overcome the effects of background noise and inaccurate system clock.The fine search uses the indoor propagation characteristics of audio and the waveform characteristics of the source signal autocorrelation to accurately estimate AAT from ADS.Static target and dynamic target positioning experiments were carried out in different scenarios such as hall,corridor,office,and underground garage,which verified that the proposed AAT estimation method is more robust than the two traditional detection methods.For the AAT detection error caused by the Doppler frequency shift caused by the target motion,a novel Doppler correction method is proposed.This method uses NRV and reconstructs the reference signal matching the received audio to achieve accurate AAT estimation in the target motion situation.The effectiveness of the proposed Doppler correction method is verified in the dynamic target positioning experiment.(2)To improve the robustness of positioning in complex scenes,various positioning algorithms are studied.First,a TDOA-based combined weighted(COM-W)positioning algorithm is proposed.Taking two-dimensional positioning as an example,this algorithm uses the weighted average of Cramer-Rao lower bound(CRLB)to obtain the final estimated target position on the basis of all three-node combined positioning.Both numerical simulation and actual experimental results show that the COM-W algorithm has excellent positioning performance under the test conditions of large measurement noise,height difference exists between the node and the target,and the target is close to the node,namely,the COM-W algorithm is relatively less affected by measurement noise and adverse geometric conditions.Second,for the error caused by the time division problem in TDoA-based positioning,a correction method using estimated speed to move node coordinates is proposed.The results of dynamic target positioning experiment show that the time-division correction method can improve the positioning accuracy by about 5 cm.Third,a search weighted(SEA-W)positioning algorithm based on TDoA+NRV is proposed.The algorithm constructs a search area by solving the solutions of all measurement combinations,and determines the solution that makes the objective function optimal in the search area as the final position estimate.The simulation results show that,under the premise of reliable historical position,the number of nodes required by the SEA-W algorithm is reduced to two,which can be used for exception handling and reliability judgment,thereby improving the robustness of the system.(3)Research on AAT exception handling and location reliability judgment.For possible abnormality of AAT estimation,an adaptive Kalman filter(AKF)and a reliability judgment(RJ)method are proposed.AKF improves the traditional Kalman filter and directly filters AAT and NRV.It can adaptively adjust the filter parameters according to the measurement results and adapt to the characteristics of strong randomness of the target motion.The RJ method combines the signal quality judgment and the estimated position consistency analysis.Only when the system meets the signal quality and estimated position consistency requirements will the reliability indication be given.The results of numerical simulation and actual experiments show that AKF can effectively suppress the adverse effects of abnormal AAT detection to a certain extent,and RJ method can effectively identify and eliminate the abnormal estimated positions.The positioning process combined with the AKF and RJ methods improves the positioning performance of the system and shows good robustness.