Research And Implementation Of Indoor Subarea Location Algorithm Based On Wi-Fi And Ambient Sound Fingerprint

In order to achieve room-level indoor location and indoor subarea location by relying on existing indoor environment and hardware conditions without adding any hardware infrastructure and base stations,this paper mainly focuses on the received signal strength Indicator of public Wi-Fi and ambient sound,combined with Android software.At the same time,in order to obtain a large number of Wi-Fi and ambient sound fingerprints,two software were developed for collecting data.Finally,the indoor subarea location algorithm based on Wi-Fi and ambient sound fingerprint is transplanted to Android software to realize indoor subarea positioning.Firstly,the Wi-Fi fingerprint map established by the received signal strength of Wi-Fi is used to realize the room-level positioning,and then the sound fingerprint map established by the acoustic features of the indoor ambient sound is used to realize the subarea location in the room.In order to deal with a large amount of Wi-Fi and ambient sound fingerprints collected by mobile phone software,this paper mainly extracts the optimal fingerprint by two similarity algorithms,reduces the memory occupied by the fingerprint map,and removes the noise and redundancy of the fingerprints.And before establishing an ambient sound fingerprint map,a variety of sound features and combined features were studied,and the sound features or combined features that are most suitable for subarea in a room are mined.The main work completed includes:(1)Developed Android mobile phone software for collecting Wi-Fi fingerprints,and assisted in the development of Android software for collecting ambient sounds.First determine the various parameters and formats of the Wi-Fi fingerprint,and then design the software framework according to the simulation requirements,and write the required software.When developing software for collecting ambient sounds,the main task is to determine various parameters for collecting sound,including audio sampling rate,channel and duration;(2)Studied the optimal fingerprint extraction algorithm for Wi-Fi and ambient sound.Using the Jaccard similarity algorithm and the Pearson similarity algorithm,the Wi-Fi optimal fingerprint of each room and the ambient sound optimal fingerprint of each subarea in the room were extracted,which solved the problems of redundancy and noise of Wi-Fi and ambient sound of fingerprints collected by multiple devices;(3)Extracted a variety of sound features from the ambient sound samples with a duration of 10 seconds,and studied the reliability and stability of these sound features and combination features used in the indoor subarea location.These sound features mainly included Chromagram,Sonagram,MFCC,power,power spectral density and fifth percentile power,and a total of 41 combinations of these six sound features.In order to improve the efficiency of subarea location and the recognition rate of subarea,a new sound feature PI05 is proposed.(4)Ported the positioning algorithm and fingerprint map to the Android mobile phone,combining theoretical research with practical applications to develop a mobile APP that can be used for real-time indoor positioning.The conclusion of this paper is that the average accuracy of room-level indoor location using Wi-Fi can reach 99.17%.When the ambient sound sample duration is 10 seconds,compared with these sound features and the combined features location results,it is concluded that the combined features of Chromagram,Sonagram and improved power spectral density have the highest recognition rate of the subarea,which can reach 85.21%.Further research on the improved feature of the fifth percentile power,PI05,and the experiment of shortening the duration of the ambient sound sample,the average recognition rate of the sound feature PI05 can reach 87.0% when the ambient sound duration is 3 seconds.