Research And Implementation Of Enclosure Scenario Positioning And Map Topology Sensing Algorithm Based On Acoustic Signal

With the continuous development of mobile intelligent sensing technology,people’s demand for accurate indoor position and map topology sensing technology is increasing.High-accuracy enclosure space location and sensing technology plays a key role in human’s behavior analysis,providing smart in-vehicle interactive services and safe driving.Traditional Radio Frequency Signal based Time of Flight algorithms suffer from environments effecting,high accuracy time synchronization,poor accuracy,etc.To provide accurate and robust positioning and map topology sensing service,researchers have proposed a variety of methods in recent years,among which positioning algorithm based on acoustic reflection signal has gained researchers’ attention due to its advantages of no need for highprecision time synchronization and wide deployment.The above methods provide a new solution for high precision enclosure scene location.But there are also challenges such as poor anti-interference ability and multipath effect in enclosure space.An enclosure scenario infrastructure-free positioning and map topology sensing algorithm based on acoustic signal reflection and deep learning technology was proposed.The proposed algorithm can localize the passengers without the need of any additional sensors or pre-installed infrastructure deployed in the vehicle.The core idea of the algorithm is to send out a series of special imperceptible acoustic signals by using the speakers embedded in the commodity mobile phone carried by the users,then a Sliding-Window FFT(Fast Fourier Transform)algorithm is used to extract corresponding time-frequency domain features and build up acoustic domain features from reflection signals;Finally,a time convolution neural network model is used to transform the enclosure positioning problem into a classification task,which implement the enclosure scenario high-accuracy positioning and map topology sensing.This project designs an enclosure scenario location subsystem and map topology sensing subsystem.And this project finally implements four modules including customer application,test application,manager application and location subsystem,which provides customer some location services like enclosure space position,system deployer professional function such as fingerprint collect and location test and real time monitoring for administrator.This project follows the principle of software engineering and is divided in five parts.In requirements analysis part,the system’s functional requirements and performance requirements are decided.Then in outline design part establish the complete architecture.In the detailed design stage,the analysis and design of system modules are finished.Finally,after completing the system development,the system would be tested in detail and correct some shortcomings.All the relate work of this project have been completed.Experimental results show that our proposed enclosure scenario location algorithms can achieve more than 94%classify accuracy,2 meters position accuracy and map topology sensing algorithm’s accuracy is less than 5 meters.