Research On Multi-Modal Indoor Localization Of Smartphone Based On Image And WiFi Signal

With the rapid growth of the performance of terminal equipment,and the popularity of location-sensing applications,the technology of indoor localization has played an important role in the mobile computing era.Providing location information continuously and reliably in indoor environments can provide a better user experience.Current indoor localization methods are difficult to balance accuracy and cost issues.Sensing methods are subject to different levels of dynamic interference of indoor activities,and high precision,easy to deploy,and robust positioning services are the key requirements of indoor mainstream applications,such as map reconstruction,indoor navigation and augmented reality.This thesis innovates in proposing a camera-based,sensor-and WiFi-assisted system for localization.The system enhances indoor localization with multi-modal sensing,and provides high-precision,easy-to-deploy,and highly stable indoor localization services,which still performs well in a sparsely characterized environment.We first model and design this multi-modal indoor localization and propose several algorithms in this model.We then implement a prototype and evaluate it in distinct indoor environments.We also discuss localization in a multiplayer environment.Our contributions in this thesis that can be summarized as follows:(1)We demonstrate the feasibility of enhancing indoor localization based on multi-modal sensing via wifi,camera and sensors of smartphones and design the system that can be used to assist reconstruction of building interior view and indoor navigation further.(2)We propose an innovative algorithm for distance and orientation estimation,making full use of images to acquire scene information,considering IMU sensors to capture the motion data,and using the CSI to calculate the distance and azimuth of the user activity through the Fresnel model.Images provide scale information and visual cues,CSI and IMU sensing fusion to extract motion information.(3)We implement a proof-of-concept localization prototype and evaluate it in various indoor environments.When only one active user indoor,the experimental results show that the 92-percentile error is within 0.2m for indoor POIs and achieves sub-meter accuracy overall.(4)We explore the possibility of indoor localization in a multi-person environment,identifying the actions of multiple users through effective methods,and further obtaining the location information of the target users to achieve indoor position in a complex environment.The experimental results show that the 92-percentile error is within 0.22 m.