Radio Tomography Indoor Positioning System Research And Implementation

This paper design and implement indoor posining system with 24 GHz radar as measurement equipment to locate and track human body in regular indoor environment.Wireless indoor posining has many advantages than other types of indoor posining systems.It does not cause invasion of privacy,not effected by indoor lighting conditions,and never interfere normal activities.There are some positioning radar for human body avaliable at present.This paper takes proven thchnology and mature devices to design and implement the core alogrithm and application software of the system,and modified existing alogrithm on other platform to adapt the radar devices.The indoor positioning radar for human body can detect multiple targets simultaneously but can not identify them,It is unable to get users' identities.Therefore,the tracking and classifing function has to be done by software.Multiple Target Tracking a Computer Vision problem and Multiple Hypothesis Tracking(MHT)is one of the earliest successful alogrithm applied to visual tracking.The image multiple target tracking depends on image identify processing and can not apply to radar system.The radar hardware can not provide fine data of body posture to build feature data,and there is no feature avaliable to tell targets from each other except their location coordinates.With location data avaliable MHT is able to establish offline tracking,but with out the online MHT the location data can not be processed at realtime.This paper modified MHT online tracking,with some precision loss the realtime tracking works and the online trajectory is appended to offline ones.In order to store and visualize trajectory data,it has to be compressed to remove redundant waypoints.When the target movement is slow the location data is centralized.If all waypoints are located in a straight line,only the two endpoint is required to represent that path.Douglas-Peucker compression algorithm is widely applied but its processing output is not capable enough to keep the origional location information.Based on Douglas-Peucker algorithm this paper takes waypoint density in to consideration,without extra time cost and compression error the compressed trajectory holds more origional position information.Radar posining error caused by distance and angle measurement may result in target coordinate stack into obstacles or thir moving trajectory cross the obstacles.To eliminate this error,radar installation error must be fixed,and extra waypoints may required to bent the path to get it out of the obstacles.This paper presents a trajectory fix algorithm based on Quick Hull,which can build convex polygon that surround scattered points.This algorithm take body size as additional parameter to build a surrounding trajectory.The positioning system should avoied undetectable corner when being deployed and use as less device as possible.Deployment location depends on indoor space and installation equipments.If any obstacle gets into detection range the radio cover area calculation becomes complex.Ray tracing method is able to build propagating cover model that is easy to calculate and simpler than the complete geometry model.By applying transforming matrix,linear interpretation that is commonly used in computer graphics and utilizing hardware and software interface,the sight covering computing can be done on GPU to get accelerated.The multiple device covering can be process by same method with extra obstacle lines added to covering area edges.After covering area at each position of candidate zone is calculated,the optimal deployment solution can be selected.