Research On Clock Synchronization Algorithm In Ultra-wideband Indoor Positioning

People are nowadays getting a higher demand for perceiving environment surrounding them and getting information from it,which results in a wider application of wireless sensor network(WSN).WSN is now being used in multiple aspects of industries and people's lives.It is considered one specific implementation of what is now a trending technology,the Internet of Things(IOT).In applications like object localization and status tracking,risk monitoring and alarming of projects and intelligent control and energy conservation of systems,WSN plays an irreplaceable role.The key component of WSN is sensor.Usually there are plenty of sensors in a network and sensors communicate with each other in order to work cooperatively and update their status timely.In most cases WSN works in place isolated from outside world and because of that,the system is lacking of reference information.To make sure that all sensor nodes work cooperatively,it is of vital importance that clocks inside each sensor node synchronize to a reference clock.WSN contains sensor nodes each of which maintains an individual clock.These clocks are all ticking in different phase as well as speed and aging rate,and the goal of clock synchronization is to keep clocks of all sensor nodes ticking in the same phase and speed.Typical clock synchronization algorithms like Reference Broadcast Synchronization(RBS)and Delay Measurement Time Synchronization(DMTS)connects each individual clock by establishing linear relative clock model: first,one reference sensor node is selected and its clock is set as standard within the network and then all the rest clocks are converted to the reference one.Factors influencing the synchronization accuracy of the above-mentioned algorithm include the hardware delay of sensor nodes,clock jump and multipath effect.Besides,describing the running of clocks with linear functions brings model error.Proper ways to handle these factors will improve the synchronization accuracy remarkably.This paper research on the factors mentioned above and provide solutions to each of them.The platform used in this paper is indoor positioning system based on Ultra Wide Band(UWB)signal and Time Difference of Arrival(TDOA)mode.Besides,a software doing the work of positioning is designed.The software works as one important module of the entire indoor positioning platform.The specific work done includes:(1)Designing an algorithm and corresponding experiment for calibrating hardware delay.Traditional algorithm is based on the two way communication between base stations,while the algorithm proposed in this paper doesn't require the change of one way communication of TDOA mode.Experiment result shows that error of hundreds of meters will be introduced if hardware delay doesn't get calibrated, while regarding the delay as invariable cause error of centimeters,which is far more precise.In TDOA localization application,correct calibration of hardware delay is an important fundamental,so the calibration algorithm proposed is of practical value;(2)Proposing step-increased detecting method to solve the problem of how to increase the accuracy of linear relative clock model.The method increases the number of clock data by one each time and error of fitting and predicting are calculated respectively.By comparing all the error,the number with the best error performance are selected.As the period of communication between stations keeps unchanged,this method actually tells how many periods of data will best fit the linear relative clock model;(3)Processing abnormal clock data with method of threshold as well as sorting and screening.Two types of abnormal clock data,clock jump and multipath value,are handled.Result shows that when the threshold are set correctly,clock jump can be removed effectively and that when method of sorting and screening are used to handle clock data deeply influenced by multipath effects,modeling error reduces by 60% and stability increases by 30%.The result also shows that when normal clock data are handled with the method,modeling accuracy won't degrade.(4)Elaborating principles and methods of how to establish a platform of UWB indoor positioning.A software is designed to work as the positioning calculating part of the platform.The software is of integrity and stability,and its functions include communicating with hardware system as well as management platform,parsing and packing original observes,positioning and error handling and reading and writing of files.By testing,the designed software can provide positioning results steadily and is capable of interacting with hardware system and management platform, which proves that the software can work as an important part of the whole UWB indoor positioning platform.(5)In order to further verify the effects of the mentioned measures,this paper compared the modeling error of optimized linear clock model applying these measures with that of RBS and modified DMTS algorithm.The result indicates that as multipath effects has limited influence,RBS algorithm expresses lower fitting accuracy and that predicting error of modified DMTS algorithm is higher than that of the other two.When there exists severe multipath effects,using optimized linear clock model will reduce error to half of that of modified DMTS algorithm.Finally,the proposed optimized linear clock model and RBS algorithm are applied in TDOA indoor positioning.The positioning result shows that in general cases,RBS algorithm reaches slightly better positioning accuracy than optimized linear clock model,which is due to the incomplete calibration of hardware delay.But as for positioning stability,result with optimized linear clock model is 2 times better than that of RBS algorithm.When the clock data is of good quality,positioning accuracy with optimized linear clock model can be better than 10 cm.