The Design And Implementation Of Improved Particle Filter Algorithm For RSSI Indoor Positioning

With the development of Internet of Things and wireless technology,outdoor positioning with GPS can no longer meet people's needs,and more and more people are paying attention to indoor positioning technology.The indoor positioning based on RSSI(Received Signal Strength Indication)of Bluetooth has become one of the research hotspots due to its low cost.However,due to the influence of the surrounding environment,it is prone to fluctuations,low precision,and unstable performance.Therefore,in order to make the RSSI-based positioning result more stable and reduce the measurement error,it is necessary to introduce a corresponding filtering algorithm.In this thesis,the particle filter was used to process the RSSI,and the smart phone was used as the positioning terminal.While mobile phone resources are limited and the calculation of particle filter is complicated.Therefore,this thesis studies how to improve the speed and accuracy of the results of the particle filter algorithm for RSSI indoor positioning.The RSSI indoor positioning algorithm based on KLD(Kullback-Leibler Distance)adaptive particle filtering is researched and improved in this thesis.The traditional particle filter algorithm for RSSI-based indoor positioning based on KLD will make the number of particles not properly adjusted according to the actual environment if the constraint error is improperly selected,resulting in the system's average positioning error becoming larger.In response to this problem,a fingerprint library with particle number information is established.In the process of collecting fingerprints,the enumeration method is used to change the constraint error in a certain range according to the fixed step size for each location,and the positioning RMSE of the different constraint errors and the positioning RMSE of fixed particle number are compared and analyzed.The constraint error with the smallest RMSE is selected and calculate the number of particles at the location,then save this number of particles to the fingerprint database.Considering the different characteristics of RSSI fluctuations in different distances during the offline fingerprint collection process,the weight of WKNN fingerprint localization algorithm is improved by introducing mean value,standard deviation and direction information to get the improved observation value with less noise.In the actual test,the unknown node obtains the initial position through the improved fingerprint algorithm firstly,and uses the optimal number of particles under the current location to perform particle filtering to obtain the location coordinates.The actual test results show that this positioning system can achieve an average positioning error less than 1.61 meters using smartphone,and the average number of particles is 361.85.The average single-step positioning time interval can be achieved in 466 ms when the Bluetooth scanning frequency is 3 Hz.In total,the result can meet the requirements of index.The improved adaptive particle filter algorithm for indoor positioning based on RSSI in this thesis finally achieves the proposed index,and it has certain practical value for the for engineering application.