Research On Wireless Signal Source Location Technology Based On Nonparametric Kernel Regression Prediction

Positioning wireless signal sources in a known range,existing positioning technologies and related research mainly focus on positioning receiving terminals.For the positioning of positioning wireless signal sources,the existing technologies and related research are few and have many limitations: TDOA(Time Difference of Arrival)and TOA(Time of Arrival)are quite strict in time synchronization;DOA(Direction Of Arrival)/AOA is higher in equipment requirements;SSR(Signal Strength Ranging)anti-interference ability Not strong.In view of the proliferation of pseudo-base stations and malicious broadcast,this paper improves the existing field strength prediction algorithm by using K nearest neighbor classification algorithm after studying the source and signal propagation characteristics.Furthermore,this paper further studies and implements a wireless signal source localization algorithm based on prediction model.The algorithm does not need time synchronization,has lower requirements on test equipment,and has certain anti-interference ability.After implementing the algorithm,our paper completed the design,development and testing of the wireless signal source location system based on the prediction model.The main work and innovation of the thesis are as follows:(1)We deeply studies the previous algorithms for electromagnetic radiation prediction and wireless communication field strength prediction,and improves the existing field strength prediction algorithm.The improvement is mainly to narrow the prediction error on the basis of the predecessors,so as to greatly improve the prediction accuracy.The experimental results show that the prediction accuracy of all signal systems involved in the test has reached more than 80%,and the prediction accuracy of most signal system is still more than 90%,and the error range is only within ±2dB.(2)Based on(1),We have already studied the characteristics of wireless signal propagation.Based on the prediction results of(1),we propose a method for locating wireless signal sources.The method uses the shrinking circle method to gradually reduce the range in which the signal source may exist,and the shrinking ring motion stops after reaching the predetermined precision requirement and gives the final result.After many tests,the program accurately found the location of the wireless signal source,and the final positioning area is no more than 1% of the original scene area.(3)Based on the algorithms in(1)and(2),for the purpose of applying algorithms,we proposes system design requirements.The system model is designed according to the design requirements,and the relationship between the module and the module is determined.The algorithm application is running on Windows7 64 bit operating system,using Microsoft Visual Studio 2015 as the development tool,Qt5 library as the interface framework,C++ as the writing language to develop the wireless signal source positioning system.The system provides functions such as loading scenarios,parameter settings,prediction,and signal source positioning to facilitate experiments and verification of algorithms in(1)and(2).(4)The algorithms of(1)and(2)and the wireless signal source localization system implemented in(3)are used as software means to test the accuracy,performance,prediction and positioning effect of the algorithm,and analyze the test results.Many experiments have proved that the improved K-nearest neighbor nonparametric kernel regression algorithm improves the prediction while reducing the error range,and contributes to the prediction of wireless communication field strength.The wireless signal source localization method proposed in this paper has been tested and verified,which provides a new idea for wireless signal source location.