Research On Single Link Passive Target Tracking Algorithm Based On WiFi

With the advent of the intelligent era,indoor passive target tracking has become a research hotspot as a key technology for intelligent applications.Because WiFi commercial devices are widely deployed and can protect user privacy well,WiFi-based passive target tracking technology stands out from numerous tracking technologies.At present,most homes and convenience stores only deploy one WiFi device and cannot form multiple wireless links to complete target tracking.In order to meet the application requirements of the above scenarios,single link passive target tracking technology based on WiFi is required.The existing single link passive target tracking technology based on WiFi has the following problems: First,the use of motion feature matching for target tracking requires the collection and maintenance of an offline database,which will consume a lot of manpower and resources,and is vulnerable to environmental changes.Second,finegrained target motion tracking requires modification of existing WiFi commercial equipment hardware,which is not conducive to large-scale deployment and use.Third,target tracking based on signal parameter estimation ignores phase errors and noise effects in the signal,resulting in low parameter estimation accuracy and large target tracking error.Based on this,this paper conducts research on single link passive target tracking technology based on WiFi.The main work is as follows:First,research on algorithms for joint estimation of multi-dimensional path parameters.Conjugate multiplication is used to reconstruct channel state information(CSI)to eliminate phase errors and static path signal components firstly,and then serial interference cancellation is used to complete the multi-dimensional parameters initialization of multiple paths.Finally the traditional Frequency Domain Space Alternating Generalized Expectation-maximization algorithm is improved to complete the joint estimation of multi-dimensional parameters of multiple paths.Second,research on the algorithms of selecting target reflection path parameters.A hybrid data association method based on the sliding window mechanism is adopted to solve the problem that the existing algorithms cannot be associated due to the inconsistent number of paths at adjacent moments.First,build a multi-path network of path parameters at a selected time window,and then complete the correlation of the path parameters in the window based on the minimum cost criterion.Next,the similarity is used to associate the path parameters at the first moment in the window with the path parameters at the previous moment.Finally,the path parameter with the largest average attenuation is selected as the target path parameter.Third,research on the algorithm of target tracking model construction.Firstly,the Time of Flight is regarded as an observable variable,and the Doppler Frequency Shift is regarded as an unobservable variable.The distance re-estimation is completed in combination with the linear Gaussian state space model.Then,the obtained distance and the angle of arrival are used to construct a geometric model to achieve target tracking.In a complex indoor environment and corridor environment,a real platform is used to collect data and verify the algorithm.The experimental results show that the algorithm can achieve a tracking accuracy of 1.2m at a confidence level of 60% without the need to collect offline databases and modify hardware devices.