Study On Location Algorithms And Nonlinear Filtering In Bistatic(Multistatic) Radars

In modern wars, bistatic (multistatic) radars have so great advantages that normal radars can't match. Especially in resisting the big four threats: electronic jamming, invisible planes,anti-radiation missiles, low altitude penetration. Bistatic radars actualize the separation of transmitting station and receiving station.Using this method, we can widely increase the bistatic (multistatic) radars'existence ability in the warfares.In the battlefield, we can dispose cooperative multistatic radars to find invisible planes(by detecting its RCS),if we put different frequency,wavelength,and the mix of impulse wave and continuous waves in multistatic radars,then we will realize comprehensive functions.As the impaction of the big four threats is getting more and more deep,bistatic(multistatic)radars strongly attract people's attention. We mainly focus on these following aspects:location,imaging,tracking,and different platforms with transmitting station.Above all, we are firstly interested in acquiring the information of target's position.In this article, we will study on localization algorithm and nonlinear filter based on bistatic(multistatic) radars.The application of bistatic (multistatic) radars is very wide, besides mainly used in military, bistatic(multistatic) radars are widely used in earth exploration, weather forecast, underground detection. So in many fields, bistatic (multistatic) radars have more and more important applications.In this paper, we first present the basic conception and interrelated parameter of bistatic (multistatic) radars, then we present the basic method and property ofestimation theory.At the end of chapter 2,we emphasize on the criterions of location's precision,these works will lay a foundation for the following research. In chapter three, we focus on the method by using time difference of arrival to get the target's position estimation. The core of the algorithm is converting nonlinear equations into linear equations.Chan algorithm and Taylor series expansion algorithm are the two classical algorithms,and they are maturely applied in the wireless location of cellular networks.In this chapter, we try to use these algorithms in the location of bistatic (multistatic) radars.Also.I will extend complanate's case to three dimensional case.By more, the improvement algorithms will be put forward: combining with Chan algorithm and Taylor algorithm in order to make full use of the two algorithms.On one hand,we can get precision iterative initial value,on the other hand,it can be adapted in non-Gaussian condition.What's more,in this chapter, a new method will be put forward in three-dimensional location of bistatic(multistatic) radars:combining TDOA with AOA.By this way, time datas and angle datas can be used fullly.Through emulation,we can prove that if getting accurate AOA datas,the total location precision can be quite high.In some instance,we usually can not get to know transcendental information of relative parameters.At this time,least square method is our best choice.In chapter four,we detailedly discuss least square method used in bistatic(multistatic) radars.During the course of study,I analysis two cases:no altitudinal difference using the method of reducing dimension and tiny altitudinal difference using iterative method.Besides,in the paper,I also analysis weighted least square method,its core is grouping the observation datas.In some extent,it can restrain the limitation that LS method can't make full use of observation datas'statistical feature.In the course of grouping datas,we stress on using observation datas repeatly owing to lacking all the observation datas.In chapter four,I analysis the instance of repeatly using slant range and azimuth of transmitting station.I prove that using observation datas repeatly will result in weighted singular matrix,so in this chapter four,I try to add a tiny amended gene into weighted matrix in order to make it a nonsingular matrix.Through Matlab emulation we can acquire quite accurate location.As the result of the use of passive positioning,bistatic(multistatic)radar's observation equations are bound to bring about strong nonlinear,so it's necessary to study the problem of nonlinear filtering.The present study of nonlinear filtering is mainly focused on two aspects:one thought is based on linear method,such as EKF;the other thought is based on sample filter such as UKF.In chapter 5,first I detailedly analysis the basic principles of EKF and UKF algorithms. EKF is simple, reserve the first order of Taylor expansion, but simple will lead to the loss of the precision,so EKF is only applicable to weak nonlinear, or less demanding on the accuracy occasions.UKF method use the idea of sampling to select a number of right points that fully reflect the mean and variance of the probability density function ,these points can reflect the characteristics of nonlinear function after being passed through nonlinear function and combining with corresponding weighted values.Compared to EKF,UKF is clearly improved in accurancy,and this is main concern of chapter five.In chapter five,two improvements about UKF algorithm are proposed:1. Propose a self adaptive UKF algorithm.Unlike previous UKF algorithm that insert a simple weighted gene into observation noise,the self adaptive UKF algorithm insert three weighted gene into three directions.By doing this,we can make it more in line with actual situation of target's maneuvering.2. Propose an improved UKF algorithm with attenuation gene.The core is adding an attenuation gene in the noise covariance(in this paper,we choose constant), by this way,we can increase the weight of the new observation datas,gradually weaken the impact of the old observation datas in data processing.