Target Localization And Tracking Study On Unreliable Bearing Measurement

In the field of target localization and tracking,the sensor network system based on acoustic arrays has the advantages of low energy-consumption and good concealment,making it important application in field monitoring,military defense,and some other fields.Because of many uncontrollable factors in actual environment and equipment,it is too difficult to ensure that each array gives completely reliable measurement.Therefore,how to obtain robust target location estimation in the case of unreliable measurement is a task that needs to be focused on.In this paper,we focus on the acoustic sensor array network system with bearing measurement.In the case of unreliable measurement,we give robust localization solutions respectively for single-target scenario,single-moving-target scenario,and multi-target scenario.The main work of this paper includes the following aspects:Firstly,in single-target scenario,the concepte of target feasible region is introduced,and an outlier removed method based on the feasible region evaluation and a l1-norm based robust localization method are designed.To solve the designed robust method fast,a searching approach based on region splitting and pruning is proposed.Secondly,in single-moving-target scenario,on the hypothesis of non-maneuvering target doing approximate uniform linear motion within short time,the array reliability and bearing measurement reliability are deduced in probabilistic way.A spatio-temporal joint l1-norm improved method based on exisiting ML and LS method is proposed.With backword evaluation mechanism,an effective anti-deceptive robust localization method is introduced.Finally,in multi-target localization scenario,the localization estimation may contain many ghosts caused by wrong measurement associations,for the measurement-target associations are unknown.To remove wrong associations,some sufficient conditions for determining the wrong measurement associations are presented.An outlier detection method is proposed by extending target feasible region evaluation mechanism.At last,we design a multi-target localization method based on clustering to further supperss the effect of few residual outliers and wrong measurement associations.