Multi-sensor Target Tracking Based On Information Geometry

With the rapid development of electronic technology,the electronic equipment in the battlefield environment is more and more diversified,and the types and quantity of sensors are increasing,which also makes the previous methods of using a single sensor to obtain information can no longer meet the operational requirements.Multi-sensor system has been favored by researchers for its advantages of complementary information and strong robustness.Sensor management is a key technology in multi-sensor target tracking system.It controls the whole system to make different actions for different tracking tasks.In this paper,the problem of sensor management in target tracking is studied.The research contents and main contributions are as follows:Firstly,considering that airborne sensor platform needs to hide itself while tracking the target in the battlefield environment,a radiation control method based on state transition equation and passive sensor measurement is proposed,which can predict the tracking performance.When the predicted tracking performance can not meet the demand,the active sensor is turned on,then the Fisher information matrix in the information geometry method is introduced to measure the tracking performance of sensors,the sensor with the largest Fisher information matrix is selected to track the target,and then the fusion algorithm based on particle filter is used to fuse the measured values of the multi-sensor to achieve the accuracy of the target tracking.The final results show that the radiation of the airborne platform is greatly reduced based on the slight increase of tracking error.Then,for a more general target tracking-oriented sensor network management problem,an information geometry method is introduced to analyze the tracking accuracy.Based on the current research on Rao distance,two new lower bounds of Rao distance are derived from the state space and measurement space respectively in the framework of target tracking,which can be applied to sensor management,and the lower bounds are taken as information incremental measurement is applied to sensor management to improve tracking performance.Finally,the research on sensor management under different parameters is carried out and the results are analyzed.After deploying different numbers of sensors in the monitoring area and conducting experiments,it was found that the tracking accuracy of the target increases with the increase in the total number of sensors deployed in the monitoring area,but the tracking accuracy is not a simple linear relationship with the total number of sensors,when the total number of sensors exceeds a certain number and then continues to increase,the improvement of tracking accuracy is extremely limited.When the number of selected sensors increases at each time,the tracking accuracy will also be improved,but the speed of improving the tracking accuracy will slow down with the increase of the number of selected sensors.At the same time,the greater the sensor noise,the smaller the Signal Noise Ratio,the worse the tracking performance.compared with the ranging noise,the tracking performance is more sensitive to changes in angular measurement noise.