UAV Localization System Based On DOA Estimation Of Radio Frequency Signals

In recent years,the civilian drone industry has developed rapidly,but the control methods for such drones are relatively lagging behind.How to effectively deal with illegally invaded drones has become a pain point for the industry.The premise that most countermeasures against drones can be effectively implemented is that the drones can be detected and located.Aiming at this problem,this paper designs a small UAV positioning system based on the Direction of Arrival(DOA)estimation of the RF signal,which can receive the RF signal from the UAV and estimate its two-dimensional position.Specifically,the main work of this paper is as follows.1.This paper proposes a UAV positioning system structure that uses an antenna array and a super-heterodyne receiver to receive signals from a specific drone band,and uses a Field Programmable Gate Array(FPGA)for data acquisition and high-speed signal processing.The implementation of the entire system have been completed.The system can not only realize DOA estimation in real time,but also have a series of functions including frequency hopping and antenna array calibration.2.This paper proposes a hardware acceleration method of DOA estimation algorithm.Specifically,DOA estimation involves eigenvalue decomposition.In order to solve the problem of using Jacobi algorithm and consuming a long time to implement eigenvalue decomposition,a parallel Jacobi-based accelerated computing method based on FPGA is proposed.Each iteration step of the traditional Jacobi algorithm requires the use of three Coordinate Rotation Digital Computer(CORDIC)algorithm cycles.In this paper,a method based on prosthaphaeresis formula and trigonometric transformation is proposed.Combined with symbol set,the time-consuming of each iteration step of Jacobi algorithm is reduced to about one period of CORDIC algorithm.The UAV positioning system proposed in this paper can locate the UAV in the range of 400 meters in two dimensions,and the root mean square error of angle estimation is within 7 degrees.The hardware implementation method of eigenvalue decomposition proposed in this paper is three times faster than that of classical algorithm.