Research On Key Technologies Of High-resolution Millimeter-wave Radar

With the advent of the era of 5G and autonomous driving,the term "millimeter wave" has slowly entered people's vision.With its unique advantages,millimeter-wave radar occupies a place in the automotive sensor market.The process technology has also evolved from the original Ga As and Si Ge processes to CMOS processes.The size reduction and cost compression are more conducive to exploring the low-end market.In addition,as people pay more attention to privacy,millimeter-wave radar has gradually emerged as a non-invasive sensor in the fields of medical equipment,security and industrial measurement.High-resolution millimeter-wave radar mainly refers to improving the angular resolution by MIMO or cascading to provide finer scene information.Firstly,the complete signal processing flow and related algorithms are discussed from the waveform timing.The form of range-Doppler matrix and the resolution of range and velocity are deduced.Focusing on the research of the target detection algorithm,in addition to the simulation of the classic CA-CFAR and OS-CFAR,it also introduces a twodimensional OSCA-CFAR algorithm that takes into account the amount of calculation and detection accuracy for crowded scenes such as roads or indoors.The simulation proves that it can overcome the shadowing effect in the direction of distance,and can guarantee the accuracy of estimating the signal-to-noise ratio in the direction of Doppler.The principle of MIMO and two orthogonal waveforms of time division multiplexing and two-phase modulation are briefly introduced.The difference between the angle estimation achieved by the classical beamforming algorithm and the angle-dimensional FFT algorithm is compared.Secondly,since high-resolution millimeter-wave radar can form multiple scattering points for a single target,an extended Kalman filter suitable for point targets is no longer feasible.According to this,the theoretical research and simulation of the multi-point multitarget tracking algorithm based on EKF,compared with more complex particle filtering,random sets and other high-order algorithms,this algorithm requires less computation and can be practically run on Embedded devices.The algorithm uses extended Kalman filtering as the core to complete the filtering of the target centroid,and discusses data association,threshold function,cluster analysis algorithm,track management,etc.Finally,a comprehensive simulation test of the multi-objective algorithm is performed,including error analysis,convergence analysis,track failure,and track merging.Then introduced the realization of multi-target tracking system based on ARM+DSP,including hardware architecture,software design and host computer writing.The design site experiment conducted an integrated test on the system.The experiment proved that the algorithm can establish tracking filters for multiple targets at the same time,and can properly deal with the situation of track failure,track crossing,and false targets.