| With the development of economy and the advancement of artificial intelligence algorithms in recent years,self-driving technology,as an effective mean to improve the traffic safety and social production,has become a significant research highlight in academia and industry.Millimeter-wave(mm-wave)radar plays a non-substitutable role among self-driving sensors due to its small size,low cost and superior performance in severe weather condition.However,limited by the cost of the system radio frequency(RF)and the baseband hardware,it is difficult for existing medium-range automotive mm-wave radars to acquire satisfactory pa-rameter estimation performance in low complexity systems.Meanwhile,compared to medium-range radars,short-range automotive mm-wave radars with higher range resolution require more advanced algorithms to achieve higher system performance.Therefore,this thesis has carried out the following researches.(1)The object detection and parameter estimation are studied for automotive mm-wave radar system.For the high computational complexity in traditional constant false alarm rate(CFAR)detection algorithm,an improved CFAR algorithm based on assistant threshold surface is proposed to reduce the number of detected units and thereby decreases the calculation complexity.For the straddle loss in the two-dimension(2D)fast Fourier transformation(FFT)results of system intermediate-frequency(IF)signal,an improved algorithm is proposed for the acquisition of object phase information.In this algorithm,the objects'range/velocity estimation values corrected by the centroid method are used to calculate the discrete Fourier transformation(DFT)rotation factors in fast/slow time dimension.Then,to decrease the loss of signal-to-noise ratio(SNR),the 2D-DFT is performed on the IF time domain signal to extract the objects'phase information.For the high computational complexity and low time utilization in existing solutions for velocity ambiguity,a novel algorithm based on signal multiplexing is proposed for wide-band systems,where the carrier frequency difference between the main signal and the sub-signal is utilized to unwrap the ambiguity.The proposed algorithm has lower complexity and does not require additional assistant signals.(2)The virtual array algorithm for automotive mm-wave radar is studied.Since the phase compensation procedure of existing time-diversity virtual array algorithm cannot be directly applied in the coherent objects case,an improved phase compensation algorithm is proposed to compensate the motion phase error between the virtual arrays via using the estimated objects' velocity value.The proposed algorithm has lower complexity and is effective for coherent objects.For the power gain loss in the time-diversity virtual array algorithm,a space-time block code(STBC)based algorithm is proposed.Via decoupling the encoded IF signal to form a virtual array with larger antenna aperture,the STBC algorithm can achieve the power and array gain by simultaneously transmitting signals from multiple antennas under the power constraint of single antenna.(3)The calibration algorithm of motional frequency spread is studied for wide-band automotive mm-wave radar.For the motional frequency due to the relative motion of object and radar in the wide-band automotive mm-radar systems,the analytical expressions of the motional frequency interference term and phase interference term in the IF signal are derived firstly.Then,the effect of frequency spread is comprehensively analyzed with the help of numerical results.Motivated by the mechanism and influence of frequency spread,three motional frequency spread calibration algorithms are proposed with different performance and complexity tradeoff,which are inverse discrete Fourier transformation frequency calibration(IDFT-FC)algorithm,spectrum cell rearrangement(SCR)algorithm,and simple frequency correction(SFC)algorithm.Simulations and complexity analysis show that IDFT-FC has the optimal calibration performance,the SCR algorithm provides a flexible tradeoff between performance and implementation cost,and the SFC algorithm can only improve the estimation accuracy of range and velocity but with nearly no additional computation and storage.(4)The implementation of the 77GHz automotive mm-wave radar system is studied for the TI-AWR1443chip.For the problem that the sampling resolution in such low-cost chip does not meet the SNR dynamic range of objects,a low-complexity dual-mode automotive mm-wave radar frame structure design method is proposed.Based on the proposed method and the chip parameters of TI-AWR1443,the transmission signal frame and the automotive mm-wave radar signal processing module are designed with the improved algorithms proposed in this thesis.The implemented automotive mm-wave radar system has a maximum detection range of180m,a range resolution of 0.75m and a detection range of azimuth±50°.The actual test and data analysis of the automotive mm-wave radar system based on TI-AWR1443 verify the effectiveness of the system. |
PDF Full Text Request