Study Of DOA Estimation And Its Hardware Implement Based On Space-Time Two-Dimensions MUSIC Algorithm

The purpose of this paper is to get the distance and angle information of obstacles in front of the radar. In this paper 14 array makes up uniform antenna linear array to send and receive the electromagnetic wave, and then to process signal data and do the frequency and angle estimation. At the same time, the theory of direction of signal arrival(DOA) and space-time two dimensions multiple signal classification(MUSIC) is adopted.Matlab is used to simulate and verify the feasibility of the project. Based on the simulation results, the resolution and accuracy of distance and angle have been gotten. In order to determine the hardware realization method, the method of dimensionality reduction of space-time two dimensions MUSIC is adopted, which consists of FFT transform and one dimension MUSIC. First, do 1024 points FFT transform for these obtained signals to get frequencies. Second, for every frequency point, one dimension MUSIC is used for angle estimation. There are four steps of the realization of one dimension MUSIC, which are the estimation of covariance matrix, eigenvalue decomposition, estimation of signal number, and peak searching.Although the simultaneous searching of frequency and angle is avoided after adopting dimensionality reduction of space-time two dimensions MUSIC, the problem that the computation is huge and the algorithm is complex still exists. In order to solve this problem, FPGA and DSP are both used as the hardware to implement the algorithm. FPGA is used to implement the FFT transform and the estimation of covariance matrix, just because it can finish huge computation which has rules. On the contrary, all of the estimation of signal number, the eigenvalue decomposition and the peak searching are complex algorithms, so DSP is used to implement them. On account of the requirement for the real time, the processing speed and the consumption of the chip resources, this design adopts Arria V chip of Altera Company as FPGA chip and TMS320C6455 chip of TI Company as DSP chip. The communication interface is EMIFA between FPGA and DSP.Ping pang operation, pipeline and state machine are used to control and improve the processing speed on FPGA, and the thought of sacrificing area in exchange for the speed is always adopted considering of the high real time requirements. For every steps running on DSP, the appropriate algorithm is selected to satisfy the implement of the chip and ensure the accuracy of the results. For example, MDL criterion is selected for the estimation of the signal number, and QR algorithm is selected for eigenvalue decomposition.All algorithms and methods being used are to cooperate with hardware to meet the requirements of the scene. This design basically completes the function of the estimation of distance and direction, and also achieves the accuracy and resolution requirements.