极大似然和MUSIC算法用于DOA估计的DSP实现

【作者】 李军；

【导师】 钟洪声；

【作者基本信息】 电子科技大学 ， 电路与系统， 2005， 硕士

【摘要】 在阵列信号处理领域,近十几年来发展起来很多波达方向(Direction of -Arrival,DOA)估计的超分辨算法。空间谱估计的算法大致分两大类:一是基于极大似然估计和最大后验概率估计统计理论的算法,包括:极大似然估计法(ML)、最大熵法等;另一类是基于对协方差矩阵进行子空间分解或投影的算法,包括:矢量特征法、多重信号分类法(MUSIC)、投影矩阵法等。其中,MUSIC 法是一种经典的空间谱估计主流算法,具有超强的分辨性能,但它无法实现对相干信号进行测向分辨。极大似然估计算法能够对相干信号源进行测向估计,但却又有运算量过于庞大的缺点。本文所涉及的空间谱估计测向系统,以浮点处理器ADSP21160 为测向处理平台,在非相干信号源情况下利用MUSIC 算法完成测向,在相干信号源情况下利用极大似然估计算法和AP(Alternating Projection)叠代法相结合的办法实现测向。采用基于信息论原理MDL 准则或者AIC 准则判断信号源数目。而对复数矩阵的特征分解则先后采用Householder 和QR 方法。DSP 与前端接收机间通过链式DMA 进行数据传输,处理结果经异步串口通信在PC 机上显示出来。上述各种算法的原理、实现步聚及程序流图都在文中进行了描述。在对方位角和仰角进行二维搜索时,本文提出了“加窗”处理办法。“窗口”为矩形窗,它能保证在二维搜索空间的任何一个“窗口”范围内最多含有一个来波的角度信息,可以避免估计出来的多个来波方向实际上为同一来波的信息;本文从原理和实验证实了这种办法的有效性。本文还介绍了在VisualDSP环境下借助ADSP21160相关的评估板和仿真器等工具,对该测向系统DSP 软件进行单独调试和软硬件联调的方法和步骤及调试中遇到的问题及其解决办法。更多还原

【Abstract】 In the recent decade many high-resolution DOA (Direction Of-Arrival) algorithms have been developed in the field of Array Signal Processing. There are mainly two type of algorithms used for spatial spectrum estimation: one is those based on Bayesian Maximum Likelihood method, like the ML (Maximum Likelihood) algorithm, Maximal Entropy method and etc., the others are based on the spatial decomposition or projection of correlation matrix, this kind of algorithm include Vector Characterization Method, MUSIC (multiple signal classification) algorithm, Projection Matrix Method, etc. MUSIC is a classic spatial spectrum estimation algorithm that has a super high resolution and is widely used today, however, it cannot estimate successfully the azimuths and elevations of signals that are correlated. The Maximum Likelihood algorithm can be used for the DOA estimation of correlated signals, but it needs a huge computation and long run time. In this paper, MUSIC is chosen as the algorithm for the DOA estimation of non-correlated signals and ML as the algorithm for the correlated signals. To determin the number of signals, we adopt the rule of MDL or AIC which are based on the information theory. And we use Householder transfer method and QR decomposition method to accomplish the eig-decomposing of Hermite matrix, which is needed in the MUSIC algorithm. The above-mentioned algorithm and their implementation procedures are fully discussed in this paper. ADSP21160 is used as the digital signal processor in this spatial spectrum estimation system. It communicates with the front-end receiver and data are transferred into the DSP RAM through chain DMA. The result of DOA estimation is transmitted to the PC by UART port communication and displayed on the terminal display unit. To improve the veracity of the result, we introduced a "window" into the 2-dimention searches of the azimuth and elevation. The "window" is a rectangle, which ensures no more than one signal appear in each “windowed”2-dimension region. By this way we can decrease the probability that more than one computed output represent the same signal while some other signals missing. And we theoretically analyzed it and prove this way effective in the paper. Take advantage of the ADSP21160-concerned Evaluator and Emulator we debugged the software system and the whole spatial spectrum estimation system, and solved all the problems that appeared.更多还原