非合作双基地雷达自适应波束形成技术研究及工程化实现

发布时间：2018-07-16 13:39

【摘要】：自适应波束形成技术是一种空域的自适应滤波技术,作为阵列信号处理的一个重要的研究方向,自适应波束形成技术经过几十年的发展,其基本理论与算法已经相当成熟,但在其工程化实现的过程当中,还存在大量问题没有得到解决。本文依托课题,以工程化实现为目标,针对波束形成过程中算法运算量与传输数据量大的问题,围绕着波束域降维、子空间降秩和宽带恒定束宽等几个方面对自适应波束形成算法进行了研究,并基于多核DSP硬件平台对提出的算法进行工程化实现。主要工作概括如下:1.采用波束域降维与子空间降秩的方式来解决窄带自适应波束形成算法计算量大的问题。首先,把最小方差无失真响应波束形成算法(MVDR)与广义旁瓣相消器(GSC)推广到波束域,在获得与传统算法相同的波束形成性能的同时,通过波束域降维处理来降低算法的计算量。然后,在GSC的框架下,将波束域降维方法与子空间降秩方法结合起来,提出了一种GSC框架下波束域快速自适应波束形成算法,并给出了该算法的实现结构。该算法通过一次特征分解降低了矩阵求逆的计算量,并通过子空间投影来改善算法性能。2.采用频域宽带恒定束宽的波束形成方法来解决宽带信号波束形成之后发生畸变的问题,并用窄带快速算法来降低宽带自适应波束形成的计算量。首先,给出了两种频率不变宽带波束形成算法——空间重采样法与FFT插值法,通过比较发现空间重采样法的聚焦性能要优于FFT插值法。然后,将频域宽带恒定束宽波束形成技术与窄带快速自适应波束形成技术结合起来,提出了一种宽带恒定束宽快速自适应波束形成算法,并给出了其两种实现结构。算法在保证恒定束宽性能的前提下,显著降低了计算量。3.依托课题背景,在多核DSP高速信号处理平台上对提出的算法进行工程化实现,设计了主从模式与数据流模式相结合的多核DSP并行处理过程。处理结果表明该方法求权精度高,波束形成速度快,易于在多核DSP上实现。
[Abstract]:Adaptive beamforming technology is a spatial adaptive filtering technology. As an important research direction of array signal processing, adaptive beamforming technology has been developed for decades, its basic theory and algorithm has been quite mature. However, in the process of its engineering, there are still a lot of problems have not been solved. Based on the subject, this paper aims at engineering realization, aiming at the problem of large amount of computation and transmission data in the beamforming process, focusing on the dimension reduction of beamspace. Adaptive beamforming algorithm is studied in several aspects, such as subspace rank reduction and broadband constant beam width, and the proposed algorithm is realized by engineering based on multi-core DSP hardware platform. The main work is summarized as follows: 1. Beamspace dimensionality reduction and subspace rank reduction are used to solve the problem of large computational complexity of narrowband adaptive beamforming algorithm. Firstly, the minimum variance distortionless response beamforming algorithm (MVDR) and the generalized sidelobe canceller (GSC) are extended to beamspace. The beamspace dimensionality reduction process is used to reduce the computational complexity of the beam-space algorithm while obtaining the same beamforming performance as the traditional beamforming algorithm. Then, a fast adaptive beamspace beamforming algorithm under GSC framework is proposed by combining beamspace dimensionality reduction method with subspace rank reduction method, and its implementation structure is given. The algorithm reduces the computation cost of matrix inversion and improves the performance of the algorithm by subspace projection. A frequency-domain beamforming method with constant beamwidth is used to solve the problem of distortion after beamforming of wideband signals, and narrow band fast algorithm is used to reduce the computational complexity of wideband adaptive beamforming. Firstly, two frequency invariant wideband beamforming algorithms, spatial resampling and FFT interpolation, are presented. It is found that the focusing performance of spatial resampling is better than that of FFT interpolation. Then, combining the frequency domain constant beamwidth beamforming technology with narrow band fast adaptive beamforming technology, a broadband constant beamwidth fast adaptive beamforming algorithm is proposed, and its two implementation structures are given. On the premise of guaranteeing the constant beam width performance, the algorithm reduces the computation cost by 3. 3. Based on the background of the project, the proposed algorithm is implemented on the multi-core DSP high-speed signal processing platform, and the parallel processing process of multi-core DSP is designed based on the combination of master-slave mode and data flow mode. The processing results show that this method has high weight accuracy, fast beamforming speed and is easy to be implemented on multi-core DSP.
【学位授予单位】：国防科学技术大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TN957.51

【参考文献】