分布式MIMO雷达目标定位与功率分配研究

发布时间：2019-03-31 14:43

【摘要】：多发多收(Multiple-Input Multiple Output,MIMO)雷达是雷达领域研究前沿的热点问题之一。按照天线间距分为集中式和分布式MIMO雷达,集中式MIMO雷达利用波形分集增益提高雷达系统的角度分辨率,分布式MIMO雷达利用空间分集增益对抗目标的RCS闪烁。分布式MIMO雷达能够从不同的观测方向对目标进行探测,为解决常规雷达面临的隐身目标难以检测以及抗干扰能力差等实际问题提供了有效解决途径,对分布式MIMO雷达所取得的初步探索成果已经显示了它在目标检测和参数估计方面的巨大优势。然而,对于分布式MIMO雷达的处理技术、性能评估以及系统设计研究成果较少,诸多概念以及关键技术问题还有待深入研究。本文围绕分布式MIMO雷达的目标定位和功率优化展开研究,主要致力于解决现有目标定位方法和功率优化分配方法所面临的关键问题。具体地,本文的研究内容包括如下几个方面:第一章为绪论,阐述了课题研究背景及意义,介绍了MIMO雷达尤其是分布式MIMO雷达的研究现状,分析了目前分布式MIMO雷达参数估计和功率分配面临的问题,并指出了开展目标定位方法和功率优化管理研究的可行性。第二章主要研究了分布式MIMO雷达的目标定位性能和它对雷达天线位置误差的敏感性问题。根据似然函数与似然比的关系得到似然函数,将目标定位问题抽象为位置参数的最大似然估计问题,推导了了目标位置参数的Fisher信息矩阵和克拉美-罗下界。针对雷达天线位置存在测量误差可能导致定位性能下降的问题,首先利用一阶泰勒近似推导了忽略天线位置误差时目标位置估计的均方误差,然后基于统计独立的思想推导了同时考虑天线位置误差和目标位置参数的联合克拉美-罗下界,定量分析了天线位置误差对定位精度的影响,为后续研究奠定了理论基础。第三章针对单目标定位场景,主要研究了基于半定规划理论的分布式MIMO雷达间接定位方法。将基于双站距离观测的间接定位建模为非线性最小二乘问题,通过一阶泰勒近似推导了线性化定位方法。为了从理论上保证获得全局最优解,借助凸优化思想提出了基于半定规划的目标定位方法。该方法通过引入多余变量,将非线性最小二乘问题转化为带约束的凸优化问题,然后通过半定松弛技术对非凸约束条件进行松弛再转化为可解的半定规划问题。在该框架下,还推导了存在天线位置误差情况下的半定规划定位方法,该方法大幅度降低了天线位置误差带来的影响,提高了目标定位的稳健性能。第四章针对多目标定位场景,主要研究了基于稀疏重构理论的分布式MIMO雷达直接定位方法。将基于匹配滤波信号的直接目标定位建模为一种块稀疏表示模型,目标定位转化为块稀疏重构问题。为解决上述问题,在块稀疏贝叶斯学习框架下发展出一种多目标定位方法,该方法利用块内的相关性,可以提高重构精度。实验结果表明,所提算法课不需要数据关联处理,具有处理密集目标、压缩采样等条件下定位问题的能力。另外,针对相参处理中可能存在的相位不同步问题,在块稀疏贝叶斯学习和最大似然估计框架下提出了一种迭代算法来解决上述问题,具体来说,迭代利用最大似然估计出相位误差和块稀疏贝叶斯学习重构出目标散射系数,该算法在相位失配情况下体现出良好的误差校正能力和较高的定位精度。第五章主要研究了分布式MIMO雷达针对目标定位的功率分配问题。针对间接定位模型,严格推导了目标位置的贝叶斯Fisher信息矩阵,详细给出了随机可观测性的粒子近似计算方法。以随机可观测性为目标函数、发射总功率为约束条件,将功率优化分配建模为非凸二次问题。为解决上述问题,阐述了合作博弈理论中沙普利值的概念、性质和求解,在该框架下提出了一种基于沙普利值的功率分配方法,给出了迭代实现流程。该方法利用加权图理论简化了沙普利值的计算过程,且具有明确的物理意义。仿真结果表明,所提算法通过合理优化分配发射功率,有效提高了目标定位精度和资源使用效率。第六章对本文的研究工作和主要创新点进行了总结,并指出了下一步可能的研究方向。
[Abstract]:Multiple-Input Multiple Output (MIMO) radar is one of the hot issues in radar field. According to the antenna spacing, the centralized and distributed MIMO radar is divided into a centralized MIMO radar and a distributed MIMO radar, the angular resolution of the radar system is improved by utilizing the waveform diversity gain, and the distributed MIMO radar uses the space diversity gain to fight the RCS of the target. The distributed MIMO radar can detect the target from different observation directions, and provides an effective solution for solving the practical problems such as difficult detection of the stealth target and poor anti-interference ability of the conventional radar, The results of the preliminary exploration on the distributed MIMO radar have shown the great advantages of the distributed MIMO radar in the aspects of target detection and parameter estimation. However, for distributed MIMO radar processing technology, performance evaluation and system design research results are less, many concepts and key technology problems remain in-depth study. In this paper, the research on the target location and power optimization of the distributed MIMO radar is mainly focused on solving the key problems of the existing target location method and the power optimization distribution method. In this paper, the research contents of this paper include the following aspects: the first chapter is the introduction, the background and significance of the subject research are set forth, the research status of the MIMO radar, especially the distributed MIMO radar, is introduced, and the problems of the current distributed MIMO radar parameter estimation and power distribution are analyzed. It also points out the feasibility of carrying out the research on the target location method and the power optimization management. The second chapter mainly studies the target positioning performance of the distributed MIMO radar and its sensitivity to the position error of the radar antenna. The likelihood function is obtained according to the relation between the likelihood function and the likelihood ratio, and the target location problem is abstracted as the maximum likelihood estimation problem of the position parameter, and the Fisher information matrix and the Cramer-Luo lower bound of the target position parameter are derived. Aiming at the problem that the measurement error in the position of the radar antenna can lead to the degradation of the positioning performance, the first-order Taylor approximation is used to derive the mean square error of the target position estimation when the position error of the antenna is ignored, Then, based on the idea of statistical independence, the lower bound of the combination of the position error of the antenna and the parameter of the target position is also taken into account, and the influence of the position error of the antenna on the positioning accuracy is quantitatively analyzed, and the theoretical foundation is laid for the follow-up research. In the third chapter, a distributed MIMO radar indirect location method based on the semi-definite programming theory is mainly studied for the single-target location scenario. In this paper, the nonlinear least-square problem is modeled by the indirect positioning based on the double-station distance observation, and the linear positioning method is derived by the first-order Taylor approximation. In order to obtain the global optimal solution theoretically, the target location method based on semi-definite programming is proposed by means of convex optimization. In this method, the non-linear least square problem is transformed into a constrained convex optimization problem by introducing the redundant variable, and then the non-convex constraint condition is relaxed and then converted into a solvable semi-definite programming problem through the semi-fixed relaxation technique. In that framework, the semi-definite plan positioning method in the case of the position error of the antenna is also derived, and the influence of the position error of the antenna is greatly reduced, and the robust performance of the target positioning is improved. In the fourth chapter, the direct location method of the distributed MIMO radar based on the sparse reconstruction theory is mainly studied for the multi-objective positioning scene. The direct target location based on the matched filtered signal is modeled as a block sparse representation model, and the target location is transformed into a block sparse reconstruction problem. In order to solve the above problems, a multi-object positioning method is developed under the block sparse Bayesian learning framework, which can improve the reconstruction precision by utilizing the correlation between the blocks. The experimental results show that the proposed algorithm does not need the data association processing, and has the ability to deal with the positioning problem under the condition of intensive target, compressed sampling and the like. In addition, an iterative algorithm is proposed to solve the above-mentioned problems in the framework of block sparse Bayesian learning and maximum likelihood estimation for the phase non-synchronization problem that may exist in the coherent processing, in particular, The target scattering coefficient is reconstructed by using the maximum likelihood estimation out of phase error and the block sparse Bayesian learning, and the algorithm exhibits good error correction capability and high positioning accuracy in the case of phase mismatch. The fifth chapter mainly studies the power distribution problem of the distributed MIMO radar aiming at the target location. In this paper, the Bayesian Fisher information matrix of the target position is derived strictly for the indirect positioning model, and the approximate calculation method of the random observability is given in detail. Taking the random observability as the objective function, the total power is the constraint condition, and the power optimization assignment is modeled as the non-convex quadratic problem. In order to solve the above problems, the concept, nature and solution of the Shapley's value in the cooperative game theory are set forth, and a power distribution method based on the value of Shapley is proposed in this framework, and the iterative realization process is given. The method simplifies the calculation process of the value of the Shapley by using the weight graph theory, and has a clear physical meaning. The simulation results show that the proposed algorithm can effectively improve the target positioning accuracy and resource efficiency by reasonably optimizing the distributed transmission power. In the sixth chapter, the research work and main innovation point of this paper are summarized, and the possible research direction is pointed out.
【学位授予单位】：国防科学技术大学
【学位级别】：博士
【学位授予年份】：2014
【分类号】：TN958

【参考文献】