一种联合优化的无源雷达实时目标检测算法

摘要

Passive radar exploits an external illuminator signal to detect targets. It has the advantages of silence, anti-interference, and counter-stealth ability. In most cases, direct and multipath clutters should be suppressed first. Then coherent detection can be made by performing a cross-ambiguity function of the remaining target echoes and the reference signal. However, under a wide-band signal, a long-integration time, or multi-beam circumstances, a large number of computations and amount of memory is required for normal processing. This paper expresses the mathematical relationships of clutter suppression algorithms based on the Minimum Mean Square Error (MMSE) principle and coherent detection algorithms based on the cross-ambiguity function. Herein, a joint-optimize and processing method is presented. This method reduces the number of computations and amount of memory required, is easy to implement on GPU devices such as CUDA, and will be useful for engineering applications. Its high-efficiency and real-time properties are validated in the experimental results.%无源雷达是一种本身不发射电磁波，而是利用外部辐射源信号进行探测的雷达系统，具有隐蔽性好、抗电磁干扰、抗隐身技术等优点。为了实现可靠目标探测，通常先对回波信号进行杂波对消，然后计算剩余回波和参考信号的互模糊函数。在基于大带宽信号源、相干累积时间较长、多波束同时处理的无源雷达系统中，常规方法需要较大的计算量和存储空间，不利于实时实现。该文研究了基于最小均方误差(Minimum Mean Square Error, MMSE)准则的杂波对消算法与基于互模糊函数的相干累积检测算法之间内在数学关系，提出一种降低计算量和存储量的联合优化方法，给出了在 NVIDIA CUDA(Computing Unified Device Architecture)平台上的实现，用实测结果论证了该文方法的高效性、实时性，并已应用于工程中。