Robust separation of background and target signals in radar cross section measurements
Robust separation of background and target signals in radar cross section measurements
复制标题
DOI：
10.1109/tim.2005.858126
复制
发表时间：
2005-11
影响因子：
5.6
通讯作者：
L. Muth;C. Wang;T. Conn
中科院分区：
文献类型：
--
作者：
L. Muth;C. Wang;T. Conn
在自由空间中沿系统视距运动的目标上的雷达横截面的相干测量将沿着以复（I，Q）平面原点为中心的圆进行。不依赖于目标位置的附加复杂背景信号（包括静止杂波、目标支持和平均目标-挂载相互作用）的存在，将把圆的原点转换为某个复杂点（I/sub 0/,Q/sub 0/）。异常值的存在（主要是由于射频干扰）可以在确定I-Q圆的半径和中心时引入重大误差。我们实现了鲁棒高效的最小中值平方和正交距离回归算法的组合，以消除或减少异常值的影响，然后分离目标和背景信号。同时，减小了噪声的影响。因此，我们既可以获得与目标无关的背景估计，也可以获得校准伪影雷达横截面的无背景估计。在低可观测目标的测量中，从测量和校准中去除背景信号可以显著提高测量精度。这种技术对于非常高频和超高频的亚波长转换特别有用，因为可用的数据弧线有限，光谱技术不适用。
Coherent measurements of radar cross-section on a target moving along the system line-of-sight in free space will trace a circle centered on the origin of the complex (I,Q) plane. The presence of additional complex background signals (including stationary clutter, target support, and averaged target-mount interactions), which do not depend on target position, will translate the origin of the circle to some complex point (I/sub 0/,Q/sub 0/). The presence of outliers (mostly due to radio-frequency interference) can introduce significant errors in the determination of the radius and center of the I-Q circle. We have implemented a combination of a robust and efficient least median square and an orthogonal-distance regression algorithm to eliminate or to reduce the influence of outliers, and then to separate the target and background signals. Concurrently, the influence of noise is also reduced. Thus, we can obtain both target-independent estimates of the background and a background-free estimates of the radar cross-sections of calibration artifacts. In measurements on low-observable targets, the subtraction of the background signal from the measurement and calibration significantly improves the measurement accuracy. This technique is especially useful for subwavelength translations at very and ultra-high-frequencies, where spectral techniques are not applicable because the available arc of data is limited.