Different transmit antenna elements in a frequency diverse array (FDA) radar transmit different frequencies. This makes the transmitted beam to continuously change its orientation, and enables an FDA to scan the space around itself in a simple manner. Furthermore, an FDA can find the range as well as the direction of each target simultaneously. The achievable resolution of direction estimation depends on the receive array geometry. The range resolution can be enhanced by decreasing the width of transmit beam, which depends on the transmit array geometry. The later property makes FDA especially attractive because it is no longer necessary to produce sharp, wideband, and high power pulses to achieve high range resolution. FDA has been studied mostly for collocated transmit and receive array, and a very little has been done on its application in bistatic radar. In this paper we propose a new signal transmission scheme for bistatic FDA radar, and demonstrate how one can detect and localize the targets accurately using a computationally efficient algorithm. The proposed signal transmission scheme is designed to produce the narrowest transmit beam for a given number of transmit antennas. This achieves significantly better detection and localization performance compared to the existing methods. Our estimation approach uses a fast algorithm to produce some initial estimates that can be used to initialize a numerical algorithm for computing the maximum likelihood estimates. We also derive the Cramér Rao bound (CRB) associated with the estimation problem, and demonstrate that the variances of the estimates produced by the proposed method are close to corresponding CRBs.

频率分集阵列（FDA）雷达中的不同发射天线元件发射不同的频率。这使透射光束不断改变其方向，并使FDA能够以简单的方式扫描其周围的空间。此外，FDA可以同时找到每个目标的范围和方向。方向估计的可实现分辨率取决于接收阵列的几何形状。可以通过减小发射波束的宽度来增强距离分辨率，这取决于发射阵列的几何形状。后者的特性使FDA特别具有吸引力，因为它不再需要产生尖锐，宽带和高功率的脉冲来实现高范围分辨率。FDA主要针对并置发射和接收阵列进行了研究，在双基地雷达中的应用还很少。在本文中，我们为双基地FDA雷达提出了一种新的信号传输方案，并演示了如何使用一种计算效率高的算法精确地检测和定位目标。拟议的信号传输方案旨在针对给定数量的发射天线产生最窄的发射波束。与现有方法相比，这可以显着提高检测和定位性能。我们的估计方法使用快速算法来生成一些初始估计，这些初始估计可用于初始化用于计算最大似然估计的数值算法。我们还推导了与估计问题相关的CramérRao界（CRB），