Among the various research branches in multiple-input multiple-output (MIMO) radar, Direction finding is an interesting topic and has attracted extensive concerns. However, many previous strategies are only effective to deal with scenarios without model error, such as well-calibrated sensor array, which is unrealistic in practice. This paper revisits the direction finding issue in a bistatic MIMO radar, in which the direction-dependent mutual coupling (MC) effect in both the transmitting array and the receiving array are considered. An estimator based on parallel factor (PARAFAC) decomposition is introduced. Benefit from the fact that the multidimensional structure of the tensor measurement can be explored, the proposed PARAFAC estimator can offer closed-form solution to direction finding without additional pairing calculation, so it is much more accurate and efficient than the state-of-the-art spectrum search method and the rotational invariance algorithm. The improved PARAFAC approach is mathematically analyzed in detail, Several computer trials are carried out to show the theoretical advantages.

在多输入多输出（MIMO）雷达的各个研究分支中，测向是一个有趣的话题，并引起了广泛的关注。然而，之前的许多策略仅对没有模型误差的场景有效，例如校准良好的传感器阵列，这在实践中是不现实的。本文重新审视了双基地 MIMO 雷达中的测向问题，其中考虑了发射阵列和接收阵列中的方向相关互耦合 (MC) 效应。介绍了一种基于并行因子（PARAFAC）分解的估计器。受益于可以探索张量测量的多维结构这一事实，所提出的 PARAFAC 估计器可以为测向提供封闭形式的解决方案，而无需额外的配对计算，因此它比最先进的谱搜索方法和旋转不变性算法更准确、更高效。对改进的PARAFAC方法进行了详细的数学分析，并进行了多次计算机试验以显示其理论优势。