The authors consider the problem of direction-of-arrival (DOA) estimation for the uniform noise (UN) or non-UN (NUN) in a passive radar. The radar waveform to be estimated is modelled as a deterministic and unknown process. Under this condition, the well-known expectation maximisation (EM) algorithm can be utilised for the estimation. In the conventional EM algorithm, the DOA is updated in every iteration, whereas the noise power ratio for each signal component is kept constant. Hence, they have to specify the noise power ratio before implementing the EM algorithm. Clearly, the DOA estimation performance of the conventional EM algorithm is sensitive to this initial noise power ratio. To deal with this problem, they propose the adaptive EM (AEM) algorithms for the UN and NUN, respectively. In the authors’ proposed AEM algorithm, the noise power and DOA are jointly estimated and updated in each iteration, which indicates that the noise power ratio for each signal component would be adaptively changed. Therefore, the DOA performance of their proposed AEM algorithm is less sensitive to the initial noise power ratio, thereby achieving better estimation results. Finally, extensive experiments are carried out to validate the effectiveness of their proposed AEM algorithm.

中文翻译：

---

自适应期望最大化算法对均匀或非均匀噪声的到达方向估计

作者考虑了无源雷达中均匀噪声（UN）或非UN（NUN）的到达方向（DOA）估计问题。将要估计的雷达波形建模为确定性和未知过程。在这种情况下，可以使用众所周知的期望最大化（EM）算法进行估计。在常规的EM算法中，DOA在每次迭代中都被更新，而每个信号分量的噪声功率比保持恒定。因此，他们必须在实施EM算法之前指定噪声功率比。显然，常规EM算法的DOA估计性能对此初始噪声功率比很敏感。为了解决这个问题，他们分别针对UN和NUN提出了自适应EM（AEM）算法。在作者提出的AEM算法中，噪声功率和DOA在每次迭代中被联合估计和更新，这表明每个信号分量的噪声功率比将被自适应地改变。因此，他们提出的AEM算法的DOA性能对初始噪声功率比不那么敏感，从而获得了更好的估计结果。最后，进行了广泛的实验以验证其提出的AEM算法的有效性。