In this paper, we explore a collaborative direction-of-arrival (DOA) estimation technique that utilizes multiple closely spaced subarrays to maximize the potential of distributed arrays while minimizing communication overhead between the subarrays and the processing center. Each subarray computes its self-covariance matrix using the full-precision data and transmits it, along with a one-bit version of the measured data, to the processing center. The processing center generates one-bit cross-covariance matrices between subarrays, which are combined with full-precision subarray self-covariance matrices to create the mixed-precision covariance matrix of the entire array for source DOA estimation. This approach utilizes the full array aperture and all available degrees of freedom of the entire distributed array. To address missing entries in the full covariance matrix, we employ matrix completion, taking into account its Toeplitz and Hermitian structure. For subarrays that are not positioned on the half-wavelength grid, we propose an iterative DOA estimation method to ensure robust DOA estimation performance. Our proposed approach outperforms scenarios where cross-covariance matrices are unavailable or the entire covariance matrix is not interpolated. With the same communication traffic limitation, it demonstrates superiority over schemes that utilize only full-precision data or only one-bit data.

中文翻译：

---

利用混合精度协方差矩阵的紧密分布阵列中的到达方向估计

在本文中，我们探索了一种协作到达方向（DOA）估计技术，该技术利用多个紧密间隔的子阵列来最大化分布式阵列的潜力，同时最小化子阵列和处理中心之间的通信开销。每个子阵列使用全精度数据计算其自协方差矩阵，并将其与测量数据的一位版本一起传输到处理中心。处理中心生成子阵之间的一位互协方差矩阵，与全精度子阵自协方差矩阵相结合，创建整个阵列的混合精度协方差矩阵，用于源DOA估计。该方法利用整个分布式阵列的全阵列孔径和所有可用的自由度。为了解决完整协方差矩阵中缺失的条目，我们采用矩阵补全，并考虑其托普利茨和埃尔米特结构。对于未位于半波长网格上的子阵，我们提出了一种迭代 DOA 估计方法，以确保稳健的 DOA 估计性能。我们提出的方法优于交叉协方差矩阵不可用或整个协方差矩阵未插值的情况。在相同的通信流量限制下，它比仅使用全精度数据或仅使用一位数据的方案表现出优越性。