论文标题
部分可观测时空混沌系统的无模型预测
A Message Passing Based Average Consensus Algorithm for Decentralized Frequency and Phase Synchronization in Distributed Phased Arrays
论文作者
论文摘要
我们考虑通过节点电态的局部广播分布式相位阵列中分散频率和相位同步的问题。分布式数组中的节点之间的频率和相位同步是为了支持波束成形，但是由于节点的局部振荡器的操作动力学，其输出信号的频率和相位会在更新间隔之间在更新间隔之间进行随机漂移和抖动。此外，频率和相位估计误差会导致总相误差，从而导致阵列中的残留相误差，从而降低了连贯的操作。 Recently, a classical decentralized frequency and phase synchronization algorithm based on consensus averaging was proposed with which the standard deviation of the residual phase errors upon convergence was reduced to $10^{-4}$ degrees for internode update intervals of $0.1$ ms, however this was obtained for arrays with at least $400$ nodes and a high connectivity ratio of $0.9$.在本文中，我们提出了一种基于消息传递的平均共识（MPAC）算法，以改善分布式阵列中节点的电态的同步。仿真结果表明，所提出的MPAC算法将残留相误差显着降低至$ 10^{ - 11} $度，仅需要$ 20 $中等连接的节点。此外，MPAC收敛速度比基于DFPC的算法更快，特别是对于具有适度连接性的较大阵列。
We consider the problem of decentralized frequency and phase synchronization in distributed phased arrays via local broadcast of the node electrical states. Frequency and phase synchronization between nodes in a distributed array is necessary to support beamforming, but due to the operational dynamics of the local oscillators of the nodes, the frequencies and phases of their output signals undergo the random drift and jitter in between the update intervals. Furthermore, frequency and phase estimation errors contribute to the total phase errors, leading to a residual phase error in the array that degrades coherent operation. Recently, a classical decentralized frequency and phase synchronization algorithm based on consensus averaging was proposed with which the standard deviation of the residual phase errors upon convergence was reduced to $10^{-4}$ degrees for internode update intervals of $0.1$ ms, however this was obtained for arrays with at least $400$ nodes and a high connectivity ratio of $0.9$. In this paper, we propose a message passing based average consensus (MPAC) algorithm to improve the synchronization of the electrical states of the nodes in distributed arrays. Simulation results show that the proposed MPAC algorithm significantly reduces the residual phase errors to about $10^{-11}$ degrees, requiring only $20$ moderately connected nodes in an array. Furthermore, MPAC converges faster than the DFPC-based algorithms, particularly for the larger arrays with a moderate connectivity.
