分布式无线网络中的仿射投影自适应算法

摘要

In this paper,we study the affine projection algorithms （APA） over distributed wireless net- works. Distributed networks exploit the cooperation of nodes and space - time structure of input data to improve the robustness of estimation results. Nodes evolve in the space domain by exchanging the infor- mation in neighborhoods and in the time domain by local update. Based on Least - Mean squared cost function of Newton method,we achieve the adaptive APA algorithms,which can be applied in distributed networks with different topologies. Simulations confirm the improved performance of the whole networks by nodes cooperation.%本文对分布式无线网络中的仿射投影自适应算法（APA）进行研究。分布式网络的工作原理是利用网络的通信合作和节点输入数据的空间域一时域特性来提高估算结果的鲁棒性，各节点通过与其通信组节点的信息交流来实现空间域演进，同时通过本地迭代来实现时域演进。此外，我们根据Newton公式的最小代价函数推导出适用于不同拓扑结构分布式网络的放射投影自适应算法，并通过仿真验证了网络节点间的通信合作对于整体网络性能的改善。