当前位置: X-MOL 学术IEEE Trans. Autom. Control › 论文详情
Our official English website, www.x-mol.net, welcomes your feedback! (Note: you will need to create a separate account there.)
Distributed Constrained Optimization Over Unbalanced Directed Networks Using Asynchronous Broadcast-Based Algorithm
IEEE Transactions on Automatic Control ( IF 6.8 ) Pub Date : 2020-05-11 , DOI: 10.1109/tac.2020.2994024
Huaqing Li , Qingguo Lu , Guo Chen , Tingwen Huang , Zhaoyang Dong

This article focuses on distributed convex optimization problems over an unbalanced directed multiagent (no central coordinator) network with inequality constraints. The goal is to cooperatively minimize the sum of all locally known convex cost functions. Every single agent in the network only knows its local objective function and local inequality constraint, and is constrained to a privately known convex set. Furthermore, we particularly discuss the scenario in which the interactions among agents over the whole network are subjected to possible link failures. To collaboratively solve the optimization problem, we mainly concentrate on an epigraph form of the original constrained optimization to overcome the unbalancedness of directed networks, and propose a new distributed asynchronous broadcast-based optimization algorithm. The algorithm allows that not only the updates of agents are asynchronous in a distributed fashion, but also the step-sizes of all agents are uncoordinated. An important characteristic of the proposed algorithm is to cope with the constrained optimization problem in the case of unbalanced directed networks whose communications are subjected to possible link failures. Under two standard assumptions that the communication network is directly strongly connected and the subgradients of all local objective functions are bounded, we provide an explicit analysis for convergence of the algorithm. Simulation results obtained by three numerical experiments substantiate the feasibility of the algorithm and validate the theoretical findings.

中文翻译:

基于异步广播的不平衡有向网络上的分布式约束优化

本文关注具有不等式约束的不平衡有向多主体(无中央协调器)网络上的分布式凸优化问题。目标是协作地最小化所有本地已知凸成本函数的总和。网络中的每个代理都仅知道其局部目标函数和局部不等式约束,并且被约束到一个私有的凸集。此外,我们特别讨论了整个网络上的代理之间的交互可能遭受链路故障的情况。为了协同解决优化问题,我们主要集中在原始约束优化的题词形式上,以克服有向网络的不平衡性,并提出了一种新的基于分布式异步广播的优化算法。该算法不仅允许代理的更新以分布式方式异步,而且所有代理的步长都是不协调的。所提出算法的重要特征是在通信受到可能的链路故障的不平衡有向网络的情况下,解决约束优化问题。在两个标准假设下,即通信网络是直接牢固连接的,并且所有局部目标函数的次梯度都是有界的,我们为算法的收敛性提供了一个明确的分析。通过三个数值实验获得的仿真结果证实了该算法的可行性并验证了理论结果。而且所有座席的步长都是不协调的。所提出算法的重要特征是在通信受到可能的链路故障的不平衡有向网络的情况下,解决约束优化问题。在两个标准假设下,即通信网络是直接牢固连接的,并且所有局部目标函数的次梯度都是有界的,我们为算法的收敛性提供了一个明确的分析。通过三个数值实验获得的仿真结果证实了该算法的可行性并验证了理论结果。而且所有座席的步长都是不协调的。所提出算法的重要特征是在通信受到可能的链路故障的不平衡有向网络的情况下,解决约束优化问题。在两个标准假设下,即通信网络是直接牢固连接的,并且所有局部目标函数的次梯度都是有界的,我们为算法的收敛性提供了一个明确的分析。通过三个数值实验获得的仿真结果证实了该算法的可行性并验证了理论结果。所提出算法的重要特征是在通信受到可能的链路故障的不平衡有向网络的情况下,解决约束优化问题。在两个标准假设下,即通信网络是直接牢固连接的,并且所有局部目标函数的次梯度都是有界的,我们为算法的收敛性提供了一个明确的分析。通过三个数值实验获得的仿真结果证实了该算法的可行性并验证了理论结果。所提出算法的重要特征是在通信受到可能的链路故障的不平衡有向网络的情况下,解决约束优化问题。在两个标准假设下,即通信网络是直接牢固连接的,并且所有局部目标函数的次梯度都是有界的,我们为算法的收敛性提供了一个明确的分析。通过三个数值实验获得的仿真结果证实了该算法的可行性并验证了理论结果。
更新日期:2020-05-11
down
wechat
bug