Distributed Constrained Optimization Over Cloud-Based Multi-agent Networks

We consider a distributed constrained optimization problem where a group of distributed agents are interconnected via a cloud center, and collaboratively minimize a network-wide objective function subject to local and global constraints. This paper devotes to developing an efficient distributed algorithm that fully utilizes the computation abilities of the cloud center and the agents, as well as avoids extensive communications between the cloud center and the agents. We address these issues by introducing a divide-and-conquer technique, which assigns the local objective functions and constraints to the agents while the global ones to the cloud center. The resultant algorithm naturally yields two layers, the agent layer and the cloud center layer. They exchange their intermediate variables so as to collaboratively obtain a network-wide optimal solution. Numerical experiments demonstrate the effectiveness of the proposed distributed constrained optimization algorithm.