Distributed Resource Allocation via ADMM over Digraphs
Wei Hua Jiang, Mohammadreza Doostmohammadian, Themistoklis Charalambous · 2022 IEEE 61st Conference on Decision and Control (CDC) · 2022
In this paper, we solve the resource allocation problem over a network of agents, with edges as communication links that can be unidirectional. The goal is to minimize the sum of allocation cost functions subject to a coupling constraint in a distributed way by using the finite-time consensus-based alternating direction method of multipliers (ADMM) technique. In contrast to the existing gradient descent (GD) based distributed algorithms, our approach can be applied to non-differentiable cost functions. Also, the proposed algorithm is initialization-free and converges at a rate of $\mathcal{O}\left( {1/k} \right)$, where k is the optimization iteration counter. The fast convergence performance related to iteration counter k compared to state-of-the-art GD based algorithms is shown via a simulation example.