We propose a continuous-time algorithm for solving a resource allocation problem cooperatively and distributedly over a uniformly jointly strongly connected graph. Particularly, a novel passivity-based perspective of the proposed algorithmic dynamic at each individual node is provided, which enables us to analyze the convergence of the overall distributed algorithm over time-varying digraphs. The parameters in the proposed algorithm rely only on local information of each individual nodes, which can be designed in a truly distributed fashion. A periodic communication mechanism is also derived using the passivity degradation over sampling of the distributed dynamics in order to avoid the introduction of the restrictive assumption of continuous-time communication among nodes.
翻译:我们建议采用连续时间算法,通过一个统一、紧密相连的图表,以合作和分配的方式解决资源分配问题。 特别是,提供了对每个节点的拟议算法动态的新式被动观点,这使我们能够分析在时间变化的分数中总体分布算法的趋同情况。 拟议的算法参数仅依赖每个节点的当地信息,而这些信息可以真正分布。 定期通信机制还利用对分布式动态的抽样的被动性退化来生成,以避免引入节点之间连续通信的限制性假设。