Asynchronous stochastic approximation algorithms for networked systems: regime-switching topologies and multiscale structure (Q2874199)

The paper develops consensus algorithms under the asynchronous communication and random computation environments. Consensus problems are related to control applications that involve coordination of multiple entities with only limited neighborhood information to reach a global goal for the entire team. The paper considers a system consisting of several components that can be handled by different agents, and the information can be shared by agents. To each single agent, the agent can start the next iteration using the newest information of iteration on other components without waiting for other agents to finish. So for each component, the times of iterations and the number of iterations to that moment are random. The modulating force of the switching process is modeled as a discrete-time Markov chain with a finite state space. It is supposed that the transition probability matrix of the Markov chain includes a small parameter \(\varepsilon \). The stochastic approximation algorithm defines its updating speed by another small parameter \(\mu \). The interplay of the two parameters introduces a multiscale system dynamics. It turns out that the difference between the parameters (\(\varepsilon = O(\mu ),\varepsilon < < \mu ,\mu < < \varepsilon \)) gives rise to qualitatively different behaviors with stark contrasts.