Distributed estimation of algebraic connectivity of directed networks

In directed network, algebraic connectivity is defined as the second smallest eigenvalue of graph Laplacian, and it captures the most conservative estimate of convergence rate and synchronicity of networked systems. In this paper, distributed estimation of algebraic connectivity of directed and connected graphs is studied using a decentralized power iteration scheme. Specifically, the proposed scheme is introduced in discrete time domain in order to take advantage of the discretized nature of information flow among networked systems and it shows that, with the knowledge of the first left eigenvector associated with trivial eigenvalue of graph Laplacian, distributed estimation of algebraic connectivity becomes possible. Moreover, it is revealed that the proposed estimation scheme still performs in estimating the complex eigenvalues. Simulation results demonstrate the effectiveness of the proposed scheme.