Distributed Computation: Consensus Algorithms in Networked Control Systems

This lecture covers the application of consensus algorithms in networked control systems, focusing on topics such as Metropolis-Hasting models, weighted adjacency matrices, node counting, and distributed computation of Least-Squares regression. The instructor explains how to compute BLUE in a distributed way, the consensus on vector-valued and matrix-valued quantities, and the distributed computation of LS regression. The lecture also delves into the explicit formulas for LS, the goal of making each node compute LS with a distributed algorithm, and the time-varying consensus algorithms' robustness to changes in graph topology, lossy communication, transmission delays, and quantization.