Mesbahi M., Egerstedt M. Graph Theoretic Methods in Multiagent Networks

Издательство Princeton University Press, 2010, -424 pp.The emergence of (relatively) cheap sensing and actuation nodes, capable of short-range communications and local decision-making, has raised a number of new system-level questions concerning how such systems should be coordinated and controlled. Arguably, the biggest challenge facing this new field of research is means by which local interaction rules lead to desired global properties, that is, given that the networked system is to accomplish a particular task, how should the interaction and control protocols be structured to ensure that the task is in fact achieved?
This newly defined area of networked systems theory has attracted wide interest during the last decade. A number of sessions are devoted to this problem at the major conferences and targeted conferences have emerged. Moreover, graduate-level courses are beginning to be taught in this general area, and major funding institutions are pursuing networked systems as integral to their missions due to the many applications where network-level questions must be addressed. These applications include sensor networks, multiagent robotics, and mobile ad hoc communication nets, in addition to such areas as social networks and quantum networks.
The particular focus of this book is on graph theoretic methods for the analysis and synthesis of networked dynamic systems. By abstracting away the complex interaction geometries associated with the sensing and communication footprints of the individual agents, and instead identifying agents with nodes in a graph and encoding the existence of an interaction between nodes as an edge, a powerful new formalism and set of tools for networked systems have become available. For instance, the graph theoretic framework described in this book provides means to examine how the structure of the underlying interaction topology among the agents leads to distinct global behavior of the system. This graph theoretic outlook also allows for examining the correspondence between system theoretic features of networked systems on one hand, and the combinatorial and algebraic attributes of the underlying network on the other. By doing this, one can, for example, address questions related to the robustness of networked systems in terms of the variation of the network topology, as well as the network synthesis problem in the context of embedded networked systems.Part 1 FoundationsGraph Theory
The Agreement Protocol: Part I–The Static
The Agreement Protocol: Part II–Lyapunov and LaSalle
Probabilistic Analysis of Networks and Protocol
Part 2 Multiagent Networks
Formation Control
Mobile Robots
Distributed Estimation
Social Networks, Epidemics, and Games
Part 3 Networks as Systems
Agreement with Inputs and Outputs
Synthesis of Networks
Dynamic Graph Processes
Higher-order Networks