Advanced Distributed Consensus for Multiagent Systems

By Magdi S. Mahmoud, Mojeed O. Oyedeji and Yuanqing Xia

Advanced Distributed Consensus for Multiagent Systems contributes to the further development of advanced distributed consensus methods for different classes of multiagent methods. The book expands the field of coordinated multiagent dynamic systems, including discussions on swarms, multi-vehicle and swarm robotics. In addition, it addresses advanced distributed methods for the important topic of multiagent systems, with a goal of providing a high-level treatment of consensus to different versions while preserving systematic analysis of the material and providing an accounting to math development in a unified way. This book is suitable for graduate courses in electrical, mechanical and computer science departments.

Consensus control in multiagent systems is becoming increasingly popular among researchers due to its applicability in analyzing and designing coordination behaviors among agents in multiagent frameworks. Multiagent systems have been a fascinating subject amongst researchers as their practical applications span multiple fields ranging from robotics, control theory, systems biology, evolutionary biology, power systems, social and political systems to mention a few.

• Gathers together the theoretical preliminaries and fundamental issues related to multiagent systems and controls
• Provides coherent results on adopting a multiagent framework for critically examining problems in smart microgrid systems
• Presents advanced analysis of multiagent systems under cyberphysical attacks and develops resilient control strategies to guarantee safe operation

• Language: English
• Publisher: Academic Press
• Release date: Dec 5, 2020
• ISBN: 9780128232033

Magdi S. Mahmoud

Magdi S. Mahmoud is a distinguished professor at King Fahd University of Petroleum and Minerals (KFUPM), Saudi Arabia. He has been faculty member at different universities worldwide including Egypt (CU, AUC), Kuwait (KU), UAE (UAEU), UK (UMIST), USA (Pitt, Case Western), Singapore (Nanyang), and Australia (Adelaide). He lectured in Venezuela (Caracas), Germany (Hanover), UK (Kent), USA (UoSA), Canada (Montreal) and China (BIT, Yanshan). He is the principal author of 51 books, inclusive book-chapters, and author/co-author of more than 610 peer-reviewed papers. He is a fellow of the IEE and a senior member of the IEEE, the CEI (UK). He is currently actively engaged in teaching and research in the development of modern methodologies to distributed control and filtering, networked control systems, fault-tolerant systems, cyberphysical systems, and information technology.

Book preview

2019

Chapter 1: An overview

Abstract

Consensus control in multiagent systems is becoming increasingly popular among researchers due to its applicability in analyzing and designing coordination behaviors among agents in a multiagent framework. This chapter provides an extensive overview on consensus control in multiagent systems from the network perspective. Specifically, this chapter provides:

•  An overview of agent models (discrete and continuous) which has been studied by earlier researchers;

•  A summary of different forms of consensus in multiagent systems;

•  An overview of recent results in consensus-related problems involving network phenomena such as time-delay, actuator failures, switching, and random networks;

•  Suggestions for future work towards designing better consensus protocols that address real-life problems in autonomous multiagent systems.

Keywords

Graph Laplacian; Adjacency matrix; Eigenvalues and eigenvectors; Models of agent dynamics; Single/double integrator model; Uncertain fully actuated model; Nonholonomic unicycle model

1.1 Introduction

During the last two decades, cooperative control of multiagent systems (MASs) has received considerable attention due to its wide applications in many fields such as formation control, sensor networks, attitude of spacecraft alignment, and so on. Nowadays, multiagent systems (MASs) technology is growing to the point where the first multiagent systems are now being transferred from the laboratory to the utility, allowing industry to gain experience in the use of MASs and also to evaluate their effectiveness.

Consensus problems have a long history in computer science and form the foundation of the field of distributed computing. In networks of agents (or dynamical systems) consensus means reaching an agreement regarding a certain quantity of interest that depends on the state of all agents. A consensus algorithm (or protocol) is an interaction rule that specifies the information exchange between an agent and all of its neighbors in the network.

The problem of reaching consensus, that is, driving the state of a set of interconnected dynamical systems towards the same value, has received much attention due to its many applications in both the modeling of natural phenomena such as flocking and in the solution of several control problems involving synchronization or agreement between dynamical systems.

Cooperative collective behaviors in networks of autonomous agents, such as synchronization, consensus, swarming, and particularly flocking, have received considerable attention in recent years due to their broad applications to biological systems, sensor networks, unmanned air vehicle formations, robotic cooperation teams, mobile communication systems, and so on. In a flock, to coordinate with other dynamical agents, every individual needs to share information with each other, and they all need to agree on a common objective of interest. In this pursuit of scientific research, two strategies are commonly adopted: centralized control and distributed control. The centralized approach assumes that a central station is available and is powerful enough to communicate with and control the whole group of mobile agents. On the contrary, the distributed approach does not require such a central unit for control and management, at the cost of becoming more complicated in both network structure and organization of multiple agents. Although both approaches are practically depending on the situations and conditions of the applications at hand, the distributed approach is generally more attractive due to the existence of many inevitable physical constraints in practice such as only locally available information, limited resources and energy, distance decay in communications, and the large scale of agent systems. This section reviews some recent progress in distributed consensus and coordination control of mobile multiagent systems over complex communication networks.

The study of distributed coordination control of mobile multiagent systems was perhaps first motivated by the works in distributed computing, management science, and statistical physics, among others. Briefly stated, research studies on distributed coordination control of mobile multiagent systems include:

1.  Consensus. This refers to the group behavior that all mobile agents asymptotically reach an agreement or alignment under a local distributed control protocol, with or without requiring some predefined common speed and orientation in their asymptotic motions.

2.  Formation control. This refers to the group behavior that all mobile agents asymptotically form a predesigned geometrical configuration through local interactions, with or without a common reference such as a target state or convergence agreement.

3.  Distributed estimation and control. This refers to designing distributed controllers for networked mobile systems, using local estimators to obtain the needed global information.

1.2 Notations

respectively denote the set of nonnegative real numbers and the nbe the set of positive integers and let In denote an nmeans that A the family of continuous functions φ .

1.3 Elements of graph theory

In this section, some preliminary knowledge of graph theory [1] is introduced to facilitate the subsequent analysis. For a system of n connected agents, its network topology can be modeled as a directed graph.

(A) be a weighted directed graph of order n, where nodes i and j the set of neighbors of node i.

(B) . A directed tree is a directed graph, in which there is exactly one parent for every node except for a node called the root. A directed spanning tree be a directed graph of order nin a directed graph. A directed graph is said to be strongly connected, if there is a directed path from every node to every other node. Moreover, a directed graph is said to have spanning trees, if there exists a node such that there is a directed path from every other node to this