Siegfried Modeling and Simulation of Complex Systems

A Framework for Efficient Agent-Based Modeling and Simulation
2014
ISBN: 978-3-658-07529-3
Verlag: Springer
Format: PDF
Kopierschutz: 1 - PDF Watermark

E-Book, Englisch, 233 Seiten

Robert Siegfried presents a framework for efficient agent-based modeling and simulation of complex systems. He compares different approaches for describing structure and dynamics of agent-based models in detail. Based on this evaluation the author introduces the 'General Reference Model for Agent-based Modeling and Simulation' (GRAMS). Furthermore he presents parallel and distributed simulation approaches for execution of agent-based models -from small scale to very large scale. The author shows how agent-based models may be executed by different simulation engines that utilize underlying hardware resources in an optimized fashion.

Robert Siegfried is Senior Consultant for IT/M&S projects. He earned his doctorate in modeling and simulation at the Universität der Bundeswehr München. His research areas are agent-based modeling and simulation, distributed simulation, and quality management. He has worked on topics like model documentation and management, distributed simulation test beds, and process models. He is active member of the NATO Modeling and Simulation Group and the Simulation Interoperability Standards Organization.

Autoren/Hrsg.

Siegfried, Robert

Inhaltsverzeichnis

1;Abstract;6
2;Zusammenfassung;8
3;Contents;10
4;List of Figures;15
5;List of Tables;17
6;1 Introduction;18
6.1;1.1 Motivation;18
6.2;1.2 Goals of this thesis;20
6.2.1;Effective and efficient model development;20
6.2.2;Effective model execution;22
6.3;1.3 Approach of this thesis and outline;23
7;Part I Preliminaries and related work;26
7.1;2 Agent-based modeling and simulation;27
7.1.1;2.1 Basic terminology;27
7.1.1.1;2.1.1 Complex system;27
7.1.1.2;2.1.2 Model;28
7.1.1.3;2.1.3 Model development process;29
7.1.1.4;2.1.4 Simulation;31
7.1.2;2.2 Agent-based modeling and simulation;33
7.1.2.1;2.2.1 Agent;33
7.1.2.2;2.2.2 Agent architecture;35
7.1.2.3;2.2.3 Relationships with other modeling paradigms;37
7.1.2.4;2.2.4 Benefits and drawbacks of agent-based modeling and simulation;39
7.1.3;2.3 Related work;41
7.1.3.1;2.3.1 Klügl (2007);42
7.1.3.2;2.3.2 Scheutz, Schermerhorn (2006);46
7.1.3.3;2.3.3 Ferber, Müller (1996) and Weyns, Holvoet (2003);49
7.1.3.4;2.3.4 DEVS and variable structure models;52
7.1.3.5;2.3.5 Agent-Object-Relationship Simulation;53
7.1.3.6;2.3.6 Further related work;54
7.1.3.7;2.3.7 Summary;56
7.1.4;2.4 The role of the environment in agent-based models;60
7.1.4.1;2.4.1 Definition and properties;61
7.1.4.2;2.4.2 Dynamics;62
7.2;3 Parallel and distributed multi-agent simulation;64
7.2.1;3.1 Basic terminology;64
7.2.2;3.2 Motivation and requirements with respect to multi-agent simulation;66
7.2.3;3.3 Existing approaches and their limitations;68
7.2.3.1;3.3.1 Pawlaszczyk (2009);69
7.2.3.2;3.3.2 Logan, Theodoropoulos (2001);70
7.2.3.3;3.3.3 P. Riley, G. Riley (2003);70
7.2.3.4;3.3.4 Scheutz, Schermerhorn (2006);71
7.2.3.5;3.3.5 Aaby et al. (2010);72
7.2.3.6;3.3.6 Further related work;73
7.3;4 Summary;76
8;Part II Effective and efficient model development;78
8.1;5 The need for a reference model for agent-based modeling and simulation;79
8.1.1;5.1 Current problems and needs of agent-based modeling and simulation;79
8.1.1.1;5.1.1 Weak definitions of basic terms and concepts;79
8.1.1.2;5.1.2 Bad support of collaborative development;81
8.1.1.3;5.1.3 Need for improvement of quality by standardization;82
8.1.2;5.2 General purpose of reference models;82
8.1.3;5.3 Requirements for a reference model;83
8.1.3.1;5.3.1 Clear distinction between model and simulation engine;84
8.1.3.2;5.3.2 Simulated environment;85
8.1.3.3;5.3.3 Simulation time;85
8.1.3.4;5.3.4 Agents and objects;86
8.1.3.5;5.3.5 Actions, effects and constraints;86
8.1.3.6;5.3.6 Abstraction from specific agent architectures;87
8.1.3.7;5.3.7 Formal specification;87
8.2;6 GRAMS – General Reference Model for Agent-based Modeling and Simulation;89
8.2.1;6.1 Aims and target audience;89
8.2.2;6.2 Basic idea of the GRAMS reference model;90
8.2.3;6.3 Definition of an agent-based model;92
8.2.4;6.4 Macro-level modeling;95
8.2.4.1;6.4.1 Simulation time;95
8.2.4.2;6.4.2 Environment;96
8.2.4.3;6.4.3 Embedding of entities;98
8.2.5;6.5 Events;99
8.2.6;6.6 Micro-level modeling;102
8.2.6.1;6.6.1 Attributes of agents and objects;103
8.2.6.2;6.6.2 Sensors and effectors;104
8.2.6.3;6.6.3 State of an agent;109
8.2.6.4;6.6.4 Constraints;109
8.2.6.5;6.6.5 Behavior of an agent;112
8.2.7;6.7 Simulation of an agent-based model;114
8.2.8;6.8 Application areas of the GRAMS reference model;116
8.2.9;6.9 Evaluation;118
8.2.9.1;6.9.1 Mapping of basic agent architectures;118
8.2.9.2;6.9.2 Case studies;122
8.3;7 Summary;125
9;Part III Effective model execution;128
9.1;8 Model partitioning and multi-level parallelization;129
9.1.1;8.1 Model partitioning strategies;129
9.1.1.1;8.1.1 Terms and definitions;129
9.1.1.2;8.1.2 Conflict detection as bottleneck;132
9.1.1.3;8.1.3 Partitioning strategy: node-level, per agent, static;134
9.1.1.4;8.1.4 Partitioning strategy: node-level, per agent, dynamic;135
9.1.1.5;8.1.5 Partitioning strategies on cluster-level;139
9.1.2;8.2 Parallel execution;141
9.1.2.1;8.2.1 Parallelization on nodeand processor-level;142
9.1.2.2;8.2.2 Parallelization on cluster-level;143
9.1.2.3;8.2.3 Multi-level parallelization;143
9.1.3;8.3 Reliability of a simulation environment;144
9.2;9 Example implementation of GRAMS;145
9.2.1;9.1 Overview;145
9.2.2;9.2 Representation of an agent-based model;148
9.2.2.1;9.2.1 Simulation time;149
9.2.2.2;9.2.2 Environment;150
9.2.2.3;9.2.3 Events;151
9.2.2.4;9.2.4 Constraints;152
9.2.2.5;9.2.5 Entities;154
9.2.3;9.3 Simulation engines;155
9.2.3.1;9.3.1 Single-threaded event-driven simulation engine;156
9.2.3.2;9.3.2 Single-threaded time-stepped simulation engine;158
9.2.3.3;9.3.3 Parallelization on node-level;159
9.2.3.4;9.3.4 Parallelization on cluster-level;163
9.2.4;9.4 Test and verification;163
9.2.5;9.5 Empirical measurements;166
9.2.5.1;9.5.1 Simulation model used for benchmarks;166
9.2.5.2;9.5.2 Benchmark suite 0: Dependency between simulation time and runtime;170
9.2.5.3;9.5.3 Benchmark suites 1 and 2: Dependency between number of agents and runtime;172
9.2.5.4;9.5.4 Benchmark suite 3: Parallelization on node-level;177
9.3;10 Summary;180
10;Part IV Conclusions;183
10.1;11 Conclusions and outlook;184
10.1.1;11.1 Summary;184
10.1.2;11.2 Scientific contributions and applicability;185
10.1.3;11.3 Outlook;188
11;Part V Appendix;189
11.1;A Case studies of the GRAMS reference model;190
11.1.1;A.1 Case study: Warehouse;190
11.1.1.1;A.1.1 Structured problem description;190
11.1.1.2;A.1.2 Conceptual model;198
11.1.2;A.2 Case study: Joint Fire Support;211
12;Bibliography;213