Design and Control of Self-organizing Systems: Facing Complexity with Adaptation and Self-organization

188 Pages·2010·2.44 MB·English

by  Carlos Gershenson

Design and Control
of Self-organizing Systems
Carlos Gershenson
New England Complex Systems Institute
and Vrije Universiteit Brussel
Mexico City Boston Vic¸osa
Madrid Cuernavaca Beijing
CopIt ArXives
2007
CopIt ArXives
MexicoCity Boston Vic¸osa
Madrid Cuernavaca Beijing
Copyright 2007 by Carlos Gershenson
Published2007byCopItArXives
Allpropertyrightsofthispublicationsbelongtotheauthorwho,
however,grantshisauthorizationtothereadertocopy,printand
distributehisworkfreely,inpartorinfull,withthesoleconditionsthat
(i)theauthornameandoriginaltitlebecitedatalltimes,(iii)thetextis
notmodifiedormixedand(iii)thefinaluseofthecontentsofthis
publicationmustbenoncommercialFailuretomeettheseconditionswill
beaviolationofthelaw.
ElectronicallyproducedusingFreeSoftware
andinaccomplishmentwithan
OpenAccessspiritforacademicpublications
A
BSTRACT
Complex systems are usually difficult to design and control. There
are several particular methods for coping with complexity, but there
is no general approach to build complex systems. In this book I pro-
poseamethodologytoaidengineersinthedesignandcontrolofcom-
plex systems. This is based on the description of systems as self-
organizing. Startingfromtheagentmetaphor,themethodologypro-
poses a conceptual framework and a series of steps to follow to find
propermechanismsthatwillpromoteelementstofindsolutionsbyac-
tivelyinteractingamongthemselves. Themainpremiseofthemethod-
ologyclaimsthatreducingthe“friction”of interactionsbetweenel-
ementsofasystemwillresultinahigher“satisfaction”ofthesystem,
i.e. betterperformance.
A general introduction to complex thinking is given, since designing
self-organizing systems requires a non-classical thought, while prac-
tical notions of complexity and self-organization are put forward. To
illustratethemethodology,Ipresentthreecasestudies. Self-organizing
traffic light controllers are proposed and studied with multi-agent
simulations, outperforming traditional methods. Methods for im-
proving communication within self-organizing bureaucracies are ad-
vanced, introducing a simple computational model to illustrate the
benefits of self-organization. In the last case study, requirements for
self-organizing artifacts in an ambient intelligence scenario are dis-
cussed. Philosophical implications of the conceptual framework are
alsoputforward.
iii
iv Abstract
C
ONTENTS
Abstract iii
Contents v
ListofFigures viii
ListofTables x
Acknowledgements xi
1 Introduction 1
1.1 Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.2 Contributions . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.3 Outline . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.3.1 HowtoReadthisBook . . . . . . . . . . . . . . . . . . 5
1.3.2 HowtheBookWasWritten . . . . . . . . . . . . . . . 6
2 Complexity 9
2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
2.2 ClassicalThinking . . . . . . . . . . . . . . . . . . . . . . . . . 10
2.3 Complexity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
2.4 Indeterminacy . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
2.5 NonlinearityandChaos . . . . . . . . . . . . . . . . . . . . . 17
2.6 AdaptingtoComplexity . . . . . . . . . . . . . . . . . . . . . 19
2.7 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
3 Self-organization 23
3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
3.2 TheRepresentation-DependentDynamicsofEntropy . . . . 24
3.3 TheRoleoftheObserver . . . . . . . . . . . . . . . . . . . . . 29
3.4 OntologicalIssues . . . . . . . . . . . . . . . . . . . . . . . . . 30
3.5 Self-organization: APracticalNotion . . . . . . . . . . . . . . 32
v
vi CONTENTS
3.5.1 Artificialself-organizingsystems . . . . . . . . . . . . 33
3.5.2 Levelsofabstraction . . . . . . . . . . . . . . . . . . . 34
3.5.3 Copingwiththeunknown . . . . . . . . . . . . . . . . 34
3.6 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
4 AGeneralMethodology 37
4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
4.2 TheConceptualFramework . . . . . . . . . . . . . . . . . . . 38
4.3 TheMethodology . . . . . . . . . . . . . . . . . . . . . . . . . 44
4.3.1 Representation . . . . . . . . . . . . . . . . . . . . . . 44
4.3.2 Modeling . . . . . . . . . . . . . . . . . . . . . . . . . 46
4.3.3 Simulation . . . . . . . . . . . . . . . . . . . . . . . . . 54
4.3.4 Application . . . . . . . . . . . . . . . . . . . . . . . . 54
4.3.5 Evaluation . . . . . . . . . . . . . . . . . . . . . . . . . 55
4.3.6 NotesontheMethodology . . . . . . . . . . . . . . . 55
4.4 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
4.5 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
5 Self-organizingTrafficLights 63
5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
5.2 ApplyingtheMethodologyI . . . . . . . . . . . . . . . . . . 65
5.3 Experiments: FirstResults . . . . . . . . . . . . . . . . . . . . 71
5.4 ApplyingtheMethodologyII . . . . . . . . . . . . . . . . . . 78
5.5 Experiments: SecondResults . . . . . . . . . . . . . . . . . . 79
5.6 ApplyingtheMethodologyIII . . . . . . . . . . . . . . . . . . 84
5.7 Experiments: ThirdResults . . . . . . . . . . . . . . . . . . . 86
5.8 ApplyingtheMethodologyIV . . . . . . . . . . . . . . . . . . 87
5.9 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
5.9.1 Adaptationoroptimization? . . . . . . . . . . . . . . 89
5.9.2 Practicalities . . . . . . . . . . . . . . . . . . . . . . . . 90
5.9.3 Environmentalbenefits . . . . . . . . . . . . . . . . . 91
5.9.4 Unattendedissues . . . . . . . . . . . . . . . . . . . . 92
5.10 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93
6 Self-organizingBureaucracies 97
6.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98
6.2 DesigningSelf-organizingBureaucracies . . . . . . . . . . . . 100
6.3 TheRoleofCommunication . . . . . . . . . . . . . . . . . . . 102
6.3.1 DecisionDelays . . . . . . . . . . . . . . . . . . . . . . 106
6.4 TheRoleofSensors . . . . . . . . . . . . . . . . . . . . . . . . 106
6.5 TheRoleofHierarchies . . . . . . . . . . . . . . . . . . . . . . 108
CONTENTS vii
6.6 TheRoleofContext . . . . . . . . . . . . . . . . . . . . . . . . 111
6.7 AToyModel: RandomAgentNetworks . . . . . . . . . . . . 112
6.7.1 Usingself-organizationtoimproveperformance . . . 114
6.7.2 SimulationResults . . . . . . . . . . . . . . . . . . . . 115
6.7.3 RANDiscussion . . . . . . . . . . . . . . . . . . . . . 117
6.8 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125
7 Self-organizingArtifacts 127
7.1 AScenario . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128
7.2 Requirementsforself-organizingartifacts . . . . . . . . . . . 129
7.3 Achievingself-organization . . . . . . . . . . . . . . . . . . . 131
7.4 Learningtocommunicate . . . . . . . . . . . . . . . . . . . . 132
7.5 Learningtocooperate . . . . . . . . . . . . . . . . . . . . . . . 133
7.6 Learningtocoordinate . . . . . . . . . . . . . . . . . . . . . . 135
7.7 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137
8 Conclusions 139
8.1 Achievements . . . . . . . . . . . . . . . . . . . . . . . . . . . 140
8.1.1 Limitations . . . . . . . . . . . . . . . . . . . . . . . . 141
8.2 FutureWork . . . . . . . . . . . . . . . . . . . . . . . . . . . . 142
8.3 PhilosophicalImplications . . . . . . . . . . . . . . . . . . . . 144
8.3.1 ObjectivityorSubjectivity? Contextuality! . . . . . . 144
8.3.2 TheBenefitsofSelf-organization . . . . . . . . . . . . 145
Bibliography 147
Glossary 169
Index 173
L F
IST OF IGURES
1.1 Bookmap . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
2.1 Isitaduck,arabbit,orboth? . . . . . . . . . . . . . . . . . . 15
2.2 Thesamesphereseenfromdifferentangles . . . . . . . . . . 16
3.1 Entropyincreasesanddecreasesforthesamesystem . . . . 27
4.1 DiagramrelatingdifferentstagesofMethodology. . . . . . . 44
4.2 DetaileddiagramofMethodology. . . . . . . . . . . . . . . . 58
5.1 Screenshotofapartofthetrafficgrid . . . . . . . . . . . . . . 67
5.2 Resultsforstandardmethods . . . . . . . . . . . . . . . . . . 73
5.3 Resultsforself-organizingmethods . . . . . . . . . . . . . . 74
5.4 Fullsynchronization . . . . . . . . . . . . . . . . . . . . . . . 77
5.5 Secondresultsforstandardmethods . . . . . . . . . . . . . . 80
5.6 Secondresultsforself-organizingmethods . . . . . . . . . . 81
5.7 Comparisonofinitialandaveragenumberofcars . . . . . . 82
5.8 SimulationoftheWetstraatandintersectingstreets . . . . . 85
5.9 Wetstraatresults . . . . . . . . . . . . . . . . . . . . . . . . . . 95
5.10 Potentialimplementationofsotl-platoon. . . . . . . . . . . . . 96
6.1 Asynchronouscommunication . . . . . . . . . . . . . . . . . 104
6.2 Responsedelay . . . . . . . . . . . . . . . . . . . . . . . . . . 105
6.3 Hierarchyrepresentedasanetwork . . . . . . . . . . . . . . 110
6.4 DynamicsofarandomagentnetworkofN = 25,K = 5 . . . 113
6.5 RANself-organizationmechanism . . . . . . . . . . . . . . . 114
6.6 ResultsforN = 15,K = 1. . . . . . . . . . . . . . . . . . . . . 116
6.7 ResultsforN = 15,K = 2. . . . . . . . . . . . . . . . . . . . . 118
6.8 ResultsforN = 15,K = 5. . . . . . . . . . . . . . . . . . . . . 119
6.9 ResultsforN = 15,K = 15. . . . . . . . . . . . . . . . . . . . 120
6.10 ResultsforN = 100,K = 1. . . . . . . . . . . . . . . . . . . . 121
6.11 ResultsforN = 100,K = 2. . . . . . . . . . . . . . . . . . . . 122
6.12 ResultsforN = 100,K = 5. . . . . . . . . . . . . . . . . . . . 123
viii
LISTOFFIGURES ix
6.13 ResultsforN = 100,K = 100. . . . . . . . . . . . . . . . . . . 124
L T
IST OF ABLES
5.1 ParametersofNetLogosimulations. . . . . . . . . . . . . . . 72
5.2 Vehiclecountperhour,Wetstraat . . . . . . . . . . . . . . . . 85
5.3 EmissionsbyidlingenginesonWetstraat . . . . . . . . . . . 92
x