URBAN DYNAMICS MODELLING USING ANT

NEST BUILDING

Rawan Ghnemat, Cyrille Bertelle, Gerard H.E. Duchamp

To cite this version:

Rawan Ghnemat, Cyrille Bertelle, Gerard H.E. Duchamp. URBAN DYNAMICS MODELLINGUSING ANT NEST BUILDING. Marwan Al-Akaidi. MESM’2008, Aug 2008, Amman, Jordan.Eurosis-ETI, pp.124-128, 2008. <hal-00431231>

HAL Id: hal-00431231

https://hal.archives-ouvertes.fr/hal-00431231

Submitted on 11 Nov 2009

HAL is a multi-disciplinary open accessarchive for the deposit and dissemination of sci-entific research documents, whether they are pub-lished or not. The documents may come fromteaching and research institutions in France orabroad, or from public or private research centers.

L’archive ouverte pluridisciplinaire HAL, estdestinee au depot et a la diffusion de documentsscientifiques de niveau recherche, publies ou non,emanant des etablissements d’enseignement et derecherche francais ou etrangers, des laboratoirespublics ou prives.

URBAN DYNAMICS MODELLING USING ANT NEST BUILDING

Rawan Ghnemat(1), Cyrille Bertelle(1), Gerard H.E. Duchamp(2)

(1)LITIS - University of Le Havre, 25 rue Philippe Lebon, BP 540, 76058 Le Havre Cedex, Franceemail: rawan.ghnemat@litislab.eu, cyrille.bertelle@litislab.eu

(2)LIPN - University of Paris XIII, 99 avenue Jean-Baptiste Clement, 93430 Villetaneuse, Franceemail: ghed@lipn.univ-paris13.fr

KEYWORDS

swarm intelligence, complex systems, self-organization,ant systems, spatial organization

ABSTRACT

Urban dynamics deal with spatial organizations where agreat complexity of interactions appears. Social and eco-nomic aspects interact and environmental objectives arenowadays a major purpose for sustainable urban devel-opment. We propose some generic modelling processesable to face with this complexity, in order to simulatethe evolution of the city centers. These organizationalcenters need a multi-criteria description for their evolu-tion, including feed-back phenomena of them over theirenvironment and components. We propose a swarmintelligence algorithm, using social-insect collective be-havior. We combine a decentralized approach, basedon emergent clustering mixed with spatial constraintsor attractions, as an extension of the ant nest buildingalgorithm with multi-center. Typically, this model iscurrently used by ourself, to model and analyze culturalequipment dynamics in urban area.

INTRODUCTION

Many natural and artificial systems have emergentproperties based on spatial development. This spatialdevelopment is both the result of some mecanisms fromthe system behavior and the actor of the system forma-tion by morphogenetic feedback. Natural ecosystemsor social organizations in urban dynamics are typicallysuch emergent spatial organizations. The goal of thispaper is to study some computable mecanisms andalgorithms able to model such spatial self-organizationprocesses, taking into account the complexity of thephenomena.

In the next section, complex system concepts are definedand their applications to urban dynamics understandingare described, then swarm intelligence algorithms are

described as methodologies to implement the complexsystems concepts, using distributed computations. Thefollowing section proposes some specific swarm intelli-gence methods based on ant systems in order to modelthe spatial organizations emergence. Experiments aregiven, using RePast muti-agent platform mixed with ageographical information system.

Figure 1: Spatial organizations Complexity Descriptionand the Conceptual Generic Model Based on Swarm In-telligence

MODELLING SPATIAL COMPLEXITY

Complex System Concepts

Complex system theory (7) is based on the fact that formany applicative domains, we can find similar processeslinking emergent global behavior and interaction net-work of constituents. The global behavior is generallynot accessible using classical analytical methods.

In classical analytical methods, the global behavior ofthe system is the description of the equations. Simula-tions from these formulations, consist in obtaining thetrajectories of the behavior predefined in the equation

formulation.

In complex systems modelling, we have to modelthe constituents of the system and the interactionnetwork or system which link these constiuents, usinga decentralized approach. So the global behavior of thesystem cannot be understood by the description of eachconstituent. In complex system modelling, the globalbehavior is an emergent property from the interactionnetwork or systems between its constituents and leadto the creation of an dynamical organization of theseconstituents

This dynamical and emergent organization retro-actson its own components. Two kinds of feedback allowto describe these phenomenon. The positive feedbackmeans that the emergence increases the organizationconstitution. the negative feedback means that theemergence has regulator properties which finally stopthe increasing organization constitution and allow thesystem stabilization.

Another major aspect of complex systems is that theycan be considered as open systems. This means thatthey are crossed by energetical fluxes that make themevolve in a continuous way. From these energeticalfluxes, complex systems can evolve through criticalstates, using bifurcation schema and attractors behav-iors. One of the major vector or support of these ener-getical fluxes is the environment itself where the com-plex systems and their entities evolve. In many naturaland artificial systems, the environement has some spa-tial effects which interact on the whole complexity of thephenomenon. This spatial environment can be modifiedby the system but he can also be the catalyst of its ownevolution. Understanding and modelling the deep struc-tural effect of the interaction between the systems andits spatial environment is the goal of the study presentedin the sequel.

Application to urban dynamics

Social and human developments are typical complexsystems. Urban development and dynamics are theperfect illustration of systems where spatial emergence,self-organization and structural interaction betweenthe system and its components occur. In figure 1, weconcentrate on the emergence of organizational systemsfrom geographical systems. The continuous dynamicdevelopment of the organization feed-back on thegeographical system which contains the organizationcomponents and their environment. The lower partof this figure explains our analysis methodology. Itconsists to describe many applicatives problems bydynamical graphs or environments in order to detectorganizations over these dynamical environment. Forthe organization detection, we use swarm intelligence

processes. We model the feed-back process of thisemergent organization on the system constituentsand its environment. To analyse or simulate urbandynamics, nowadays, we can use the great amount ofgeographical databases directly available for compu-tational treatment within Geographical InformationSystems. On the organizational level description, thenew development of multiagent systems (MAS) allowsnowadays to develop suitable models and efficientsimulations.

The applications we focuss on in the models thatwe will propose in the following concerns specificallythe multi-center (or multi-organizational) phenomonainside urban development. As an artificial ecosystems,the city development has to deal with many chal-lenges, specifically for sustainable development, mixingeconomical, social and environmental aspects. Thedecentralized methodology proposed in the followingallows to deal with multi-criteria problems, leadingto propose a decision making assistance, based onsimulation analysis.

Cultural dynamics processes in urban can be modelledusing such methodology. It is typically a multi-criteriaself-organization processes where appears emergent us-age oo urban equipment. A modelling of these dynamicsis presented latter in this paper.

Swarm Intelligence and Spatial Environment

Decentralized algorithms have been implemented formany years for various purposes. In this algorithmcategory, multi-agent systems can be considered asgeneric methods (11). We deal, in this paper, withagents based on reactive architecture and expected tobe used inside numerous entity-based systems. The aimof programs using such architectures, is to deal withemergent organizations using specific algorithms calledemergent computing algorithms. Swarm Intelligenceis the terminology used to point out such reactiveagent-based methods where each entity is built withthe same basis of behavior, but reacts in autonoumousway.

Based on swarm intelligence concepts, Ant systems (3)propose some bio-inspired method from social insectsmecanisms. For exemple, Ant Colony Optimization(ACO) methods is a such method family where the ba-sic entities are virtual ants which cooperate to find thesolution of graph-based problems, like network routingproblems, for example. Using indirect communications,based on pheromon deposites over the environment(here a graph), the virtual ants react in elementaryway by a probabilistic choice of path weighted with twocoefficients, one comes from the problem heuristic andthe other represent the pheromon rate deposit by all

the ants until now. The feed-back process of the wholesystem over the entities is modelled by the pheromonaction on the ants themselves.

Particule Swarm Optimization (PSO) is a metaheuris-tic method initially proposed by J. Kennedy and R.Ebenhart (6) which is also based on swarm intelligenceconcepts. This method is initialized with a virtualparticle set which can move over the space of solutionscorresponding to a specific optimization problem.The method can be considered as an extension of abird flocking model, like the BOIDS simulation fromC.W. Reynolds (9). In PSO algorithm, each virtualparticle moves according to its current velocity, its bestprevious position and the best position obtained fromthe particles of its neighborhood. The feed-back processof the whole system over the entities is modelled bythe storage of this two best positions as the result ofcommunications between the system entities.

Other swarm optimization methods have been de-velopped like Artificial Immune Systems (4) whichis based on the metaphor of immune system as acollective intelligence process. F. Schweitzer proposesalso a generic method based on distributed agents,using approaches of statistical many-particle physics(10).

The method proposed in this paper is based on Ant sys-tems, allowing to deal with self-organization processesemerging from spatial constraints and attractive areas.

MULTI-CENTER ANT NEST BUILDING TO

MODEL EMERGENT SPATIAL ORGANIZA-

TIONS

We will describe in this section, the general algorithmwhich is proposed to model emergent spatial organiza-tions. This algorithm is based on the ant clustering.We introduce pheromon template to spatially controlthe clustering from local attraction. This method is adecentralized approach which allows to combinate multi-center and multi-criteria problem and we will show howwe can apply it to model cultural dynamics in urbanareas.

Ant clustering

Ant clustering algorithms are inspired by the corposesor larvea classification and aggregation that the antscolony are able to do in the real life. The ants aremoving inside a closed area and are able to move somematerial which are randomly put on this area. Aftera while, and without any kind of centralized coordo-nation, the ants success to create some material clusters.

The algorithm is based on the following and very simple

behavioral rules that each ant implements :

• When an ant is moving without carrying yet ma-terial and find some material, the ant will take thematerial respecting the probability number :

Pp =

(

k1

k1 + f

)2

(1)

where f is the material density that the ant per-ceives locally around itself and k1 is the treshold.It is easy to check that if f << k1 then Pp is nearthe value 1 and if f >> k1 then Pp is near thevalue 0.

• When an ant is moving when carrying some mate-rial, the probability to deposite it is computed by :

Pd =

(

f

k2 + f

)2

(2)

where f is still the material density that the ant per-ceives locally around itself and k2 is another tresh-old. It is easy to check that if f << k2 then Pd isnear the value 0 and if f >> k2 then Pd is near thevalue 1.

Spatial constraints using template

The ant clustering shows some spatial self-organizationsbut has the specificity to generate clusters at randomplaces. According to the first random moves that theants start to do in the beginning of the algorithm, somematerial will initiate aggregation and the clusteringprocessus will complete this aggregation from theseinitial random first aggregations. To simulate someurban dynamics, we need to introduce specific locationwith respect to city center for example or culturalequipments. The clustering here will represent thepeople usage of these centers or equipments and we needto introduce an attractive effect by using a pheromontemplate. This method follow the algorithm knownas Ant Nest Building (3). In ant colonies, the centercorresponds to the position of the queen which needsto build the nest and the ant colony moves around itto protect the nest by various material taken on theground. The queen emits a pheromon which allowsto attract the ants during their building. The ant hasto deposite the material carried only if the pheromonquantity perceived belongs to a specific range. We usean attractive fonction called Pt, corresponding to apheromon template and represented by the part (a) ofthe figure 2.

Using this template function, we remplace in the clus-tering algorithm, the two provious probabilities definedin equation (1) and equation (2) by

(a) Template function

(b) Simulation on RePast

Figure 2: Ant nest building with one center usingRePast MAS paltform over OpenMap GIS

P ′

p = Pp(1 − Pt) (3)

P ′

d = PdPt (4)

In figure 2, we show an implementation of this algo-rithm using the multi-agent platform called Repast (8).The java version of this platform includes some pack-ages allowing to interface with geographical databaseand geographical information systems (GIS). The graph-ical output windows is made under OpenMap which is aGIS developped in Java. In figure, the materials movedby the ants are the small grey circles, the ant movingwithout material are the green circles, the ant carryingmaterial are the red circles and the queen location is theyellow circle.

Multi-template modelling

The previous subsection describes one local attractiveprocess characterized by the queen and its pheromontemplate emission. The advantage of this method is tobe able to combine the solutions of multi-center and

(a) Template functions

(b) Simulation on RePast

Figure 3: Multi-center ant nest buiding using RePastover OpenMap

multi-criteria problems, using interactive processes,each one is represented by a queen and its pheromontemplate.

On the figure (3), we can see a simulation with twoqueens and two pheromon templates. It is possiblealso for each queen to emit many different kinds ofpheromons : we called them colored pheromons. Eachcolored pheromon will attract only the ants associatedto its color.

Application to cultural equipment dynamics

The multi-template modelling can be used to model cul-tural equipment dynamics as described in the figure (4).On this figure, we associate to each cultural center (cin-ema, theatre, ...) a queen. Each queen will emit manypheromon templates, each template is associated to aspecific criterium (according to age, sex, ...). Initially,we put the material in the residential place. Each mate-rial has some characteristics, corresponding to the peo-ple living in this residential area. The simulation showsthe self-organization processus as the result of the set

Figure 4: Cultural Equipment Dynamics Modelling

of the attractive effect of all the center and all the tem-plates.

Conclusion and Perspectives

The paper develops some specific swarm intelligence al-gorithms based on ant colonies processes. Using such de-centralized methods, we can model complex multi-centerand multi-criteria self-organizations. Urban dynamicsare one of the most relevent problems where these ap-proaches can be efficient. These studies are supportedby a french regional project (Haute-Normandie) dealingwith the study of cultural dynamics over urban area.

REFERENCES

[1] Benenson, I., Torrens, P.M. (2004) Geosimulation -Automata-based modeling of urban phenomena, Wi-ley.

[2] C. Bertelle, G.H.E. Duchamp, H. Kadri-Dahmani(eds), Complex Systems and Self-Organization Mod-elling, “Understanding Complex Systems” series,Springer, 2008 (in press).

[3] Bonabeau, E., Dorigo, M., Theraulaz, G. (1999)Swam Intelligence, from natural to artificial systems,a volume in the Santa Fe Institute Studies in the Sci-ences of Complexity, Oxford University Press.

[4] De Castro, L.N., Timmis, J. (2002) Artificialimmune system: a new computational approach,Springer-Verlag, London, U.K.

[5] Ghnemat, R., Bertelle, C., Duchamp, G.H.E. (2007)Adaptive Automata Community Detection and Clus-tering, a generic methodology, in Proceedings ofWorld Congress on Engineering 2007, InternationalConference of Computational Intelligence and Intel-ligent Systems, pp 25-30, London, U.K., 2-4th July2007.

[6] Kennedy, J., Eberhart, R.C. (1995) Particle SwarmOptimization In Proceedings of IEEE InternationalConference on Neural Networks (Perth, Australia),IEEE Service Center, Piscataway, NJ, 5(3), pp 1942-1948.

[7] Le Moigne, J.-L. (1999) La modelisation dessystemes complexes, Dunod.

[8] Repast web site (2008)http://repast.sourceforge.net.

[9] Reynolds, C.W. (1987) Flocks, Herds and Schools: adistributed behavioral model In Computer Graphics,21(4) (SIGGRAPH’87 Conference Proceedings), pp25-34.

[10] Schweitzer, F. (2003) Brownian Agents and ActiveParticles, Springer.

[11] Weiss, G. (ed.) (1999) Multiagent Systems, MITPress.

[12] Xiao, N. (2005) Geographic optimization using evo-lutionary algorithms In 8th International Conferenceon GeoComputation, University of Michigan, USA.