Steps Towards Simulating Smart Cities and Smart Islands with a Shared

Generic Framework

A Case Study of London and Reunion Island

Tahina Ralitera

, Maxime Ferard

, Gonzalo Bustos-Turu

, Koen H. Van Dam

and R

emy Courdier

Laboratoire d’Informatique et de Math

ematiques (LIM), University of Reunion Island,

Parc Technologique Universitaire, Saint-Denis, France

Department of Chemical Engineering, Imperial College London, South Kensington, London, U.K.

Keywords:

Agent based Simulation, Smart Island, Smart City, Model Re-use, Replication, Software Reﬁnement.

Abstract:

Simulation models can be used as decision support tools for smart city design and planning. They allow to

evaluate the possible consequences of projects, before their implementation in the real world. Decision makers

could beneﬁt from replicable ones that can be relevant and easily transferable from one territory to another so

solutions can be compared and re-use of model components can save time. In this paper we consider the case

of citizen’s mobility ﬂow simulation. However, most of such simulation models are designed to be suitable

for a speciﬁc kind of territory. Some of them are reusable, but in a context that does not differ much from

the original one for which they were designed, or require lots of changes to be relevant in another context.

We classify those contexts into urban and insular and we show that despite their difference, they could be

complementary. We demonstrate that testing a simulation model designed for an urban context, in a context

with strong constraints can help in its consolidation. Thereby, after testing an Agent Based Simulation Model

originally applied to a case study in London, in Reunion Island, we present a more generic simulation model

that works for both systems.

1 INTRODUCTION

Currently, combining competitiveness and sustain-

able urban development in cities becomes a challenge.

Consequently, government, citizens and stakeholders

have to search for technical solutions to reduce eco-

nomic and environmental crises that come with the

growth of urbanization. On the other hand, since these

last decades, the development of ICTs has contributed

to great changes in cities in all points of view (eco-

nomic, cultural, transport...). Those trends have led

to the growing popularity of the concept of ”smart

city” (Dijkstra et al., 2013), deﬁned by (Longo et al.,

2014) as a ”place and territorial context, where use of

planned and wise of the human and natural resources,

properly managed and integrated through the various

ICT technologies already available, allows for the cre-

ation of an ecosystem that can be used of resources

and to provide integrated and more intelligent sys-

tems”.

A city system is considered as a complex system

(Batty et al., 2012) characterised by:

• Micro-interactions of both human-human and

human-environment;

• Emergence of unexpected phenomena that arise

from the behaviour of independent units;

• Non-linear dynamics i.e. it is difﬁcult to predict

the output of the system from its inputs;

• Feedback loops.

It is composed of different recognizable subsets, in-

cluding individuals with various behaviours, who are

no more considered only as users of the system, but as

part of it. Proposed smart city solutions should there-

fore consider those characteristics and should be able

to adapt to the individuals and to the environment in

which they are. Agent-Based Modelling is a promis-

ing solution in developing such a complex system. It

allows to break down city into smaller simpler subsets

handled by several autonomous, social, reactive and

proactive entities. An approach that allows to model

non-linearity and favors the emergence of necessary

standards and protocols needed for the Smart Cities

(Roscia et al., 2013).

Ralitera, T., Ferard, M., Bustos-Turu, G., Dam, K. and Courdier, R.

Steps Towards Simulating Smart Cities and Smart Islands with a Shared Generic Framework - A Case Study of London and Reunion Island.

DOI: 10.5220/0006371203290336

In Proceedings of the 6th International Conference on Smart Cities and Green ICT Systems (SMARTGREENS 2017), pages 329-336

ISBN: 978-989-758-241-7

329

Different projects are done and simulation models

are often used by policy makers to evaluate their pos-

sible consequences before their application in the real

world. It would be therefore interesting to have repli-

cable simulation models that could be easily trans-

ferable from one context to another. However, cur-

rently, many of them are designed for a speciﬁc en-

vironment and are only transferable to a context that

do not differ much from the original one. Making

experimentations in insular territory such as islands

could be interesting for the consolidation of a simu-

lation model designed for urban territories, as islands

might be considered as an ideal candidate for tests un-

der real-life conditions before later implementation in

urban areas. In this paper, the SmartCityModel, an

agent based simulation model, originally applied to a

context of urban systems such as London, is tested.

Insights gained from applying this model to Reunion

Island help develop this framework further to be more

widely usable in different contexts and thereby con-

tributes to a generic simulation model presented here.

This paper is structured as follows, section 2 gives

an overview of smart cities, smart islands and agent

based simulations of mobility ﬂow as stated in the

literature. Section 3 is about a comparative study

between urban and island systems and section 4 de-

scribes the experimentations. Finally, Section 5 re-

ﬂects the approach and discusses the contributions.

2 RELATED WORK

2.1 Smart Cities and Smart Islands

The ”smart city” is considered as a solution that could

solve the problems linked to the growth of the urban-

ization and the increase of the environmental and eco-

nomic crises in the city. Currently, there is still no uni-

versal deﬁnition, but (Longo et al., 2014) deﬁne it as

a ”place and territorial context, where use of planned

and wise of the human and natural resources, prop-

erly managed and integrated through the various ICT

technologies already available, allows for the creation

of an ecosystem that can be used of resources and to

provide integrated and more intelligent systems”. A

smart city is well performed in 6 key ﬁelds (Gifﬁn-

ger, 2011), related to some urban life aspects (Albino

et al., 2015): smart economy, smart mobility, smart

environment, smart people, smart living, smart gover-

nance. The experimentation described in this paper is

focused on the smart mobility part of this smart city,

but could also be applied to the other parts.

Islands are also facing the same problems as cities,

to which additional constraints are added. Indeed, due

to their remoteness, their isolation and other speciﬁc

geographical constraints, islands are facing multiple

challenges in managing energy, resources, transport...

Those lead to think about ”smart islands”. Currently,

to our knowledge, the concept of smart city is sub-

ject to lots of study, however few works are talking

about smart islands. And most of the time, the given

solutions suit to a particular territory (Choo, 1997)

(Gioda, 2015). Sometimes, however, it is interest-

ing to have a re-usable simulation model that could

be easily transferable from one territory to another, by

taking into account characteristics which are proper to

different types of territories such as cities and islands

(Comparison will be discussed later, at Section 3).

2.2 Agent based Simulation Models

Different approaches were used in the literature for

simulating transport mobility: trip-based approach,

tour-based approach and activity-based approach.

The proposed simulation model uses activity-based

approach which supposes that travel derived from

the demand for personal activities (work, shopping,

leisure...) that individuals need or wish to perform

(McNally and Rindt, 2007). In the example, the

SmartCityModel (Bustos-Turu et al., 2014) (see Sec-

tion 4.1) is used to simulate the electric transport and

the impact of that on the electricity consumption, de-

pending on agents activity schedule. Similar works

can be found in the literature (Sweda and Klabjan,

2011) but none take into account the speciﬁc con-

straints that could be very important for some kind of

territory such as oceanic islands. For example, they

do not consider the geographical parameters such as

road slopes (Maia et al., 2011). Also, transport be-

haviour considerations such as vehicle speed or the

use of comfort items, that could also inﬂuence the

energy consumption (Bingham et al., 2012) (Karaba-

soglu and Michalek, 2013), are usually neglected. So,

such simulation models may not be transferable in

the context of a territory where these missing param-

eters could strongly inﬂuence the electric vehicle’s

consumption. Consequently, testing such simulation

models in a very different context, such as an oceanic

island one, could help to improve their generic nature

by detecting such missing parameters. It is also inter-

esting to explore how these features could be added to

an existing approach.

To support these proposals, a comparative study

between urban and insular context is done in the next

section, followed by experimentations on the particu-

lar case of London and Reunion Island.

SMARTGREENS 2017 - 6th International Conference on Smart Cities and Green ICT Systems

330

3 COMPARATIVE STUDY

3.1 Urban and Insular Systems

An urban area generally contains one or several of

the following features: local municipality, large pop-

ulation, high population density, and economic ac-

tivity which does not mainly depend on primary re-

sources exploitation (Unicef et al., 2012). However, it

is strictly deﬁned at a national level, but there is no in-

ternational agreement on this topic. This raises impor-

tant issues in the comparison of urban zones located in

different countries, as diverging city deﬁnitions might

create misleading results. It must also be noticed that

the administrative deﬁnitions of cities boundaries are

not covering the entire real city. However, the major

part of the data are from the local government, and are

therefore based on that deﬁnitions. As a result, using

another boundary would strongly increase the com-

plexity of the research, both in terms of geographic

limits setting and data collection.

Many simulation models are tested only on ur-

ban systems (Badariotti and Weber, 2002) (Sweda and

Klabjan, 2011) (

Certick

y et al., 2015), so there could

be a problem in translating them in systems with dif-

ferent geographical and socio-cultural characteristics

such as insular systems.

Insularity refers to areas with more or less marked

spatial boundaries that identify, at least in theory, a

set of elements from diverse origin, participating in

the same territorial dynamics. It is therefore favor-

able to the study of the functioning and the viabil-

ity of systems, and the islands are interesting exam-

ples (Magnan, 2009). Indeed, islands are pieces of

lands surrounded by water and which are above wa-

ter at high tide (Nations, 1994). Their boundaries are

already set by geographic constraints and are time-

independent. Moreover, unless they have a direct link

with the mainland (e.g. electrical network, bridge),

every ﬂows that come from and to the islands are eas-

ily traceable as they occur by plane or boat. The usual

classiﬁcation of islands is also relatively clear (ocean

and continental) and independent of the local govern-

ment. Apart from these facts, islands can be home to

different cities and towns. In this respect, the previ-

ous considerations about urban system characterisa-

tion are also valid for islands having high population

levels and urban centres.

3.2 Comparison

When comparing them, island and urban systems

present very different structural features. Because of

their relative differences, the two systems could re-

act in very different ways to the same changes. Ta-

ble 1 summarizes the comparison of the two systems.

Indeed, urban areas are hard to deﬁne and to break

down into elementary units, whereas islands have

ﬁxed boundaries and a more common structure due

to their geographical characteristics. From a system

perspective, islands appear to be easier to deﬁne and

study. They might be considered as an ideal candidate

for tests under real-life conditions before later imple-

mentation in urban areas. However, despite their dif-

ferences, islands and cities could be complementary.

Therefore, making experiments on both cities and is-

lands can generate useful insights for the discussion

of transferability of smart city solutions from islands

to cities.

4 CASE STUDY

4.1 The Simulation Model

Proposed simulation SmartCityModel is built upon

the free and open source Agent-Based Modelling and

Simulation platform Repast Simphony (North et al.,

2013). It was used for different case study around

urban systems, such as for simulating the interac-

tions between land use, transport and electric ve-

hicle charging demand (Bustos-Turu et al., 2015),

residential electricity and heat demand (Bustos-Turu

et al., 2016), estimating plug-in electric vehicle de-

mand ﬂexibility (Bustos-Turu et al., 2014) or for mod-

elling water and sanitation infrastructure use in urban

systems (The Ecological Sequestration Trust (TEST),

2016).

In this paper, the application area is electric trans-

port and the example simulate Plug-in Electric Ve-

hicle (PEV) ﬂow in the city. PEV owners are mod-

elled as agents who take travel and charging decisions

based on their perceptions and memories. They use

an activity schedule and are capable to adapt their be-

haviour given a particular situation such as a reduc-

tion in the charging tariff. The environment, in which

they are situated and act, is represented by a collec-

tion of objects, each representing a fragment of the

modelled physical reality. It is deﬁned using a GIS

representation (shape ﬁles).

To summarise, the simulation model takes, as in-

put, GIS and statistical data. As output, it generates

metrics such as electricity consumption and charging

station usage proﬁles, which are stored in .csv ﬁles.

Those .csv ﬁles could be used for further studies, us-

ing post-processing software. Figure 1 shows the ini-

tial experiment process of the simulation model.

Steps Towards Simulating Smart Cities and Smart Islands with a Shared Generic Framework - A Case Study of London and Reunion Island

331

Table 1: Key structural differences between islands and cities.

Islands Cities

Physical boundaries Set by geography, Constant Many possibilities, Time-dependant

Geographical constraints Often strong in oceanic island Often small or null

In/out ﬂows Easy to evaluate Hard to determine

Figure 1: SmartCityModel initial experiment process.

Experiments are carried out in London and Re-

union Island, in 2 steps: a comparative analysis be-

tween the 2 contexts and experiments by simulation.

The approach adopted follow the three steps exper-

iment process used in (

Certick

y et al., 2015): ex-

perimentation speciﬁcation (scenario deﬁnition and

setup), simulation execution, result analysis and vi-

sualisation. This allows not only to make adjustments

in the parameterization of the simulation, but also to

take into account all the aspects going from the initial-

ization to the interpretation of the results. Aspects that

seem important to consider because they are all linked

and contribute to the generic nature and the relevance

of the simulation model.

4.2 London and Reunion Island

London (Figure 2) is the capital of the United King-

dom. It is an urban area of 1,572 km

, populated by

8,538,700 inhabitants (in 2014). Situated in the north

hemisphere, it has a temperate oceanic climate.

Figure 2: London map.

Reunion Island (Figure 3) is a French island sit-

uated in the south hemisphere, in the Indian Ocean,

nearby Madagascar and Mauritius. It is an oceanic

island of 2,512 km

, with a tropical climate and pop-

ulated by 843,529 inhabitants (in 2015).

Figure 3: Reunion Island map.

First of all, as opposed to London, which has mov-

ing boundaries, Reunion Island has boundaries that

are delimited by the sea. Moreover, it is a completely

remote island and in and out ﬂows are easier to de-

ﬁne. The center of the island is a national park, with

a very high elevation, so population housing and mo-

bility are constrained by the physical characteristics.

Moreover, Reunion Island has a contrasted relief, with

an altitude variation from 0 to 3070.50 m (the Piton

des Neiges), while in London the slope is negligi-

ble. Consequently, if in London, the population and

their mobility are scattered across the territory, in Re-

union Island, they are concentrated in the coastal ar-

eas. Second, due to their different geographical posi-

SMARTGREENS 2017 - 6th International Conference on Smart Cities and Green ICT Systems

332

tion, their environment, their relief and their climate

are also different. The average temperature in London

is 12.4

◦

C against 27.7

◦

C in Reunion Island. However,

in Reunion Island the temperature can be lower in

the mountainous areas and higher in the coastal areas.

The average precipitation is also higher in Reunion is-

land: around 10 000 mm per year against 622.5 mm.

Therefore, the use of comfort items in vehicles such

as air conditioning, heating, wipers, etc. is not the

same. Finally, compared to its small size, London is

10 times more populated than Reunion Island.

To sum up, to be transferable to a context sim-

ilar to Reunion Island, the simulation model should

take into account the parameters listed in table 2 (non-

exhaustive list).

Table 2: Non-exhaustive list of parameters.

Classiﬁcation Examples

Geographical Slope

Socio-cultural

Population density and

geographical distribution, speed

Climate related

Use of comfort items,

temperature variation

4.3 Experimentations

In this experimentation, the SmartCityModel is used

to implement a Reunion Island case, what entails

changes in the maps, the statistical data used and the

agent behaviour (mobility concentrated in the coastal

roads for example).

4.4 First Improvements

First experimentations on Reunion Island revealed

some limitations. Limitations that could also appear

in other simulation models and in similar territories

(oceanic islands or territories with high geographical

constraints). The experimentations follows the three

steps cited in section 4.2. The results and the dis-

covered limitations and problems are summarised in

Table 3. Improvements are done following the same

3 previous levels:

4.4.1 Initialisation of the Simulation

The initialization process of the simulation model is

reorganized, based on works done by (

Certick

y et al.,

2015). Now, a .csv ﬁle is used to deﬁne the sce-

nario parameters and the display is set up automati-

cally. Consequently, the user will no longer have to

change the source code of the program or the display,

manually. Moreover, wrong ﬁle, variable formats and

missing attributes, are checked at this level, in order

to avoid errors that could appear during the simulation

run.

4.4.2 Simulation Core

Improvements can be classiﬁed in 2 categories:

First, there are improvements in the algorithm. In

fact, to address the problem of unrealistic geographi-

cal population distribution (see Table 3), a buffer zone

is added around the road network. It allows to gener-

ate agents location and activities areas nearby roads.

Therefore, population distribution is more realistic:

concentrated in the coastal areas and the few people

who are in the center of the island live nearby roads.

The following pseudo-code show how the buffer zone

is used to generate a random location nearby a road:

set BUFFER_DISTANCE to buffer radius

Initialise nearestRoad to null

while nearestRoad==null do

set coords = random coordinate inside

a building

set nearestRoad = the road where the

distance from coords is equal

to BUFFER_DISTANCE

end do

return coords;

Then, an alternative method that consists of deﬁn-

ing the activity area locations from shape ﬁles (one

shape ﬁle for an activity zone) is created to address

the problem of missing land use distribution data used

for the generation of the activity areas location. Gen-

erated locations correspond to the precise location of

the activity zones in the real world. However, we kept

both methods. The method used is chosen automati-

cally at the initialisation of the simulation, according

to available data.

Second, there are improvements in the simulation

setting. It is about the implementation of the param-

eters listed in Table 2. Parameters that, according to

studies described in Section 4.2 could inﬂuence the

relevance of the simulation results. This part is still in

progress, but currently the speed variation is already

taken into account in the simulation. For that, a new

attribute is added to the road shape ﬁles. This new at-

tribute takes the value of ”majorRoad” if the road is

a major road. Then a speed advantage is assigned for

car drivers when driving on a major road.

4.4.3 Visualisation of the Simulation Results

It is interesting to link the simulation platform with

post-processing tools in order to have an automated

treatment as well as a more adequate visual exploita-

tion of simulation results. (Augusseau et al., 2013)

Steps Towards Simulating Smart Cities and Smart Islands with a Shared Generic Framework - A Case Study of London and Reunion Island

333

Table 3: Experimentations on Reunion Island.

Step Details Limitations

Simulation initialisation

File processing

Shape ﬁles have to be contained in a speciﬁc folder, if not, need to change the path in the source code

Shape ﬁles content are not checked so that induce errors during the simulation execution

Display conﬁguration Manual

Scenarios deﬁnition Need to set global variables value in the source code

Simulation core

Population distribution

Unrealistic population distribution

Large number of agents are located in the center of the island which in reality is a very sparsely populated area

Data availability Some important data are not available. For example: land use distribution (percentage of the different activity zones per commune)

Simulation conﬁguration Geographical, socio-cultural and other parameters not taken into account (slope, speed, etc.)

Results

A post-processing step must be done for analysis and visualization, using different tools

and (

Certick

y et al., 2015) show that having a spa-

tialized display of the simulation metrics in an earth

browser such as Google Earth can be interesting. It

gives the possibility to replay a previous simulation

run, particularly useful for validation by domain ex-

perts. Indeed, recording a model as a video preserves

the movement, but removes inspection capabilities.

Moreover, it is then difﬁcult to include the animation

in web applications and to layer it with other spatial

data sets. To address this, an export mechanism which

records the simulation and its metrics and exports it

to KML format is developped. It allow to incorporate

data in displays using an interactive geobrowser. This

export function can be used to create a web interface

showing the output of a simulation run and the varia-

tion in the time of the metrics across different parts of

the simulated area. An example of display is shown

on ﬁgure 4.

Figure 4: An example of results display on Google Earth.

4.5 Results

Figure 5 shows the new process of the SmartCity-

Model. The 3 parts (scenario deﬁnition, simulation

execution, result analysis and visualisation) are sepa-

rated into 3 connected blocks. This new organization

make the initialisation of the simulation and the sce-

nario deﬁnition simpler, so the user do not need to be

familiar with Repast Simphony. Changing the simu-

lated territory is easier as the user only have to add

the shape ﬁles and the display is conﬁgured automat-

ically. Moreover, results visualization is better and

there is an automatic ﬁrst results post processing.

5 DISCUSSION AND FUTURE

WORKS

The paper present experimentations done on an Agent

Based Simulation model originally applied to a case

study in urban systems, on strong constraints systems

such as insular systems. By taking a model previously

used in a London case study and applying it to Re-

union Island by a new team of modellers, some limi-

tations were detected and improvements are done on

3 levels:

• The scenario deﬁnition and setup;

• The simulation execution;

• The result analysis and visualisation.

They show that making such experimentations could

generate useful insight that can help at improving the

transferability of such simulation models from one

context to another. Most of the functionality imple-

mented are generic enough to be easily transferable

to other contexts. However, there are still some fur-

ther improvements and veriﬁcations that should be

done to validate the simulation model, in particular,

SMARTGREENS 2017 - 6th International Conference on Smart Cities and Green ICT Systems

334

Figure 5: SmartCityModel 3 steps process, after improvements.

in the simulation core by implementing missing es-

sential geographical and socio-cultural parameters.

One of the most important is the slope. Indeed,

currently, in the SmartCityModel, the agent route

is chosen following the shortest path algorithm, a

method that could be improved by adding more cri-

teria for the choice of the road. For example, for

energy saving and vehicle power reason, an agent

could choose a road with less slope even if it is longer

in distance instead a short one but with high slope.

This could produce the emergence of a collective phe-

nomenon that lead to the concentration of the trafﬁc in

coastal roads, where slopes are lower. A phenomenon

that corresponds to the reality in the Island. More-

over, it has also impact on efﬁciency, especially with

EV used as case. As we need to know how much en-

ergy is used, maybe it is beneﬁcial to charge at the top

of the mountain than at the bottom. And it is applica-

ble not only for islands case but also for cities with big

hills or even mountains such as Rio de Janeiro of Lis-

bon where these improvements following the islands

application will be relevant.

So, the future work will be ﬁrst of all focused on

the implementation of those geographical and socio-

cultural parameters. Then, we can proceed to vali-

dation by experimentations on other urban and island

systems. Finally, veriﬁcation of the simulation results

could be done by domain experts.

REFERENCES

Albino, V., Berardi, U., and Dangelico, R. M. (2015). Smart

cities: Deﬁnitions, dimensions, performance, and ini-

tiatives. Journal of Urban Technology, 22(1):3–21.

Augusseau, X., Aulanier, F., Begue, A., Bourgoin, J.,

David, D., David, G., Degenne, P., Fargette, M.,

Lagabrielle, E., Lajoie, G., Laurent, J.-B., Lebour-

geois, V., Lestrelin, G., Lo Seen, D., Loireau, M.,

Tillard, E., Todoroff, P., Tonneau, J.-P., and Vitry,

F. (2013). Simulation cartographique au service

de l’innovation collective. In SAGEO’2013, Brest,

France. 16 p.

Badariotti, D. and Weber, C. (2002). La mobilit

esidentielle en ville. mod

elisation par automates cel-

lulaires et syst

eme multi-agents

a bogota. Lespace

eographique, 31(2):97–108.

Batty, M., Axhausen, K. W., Giannotti, F., Pozdnoukhov,

A., Bazzani, A., Wachowicz, M., Ouzounis, G.,

and Portugali, Y. (2012). Smart cities of the fu-

ture. The European Physical Journal Special Topics,

214(1):481–518.

Bingham, C., Walsh, C., and Carroll, S. (2012). Impact of

driving characteristics on electric vehicle energy con-

sumption and range. Intelligent Transport Systems,

IET, 6(1):29–35.

Bustos-Turu, G., Van Dam, K., Acha, S., and Shah, N.

(2015). Integrated planning of distribution networks:

interactions between land use, transport and electric

vehicle charging demand. In 23rd International Con-

ference on Electricity Distribution, Lyon, France.

Bustos-Turu, G., van Dam, K. H., Acha, S., Markides,

C. N., and Shah, N. (2016). Simulating residential

electricity and heat demand in urban areas using an

agent-based modelling approach. In Energy Con-

ference (ENERGYCON), 2016 IEEE International,

pages 1–6. IEEE.

Bustos-Turu, G., van Dam, K. H., Acha, S., and Shah, N.

(2014). Estimating plug-in electric vehicle demand

ﬂexibility through an agent-based simulation model.

In Innovative Smart Grid Technologies Conference

Europe (ISGT-Europe), 2014 IEEE PES, pages 1–6.

IEEE.

Certick

y, M., Jakob, M., and P

ıbil, R. (2015). Analyzing

on-demand mobility services by agent-based simula-

tion. Journal of Ubiquitous Systems & Pervasive Net-

works, 6(1):17–26.

Choo, C. W. (1997). It2000: Singapore’s vision of an in-

telligent island. In Droege, P., editor, Intelligent Envi-

ronments, pages 49 – 65. North-Holland, Amsterdam.

Dijkstra, L., Garcilazo, E., and McCann, P. (2013). The

economic performance of european cities and city re-

gions: Myths and realities. European Planning Stud-

ies, 21(3):334–354.

Gifﬁnger, R. (2011). European smart cities: the need for

a place related understanding. Department of Spatial

Development, Infrastructure and Environmental Plan-

ning of Vienna university of Technology.

Gioda, A. (2015). El hierro (canaries): une

ıle et le choix

Steps Towards Simulating Smart Cities and Smart Islands with a Shared Generic Framework - A Case Study of London and Reunion Island

335

des transitions

energ

etique et

ecologique. VertigO-

la revue

electronique en sciences de l’environnement,

14(3).

Karabasoglu, O. and Michalek, J. (2013). Inﬂuence of driv-

ing patterns on life cycle cost and emissions of hybrid

and plug-in electric vehicle powertrains. Energy Pol-

icy, 60:445–461.

Longo, M., Roscia, M., and Lazaroiu, G. C. (2014). Inno-

vating multi-agent systems applied to smart city. Re-

search Journal of Applied Sciences, Engineering and

Technology, 7(20):4296–4302.

Magnan, A. (2009). Syst

emes insulaires, repr

esentations

pyramidales et soutenabilit

e: approche comparative

ean indien/petites antilles. Les Cahiers dOutre-Mer.

Revue de g

eographie de Bordeaux, 62(248):529–548.

Maia, R., Silva, M., Ara

ujo, R., and Nunes, U. (2011). Elec-

tric vehicle simulator for energy consumption studies

in electric mobility systems. In Integrated and Sus-

tainable Transportation System (FISTS), 2011 IEEE

Forum on, pages 227–232. IEEE.

McNally, M. G. and Rindt, C. R. (2007). The activity-based

approach. In Handbook of Transport Modelling, pages

55–73. Emerald.

Nations, U. (1994). United nations convention on the law

of the sea. Technical report.

North, M. J., Collier, N. T., Ozik, J., Tatara, E. R., Macal,

C. M., Bragen, M., and Sydelko, P. (2013). Com-

plex adaptive systems modeling with repast simphony.

Complex adaptive systems modeling, 1(1):3.

Roscia, M., Longo, M., and Lazaroiu, G. C. (2013). Smart

city by multi-agent systems. In Renewable Energy

Research and Applications (ICRERA), 2013 Interna-

tional Conference on, pages 371–376.

Sweda, T. and Klabjan, D. (2011). An agent-based deci-

sion support system for electric vehicle charging in-

frastructure deployment. In 2011 IEEE Vehicle Power

and Propulsion Conference, pages 1–5.

The Ecological Sequestration Trust (TEST) (2016). The

resilience.io platform. Accessed: 2017-01-23.

Unicef et al. (2012). The state of the world’s children 2012:

children in an urban world. Technical report, eSo-

cialSciences.

SMARTGREENS 2017 - 6th International Conference on Smart Cities and Green ICT Systems

336