Modeling Networks of Interdependent Infrastructure in Complex Urban

Environments Using Open-Data

Antonio Di Pietro

1 a

, Francesco Cavedon

, Vittorio Rosato

3 b

and George Stergiopoulos

4 c

Department of Energy Technologies and Renewable Sources, ENEA, Via Anguillarese 301, Rome, Italy

University of Verona, Department of Computer Science, Strada le Grazie 15 - 37134 Verona, Italy

Department of Engineering, University Campus Bio-Medico of Roma, Via Alvaro del Portillo, 21 - 00128 Roma, Italy

Department of Information and Communication Systems Engineering, University of Aegean, Greece

Keywords:

Cascading Failure, Open-Data, Graph Critical Infrastructure, Risk Analysis.

Abstract:

Dependency effects between Critical Infrastructure (CI) elements represent key information needed to predict

and analyze the impact of natural (or man-made) disturbances. The dependency links among CI elements and

their associated weight are data whose availability is often complex to determine and are usually not available.

Leveraging on several data supporting US and EU Directives for the Resilience and the Protection of CIs, the

objective of the present work is to deﬁne a dependency network of elements of critical sectors extracted from

available open-data. The resulting network is then studied in terms of its basic topological properties. The

analysis of the network provides interesting clues about the properties and locations of critical points that can

cause cascading failures. In addition, this information can form the basis for planning actions that mitigate the

risk of cascading effects.

1 INTRODUCTION

Resilience is a key, systemic, property which is of pri-

mary relevance in Smart Cities which are complex

environments characterized by the superposition of

infrastructure, structural assets, citizens, and host all

primary services which citizen must have constantly

available. All these components, however, are fully

dependent on each other, leading their management

and protection a complex task. Whereas repositories

of system dependencies already exist, mapping urban

dependencies is a challenge hindered by several fac-

tors including lack of integration of knowledge held

by different critical infrastructure (CI) operators and

privacy restrictions. Furthermore, data are constantly

changing and are difﬁcult to collect because they are

kept by different stakeholders.

One way to analyze CI dependencies is to con-

sider Dependency Risk Graphs (DRGs) (Stergiopou-

los et al., 2015) whose nodes represent CI compo-

nents and directed edges represent the potential risk

that the destination node may suffer due to its de-

https://orcid.org/0000-0003-4041-4913

https://orcid.org/0000-0003-4784-0069

https://orcid.org/0000-0002-5336-6765

pendency on the source node in the event of a source

node failure. Throughout this work, we will use the

acronym POI (Point of Interest) to identify a whatso-

ever element or installation of a speciﬁc technological

system belonging to a given network which is ”ex-

pression” of a speciﬁc Sector. Sectors, in turn, are

functional domains which produce goods and/or ser-

vices to citizens. Sectors are Government functions,

Energy production, Water distribution, Finance activ-

ities etc. Resulting DRG can be considered for the

analysis of possible cascading failures due to infras-

tructure dependency chains at an urban scale.

The MARIS (Modeling infrAstructuRe dependen-

cIes at an urban Scale) methodology (Di Pietro et al.,

2023), previously used for the prediction of depen-

dency cascading outages in CI, can also be imple-

mented by producing DRGs by using open-data (in

terms of POI) collected at an urban scale. This paper

extends the methodology by considering a broader set

of POI types and dependencies. It further propose a

strategy to unveal unknown dependencies in order to

produce DRGs that can be analyzed to identify possi-

ble mitigation actions to cascading failures. This can

be achieved by using the results under the form of at-

tachments of new Directives which are proposed, at

national or multi-national level, to deal with the task

Di Pietro, A., Cavedon, F., Rosato, V. and Stergiopoulos, G.

Modeling Networks of Interdependent Infrastructure in Complex Urban Environments Using Open-Data.

DOI: 10.5220/0012743800003708

Paper published under CC license (CC BY-NC-ND 4.0)

In Proceedings of the 9th International Conference on Complexity, Future Information Systems and Risk (COMPLEXIS 2024), pages 62-73

ISBN: 978-989-758-698-9; ISSN: 2184-5034

of protecting and enhancing resilience of critical as-

sets. They contains the understanding of the complex

infrastructure dependencies sedimented by the experi-

ence of infrastructure operators, stakeholders and the

scientiﬁc community than, since at least two decades,

have addressed the problem of technogical system’s

dependencies.

2 DEPENDENCY NETWORKS

AND THE MARIS

METHODOLOGY

There has been a great deal of effort to identify depen-

dency mechanisms among infrastructure, leading to

perturbation ﬂows from one infrastructure to the oth-

ers, which trigger the onset of the so called ”cascading

failures”. Dependency links represent the functional

interaction among infrastructure whose services (or

products) supply other infrastructure providing them

some necessary input (energy, water, personnel, data

etc.) for the infrastructure functioning. The loss

of those inputs reduces (or completely inhibits) their

functioning. However there is a graduality in the in-

frastructure dependencies. Some inputs are more rel-

evant than others. In some cases, for instance, prod-

ucts to be received could be appropriately stored and

used when missing (like e.g. energy buffer or wa-

ter reservoirs). Other inputs, when missing, start pro-

ducing an impact only after a certain time; others can

be rapidly supplied if appropriate redundancy strate-

gies have been prepared. Some of these dependen-

cies can also be bi-directional: the fading of one in-

frastructure’s functioning can provide a negative feed-

back to some feeding infrastructure, which will then

further reduce its capability of producing the needed

input to the former, thus producing a self-consistent,

multi-infrastructure negative feedback and an increas-

ing large functional crisis (inter-dependency effects).

This is to say that dependency dynamics is quite com-

plex and establishes along different time and geo-

graphical scales.

Related work in this ﬁeld touches both engineer-

ing perspectives for smart city development, network

dynamics as well as information modeling. From the

engineering perspective, multiple publications delve

on the subject of dependencies inside cities. Au-

thors in ((R., 2019) elaborate on dependencies of crit-

ical infrastructures by highlighting the ”double di-

mension” of efﬁciency and susceptibility of elements

against disruptions due to their dependencies. A re-

cent work introduces measures to identify critical net-

work components inﬂuencing the resilience of inter-

dependent infrastructure networks and their recovery

(Almoghathawi, 2019).

From a network dynamics perspective, authors

utilize models to analyze the interrelations between

infrastructures. From the early work of (Kotzaniko-

laou, 2013) on n-order cascading effects analysis that

laid the foundations on modeling cascading effect

Risk, to (Stergiopoulos, 2015) that extended previ-

ous work to incorporate time dimensions and cen-

trality metrics, to today’s publications, this work em-

phasizes the need for modeling of information to un-

derstand cascading effects. Other authors (such as

(Abdelgawad, 2019)) used MATLAB code to iden-

tify and quantify the ampliﬁcation of cascading ef-

fects due to interdependencies among CIs, aiming to

pinpointing weak points for better preparedness and

mitigation. In (Mbanaso, 2021), authors analyze the

Nth-order dependency effects between critical infras-

tructures using Neural Networks and create a model to

analyze the criticality and dependency of CI elements,

emphasizing on cascading effects. Similarly, authors

in (Zhou and Bashan, 2020) explore targeted attack

strategies based on dependencies within interdepen-

dent networks, ﬁnding that dependency-last attacks

are more effective than dependency-ﬁrst or random

attacks in triggering cascading failures, while (Aida

and Katsikas, 2021) analyze dependencies in large-

scale critical infrastructures, offering quantitative as-

sessment parameters for multi-order dependencies be-

tween cyber-physical systems.

Major breakthroughs in this domain have been

provided by a number of works (like e.g. (Franchina

et al., 2011) and (Laug

e et al., 2015)) which, by us-

ing different approaches, have attempted to produce

a qualitative, and in some cases quantitative, estimate

of sectorial dependencies, their relevance and the tim-

ing during which dependency effects take place.

The net result of these works has been the iden-

tiﬁcation of a fault tree indicating, starting from the

failure of an infrastructure belonging to a speciﬁc Sec-

tor, which other infrastructure of other Sectors would

be compromised (after which amount of time) and

for each new perturbed infrastructure, the number of

other infrastructure attained. Such ”dependency” ex-

ercise has been realized up to a small number of steps

in the hierarchical dependency tree and/or up to a cer-

tain time after the occurrence of the main fault event

in the cascade-initiating Sector.

After the identiﬁcation of the dependency tree for

each infrastructure Sector, it is possible to build up the

inter-sectorial network in a speciﬁc area, by appro-

priately linking elements of different Sectors among

them, by using the hierarchical dependency tree as a

guide to connect different nodes. Each link connect-

Modeling Networks of Interdependent Infrastructure in Complex Urban Environments Using Open-Data

ing different nodes is oriented: it goes from the feed-

ing node to the fed one. Links, in principle, should

also be weighted in order to identify, in a ”static” per-

spective, the relevance of a speciﬁc dependence for

the fate of the fed node; higher the weight, higher the

impact of the loss of the feeding node for the func-

tionalities of the fed one . Lightness, in a ”dynami-

cal” viewpoint, would correspond to a higher latency

(i.e. a longer time) for the functional impact of the hit

node to the others.

With weighted links, it is possible to ”simulate”

dynamic events in cascading failures, as a network

model of such type would allow to emulate the be-

havior of the propagation failure from one node to

the others. This is essentially the objectives of the

MARIS methodology. An example of such methodol-

ogy on a synthetic scenario has been given in (Rosato

et al., 2021a).

The main objective of the present work is to

propose the application of this methodology which,

starting from Public Open Data and the availability

of appropriate hierarchical dependency trees, is able

to construct realistic infrastructure networks which

could be useful for further analysis on both their

”static” (i.e. topological, functional) and ”dynamical”

(reproduction of fault scenarios) behavior.

The MARIS methodology leverages on three

building blocks: (i) the mapping of POI data into the

CI sectors (to attribute the dependency of POI to the

geographically proximal CI element); (ii) the map-

ping of CI sector dependencies into POI dependen-

cies (to establish the way in which perturbations ﬂow

from one CI to another, thus ultimately affection POI

functionality); and (iii) the creation of DRGs of POI

facilities.

A signiﬁcant contribution of MARIS methodol-

ogy resides on the model’s ability to easily incor-

porate Open Data sources, enhancing scalability and

adaptability to different urban contexts.

2.1 POI Facilities Classiﬁcation

In order to classify data, the ﬁrst problem is to allot

the different elements (POI) to a speciﬁc functional

class (Sectors). POIs will thus become nodes of a

network, where links will represent the unknown de-

pendency of the services exchanged by two different

nodes.

POI classiﬁcation can be dealt with using ”legal”

deﬁnition of Sectors, by using appropriate national

deﬁnition of elements belonging to different societal,

industrial and economic activities. Links will be then

deduced by ”dependency maps” which have been pro-

duced for the production of Directive to manage the

complex domain of Critical Infrastructure.

2.1.1 The CER and CISR Directives

A clue for identifying the Sectors in which speciﬁc

POI can be classiﬁed is provided by laws (or Direc-

tive) issued ad national or federal scale.

The Critical Entities Resilience (CER) Directive

(Council of European Union, 2022) 2557/2022 has

identiﬁed a resilience strategy for the EU Member

States (MS) with the aim of realizing a common and

homogeneous security space of all CI with respect to

natural and anthropic hazards. Such a strategy con-

sists of a number of prescriptions to MS which will

be called to set in place speciﬁc measures to ensure

the resilience of essential services in case of critical

situation (essential services, economic and industrial

activities). CER Directive follows a similar initiative

established in US, the Critical Infrastructure Security

and Resilience (CISR) Directive (Presidential Policy

Directive, 2013) which manages all Federal Depart-

ments and Agencies to identify, prioritize and provide

a plan to protect their physical and cyber critical in-

frastructure. Both Directives provide a list of essential

services (and their related ”entities”, such as both in-

frastructure and their managing organizations): they

consist in 11 (for EU) and 16 (for US) sectors, as

shown in Table 1.

The CISR Directive inspired the further delivery

of 16 Sector-Speciﬁc Plan (SSP) documents (within

the framework of the National Infrastructure Protec-

tion Plan (NIPP)) that provide a detailed description

of all CI sectors identiﬁed in the CISR directive as

well as of their dependencies (NIPP, 2013). Consid-

ering the high level of detail of all CI sectors and sub-

sectors and the dependency analysis of all CI sectors,

SSP documents were used as input to associate: (i)

CISR sectors to POI types and (ii) CISR sector de-

pendencies into POI dependencies.

2.1.2 POI Type Acquisition

The POI data used in this study were acquired from

OpenStreetMap (OSM) which is one of the most

widely used map services in the world, providing free

geospatial information of real world features, with

more than 10 million users. Based on the initial 29

primary types or features of POI provided by OSM,

the data used in the analysis were simpliﬁed to 20 dif-

ferent features to address a subset of the CISR sec-

tors and subsectors (Figure 1). Each POI data con-

tains attribute information including geographical en-

tity name, address, afﬁliation type, longitude, and lat-

itude, administrative region and were stored locally

into a geospatial database.

COMPLEXIS 2024 - 9th International Conference on Complexity, Future Information Systems and Risk

Table 1: Classiﬁcation of the type of Points of Interests ac-

quired for each CISR sector.

Sector (CISR) Sector (CER) POI types

Energy x 1

Transport x 4

Finance x 3

Healthcare x 9

Water x 7

Information Technology x -

Communications x 2

Government x 15

Defense Industrial x -

Food and Agriculture x 27

Nuclear - -

Commercial facilities - 42

Critical Manufacturing - -

Dams - -

Emergency services - 6

Chemical - -

Table 3 lists a detailed description of each POI

considered in the analysis.

post oﬃce,

public building,

townhall, prison,

diplomatic, courthouse,

telephone,

community centre,

nursing home,

social facility, university

school, kindergarten

college

Government

hospital, clinic, doctors,

dentist,

physiotherapist,

psychotherapist,

veterinary, laboratory,

pharmacy,

Healthcare

police, ﬁre station,

phone, siren,

ambulance station,

Emergency services

communications tower,

connection point,

Communications

Drinking water (water

tap, fountain,

ﬁre hydrant,

water tower, water well,

water works),

Waste water

(wastewater),

Water

Retail (ﬂorist, mall,

department store,

general, clothes,

fashion accessories,

shoes, bicycle rental),

Sports (sports centre,

pitch, swimming pool,

tennis),

Accommodation (

hotel, motel,

guest house, hostel

apartment, chalet,

caravan site),

nightclub, park,

Public assembly

(playground, arts

centre, cinema,

refugee site, museum,

library, shelter,

stadium, theatre),

Outdoor events

(water park, running

fairground),

Real Estate (property

management,

apartments, detached

house, residential,

place of worship,

semidetached house)

Gaming (casino,

gambling),

Commercial facilities

Distribution

(restaurant, fast food,

cafe, pub, bar,

biergarten,

food court,

supermarket, bakery,

kiosk, greengrocer,

agrarian),

Storage (barn, silo),

Processing,

Packaging and

Production (

cowshed, distillery,

brewery, winery,

stable),

Supply (windmill,

greenhouse, farm,

oil mill, allotments,

farmyard, farmland)

Food and Agriculture

substation,

Energy

stop position, port,

motorway, airport,

Transportation

payment terminal,

bank, atm

Financial services

Figure 1: List of POI grouped by CISR sector.

2.2 POI Dependency Setup

Based on the sector dependencies deﬁned in the SSP

documents (where the type of dependency is ex-

pressed as physical, logical or cyber (Marashi et al.,

2021), a mapping of the CISR sector dependencies in

terms of POIs was implemented. The resulting POI

dependency mapping is shown in Figure 2 in terms of

consumer POIs i.e. elements that are dependent on

other elements to work and producer POIs i.e. ele-

ments that provide resources or services to other ele-

ments.

2.2.1 Assumptions

In order to focus the analysis on cascading failures

i.e., disruptions occurring in one CI which causes

the failure of a component in one or more CIs (CI-

Pedia(c), (n.d.)., 2024), intra-dependencies, i.e. the

relations among subsystems of the same sector, were

not considered. For example, considering the Trans-

port sector, the dependencies existing among the POI

facilities airport and motorway were not considered

although an airport may require a motorway to be

connected with the nearby territory and to be fed by

the needed resources (personnel, supply of different

goods etc.).

Regarding the number of dependencies selected

for the POIs of each sector, where possible, the same

dependencies for multiple POIs were grouped to-

gether. In addition, considering that the ultimate goal

of this approach is to create an urban level of de-

pendencies, only direct (i.e. ﬁrst-order) dependencies

were considered.

In the following, we analyse the dependency con-

nections which are established among different ele-

ments of different sectors which represent the territo-

rial dependency map of CIs. In particular, considering

that most POIs depend on the supply of energy, water,

communications, transportation, emergency, health-

care and ﬁnancial services (Figure 2), only dependen-

cies not included in those mentioned above will be

discussed.

The proposed method makes use of a multi-step

approach for data veriﬁcation and transposition.

1. Initially, a data quality assessment evaluates the

accuracy, timeliness, and relevance of data col-

lected from various open sources using a 70-30

sampling approach per type of POI, including ge-

ographical information systems (GIS) and real-

time infrastructure performance metrics.

2. We then apply data cleaning and preprocessing

techniques to mitigate the impact of incomplete

or erroneous data entries.

Since the use of data is not driven by any classiﬁer

but rather as input for modeling a wider urban area

and the reliability analysis of dependencies that arise

in the infrastructure, the notion of a biased dataset is

not as relevant. We bridged data gaps by determining

the type of data missing and reconstructing the en-

tries from other, identical POIs that include any miss-

ing data. This way the methodology ensures a high

level of consistency and robustness in modeling the

surrounding areas and allowing for optimal analysis

of cascading effects.

Modeling Networks of Interdependent Infrastructure in Complex Urban Environments Using Open-Data

2.2.2 Energy POIs

A Substation represents the main facility of a High

or Medium Voltage electrical network that provides

electricity to an urban environment. It depends on

communication tower and connection point since

these components provide telecommunications ser-

vices required by SCADA and telecontrol operations.

2.2.3 Transportation POIs

A Stop position represents a public transport stop. It

receives energy from a substation to support critical

facility functions, such as lighting and it also relies on

a communication tower, to enable Internet connec-

tion to customers.

A Motorway is a main road that requires energy

from a substation to ensure lighting. Both Motor-

way, Port and airport facilities also rely on ambu-

lance station, police, phone, ﬁre station, hospital

and communication tower to ensure emergency re-

sponse.

2.2.4 Financial Service POIs

A Payment terminal and atm are associated to

bank facilities that require the supply of energy,

telecommunication, medical assistance, transporta-

tion, wastewater and emergency services to enable or-

dinary and emergency operations.

2.2.5 Healthcare POIs

Hospital and clinic facilites require the supply of en-

ergy, telecommunication, water and wastewater to op-

erate. In addition, the dependency on the Transporta-

tion sector ensures the efﬁcient shipment of supplies,

without which the sector cannot provide healthcare

services. The availability of telecommunication is vi-

tal to enable routine operations, maintain situational

awareness and coordinate healthcare activities during

steady state and emergency response. Besides, the de-

pendency on Emergency Services provide life safety

and security.

2.2.6 Water POIs

A ﬁre hydrant represents an active ﬁre protection

measure and a source of water provided whereas a

water well is a structural facility to access ground

water, created by digging or drilling. Both depend on

electricity for the functioning of pumps, wells and for

treatment operations and on Medical and ﬁreﬁghting

responders in emergency scenarios.

A water works represents a facility where wa-

ter is treated to make it suitable for human consump-

tion and whereas a wastewater indicates a clariﬁer or

settling basin of a wastewater treatment plant. Both

depend on electicity and Emergency services to res-

cue people and coordinate with public health agen-

cies during emergency and other water quality-related

events.

2.2.7 Communications POIs

A communications tower is a structure in which a

Base Transceiver Station (BTS) and other devices are

mounted that enable mobile and Internet connectiv-

ity. Communication devices rely on energy to power

its backup generators and the Transportation Sector to

deliver those fuels. Water sources are also necessary

for cooling and other processes.

2.2.8 Government POIs

Government POIs include ofﬁces and building com-

plexes (post ofﬁces, public buildings), correctional

facilities (prison), embassies, courthouses, education

facilities (university, school, kindergarten, college).

2.2.9 Food and Agriculture POIs

Food and Agriculture POIs complex production, pro-

cessing, and delivery systems and has the capacity to

feed people and animals. Such facilities are almost

entirely under private ownership and operate in highly

competitive global markets.

2.2.10 Commercial Facility POIs

Commercial facilities include an extremely diverse

range of sites and assets where large numbers of peo-

ple congregate daily to conduct business, purchase re-

tail products, and enjoy recreational events and ac-

commodations.

2.2.11 Emergency Services POIs

Emergency Services facilities include human (ﬁre,

police and ambulance stations) and equipment for

daily operations (phones, sirens).

2.3 Creating Dependency Risk Graphs

As above stated, whereas for CI it is possible to get

an accurate dependency map (pointing functional de-

pendency and dependency links), other POI data are

not accurately speciﬁed in terms of their functional

dependencies and dependency links. For instance, it

might be undeﬁned the dependency of a POI (i.e. an

COMPLEXIS 2024 - 9th International Conference on Complexity, Future Information Systems and Risk

public building,

townhall, prison, social

facility, community

centre,

Government

(POI consumer)

substation, water tower, wastewater, police,

ﬁre station, connection point, stop position,

clinic, atm, general, motorway

Other sectors

(POI producer)

post oﬃce,

substation, water tower, wastewater, police,

connection point, stop position, hospital,

atm, bank, general,

grave yard,

substation, water tower, wastewater, stop

position, ﬂorist

school, kindergarten,

college, university,

substation, water tower, wastewater, ﬁre

station, connection point, stop position,

hospital, atm, library

nursing home,

substation, water tower, wastewater,

ambulance station, ﬁre station,

connection point, stop position, clinic, atm,

general,

telephone, connection point, substation

diplomatic, courthouse,

substation, water tower, wastewater, police,

connection point, stop position, bank,

general, airport

doctors, dentist,

physiotherapist,

psychotherapist,

veterinary, laboratory,

pharmacy,

Healthcare

(POI consumer)

substation, stop position, water tower,

wastewater, ﬁre station, connection point,

motor way,

Other sectors

(POI producer)

hospital, clinic

substation, stop position, motorway, water

tower, wastewater, ambulance station, ﬁre

station, police, communication tower,

connection point,

police, ﬁre station,

phone,

ambulance station,

Emergency services

(POI consumer)

substation, stop position, motorway,

connection point, water tower, wastewater,

hospital,

Other sectors

(POI producer)

siren substation, communication tower,

communication tower,

Communications

(POI consumer)

substation, stop position, motorway, water

tower,

Other sectors

(POI producer)

connection point, substation, stop position, water tower,

water tap, fountain

Water

(POI consumer)

substation, ambulance station,

Other sectors

(POI producer)

ﬁre hydrant, water well,

substation, ambulance station, ﬁre station,

connection point, bank, laboratory,

motorway,

retail*, accommodation*,

sports*, public

assembly*, gaming*,

Commercial facilities

(POI consumer)

substation, water tower, wastewater, ﬁre

station, police, connection point, stop

position, motorway, hospital, bank, atm

Other sectors

(POI producer)

playground, park,

substation, water tower, wastewater, ﬁre

station, police, stop position, motorway,

hospital,

distribution*,

processing &

packaging &

production*

Food and Agriculture

(POI consumer)

substation, water tower, wastewater, ﬁre

station, police, connection point, stop

position, motorway, hospital, bank, atm

Other sectors

(POI producer)

barn, silo,

substation, water tower,

supply*,

substation, water tower, wastewater,

motorway, airport

substation,

Energy

(POI consumer)

connection point, communication tower,

Other sectors

(POI producer)

port, airport,

Transportation

(POI consumer)

substation, water tower, wastewater, police,

ﬁre station, connection point, hospital,

bank, atm,

Other sectors

(POI producer)

stop position,

substation, communication tower,

motorway,

substation, wastewater, ﬁre station,

connection point, police, hospital,

communication tower, ambulance station,

phone,

payment terminal,

bank, atm,

Financial services

(POI consumer)

substation, wastewater, connection point,

police, hospital, communication tower,

Other sectors

(POI producer)

Figure 2: POI dependencies grouped by CISR Sector.

hospital) with telecommunication and electrical net-

works elements. Thus, we should reconstruct the de-

pendency map by using other criteria (for instance, a

proximity criteria, that is we decide that the functional

elements of other CI which are responsible to provide

speciﬁc functionality to the Hospital are the geogra-

phycally clostest ones). This section will deal with

the identiﬁcation of the different strategies adopted

to identify dependencies whereas not explicitly pro-

vided. The created DRG can be used as a basis for

the use of risk analysis models that allow assessing

the criticality of speciﬁc dependency chains that can

exacerbate the occurrence of cascade effects ((Rosato

et al., 2021b)).

3 CASE STUDY: THE MARCHE

REGION

As a case study, we created a DRG by using open-

data of a speciﬁc area of interest and applying the

methodology described in the previous paragraphs. In

particular, we considered the Marche Region, located

in central Italy, with near 1.5 million inhabitants in

an area of 9.350 km

. The resulting regional scenario

contains N = 27.324 nodes belonging to 10 CISR sec-

tors (as shown in Table 2) where the dependencies

among nodes were established by using a proximity

strategy. The resulting network of total set of N nodes

contains M = 279.663 oriented links. In particular, a

generic nodes will result to have incoming links (com-

ing from producer nodes) and outgoing links (going

to consumer nodes). At this stage of elaboration, the

Modeling Networks of Interdependent Infrastructure in Complex Urban Environments Using Open-Data

Table 2: Major topological properties of POIs for the differ-

ent CISR sectors: d

is the node degree of incoming links

(constant), d

out−MAX

is the maximum degree of outgoing

links.

Sector POI instances d

out −MAX

Energy 281 2 1506

Transp. 426 5,4 1.748

Finance 461 6 2.033

Health. 599 6,8 3.640

Water 493 4,2 3.130

Commun. 9 3,1 12.895

Governm. 840 - -

Food&A. 5.497 - -

Commerc. 18.594 0,04 118

Emergen. 201 7 4.642

weight of each link is unitary everywhere. However,

this value could be modulated based on the relevance

of the connection, or even the latency of its depen-

dency. In fact, a speciﬁc link may represent a func-

tional dependency that occurs instantaneously, or that

requires a longer latency.

A ﬁrst analysis can be provided by studying the

basic properties of the network: the degree distribu-

tion of the nodes and the main centrality properties of

the nodes. Table 3 shows that, on average, the number

of incoming links is smaller than the number of out-

going ones. In other words, a node, on average, is the

provider of a large number of nodes in other sectors

that need to be powered by its service. Note that the

nodes in the Government and Food and Agriculture

sectors are only consumer nodes, since they have no

outgoing connections.

Outgoing links count the number of elements of

other sectors which are supplied by a speciﬁc node, in

order to have an appropriate functionality. As stated

previously, this counting could be done more appro-

priately by weighting each link with a speciﬁc weight

that takes into account the relevance of that speciﬁc

link to ensure the functionality of the provisioned

node. Although this may favor a more accurate evalu-

ation of the dependency mechanisms, a simple count-

ing with unit weight of each link (as in our case) al-

ready provides a clear identiﬁcation of the nodes that

are critical in guaranteeing the functionality of all the

others.

Figure 3 shows the distribution of outgoing links

for the nodes belonging to 4 of the most important

sectors. Energy nodes consist of electric substations

and exhibit a large spread of outgoing degree. Similar

results concern the Emergency, Finance and Transport

sectors where ﬁre stations, atm and public transport

stops respectively are source nodes for many sectors,

especially in urban areas. In particular, the POI in-

stances that exhibit the higher outbound degree are:

a connection point, a hospital, a water tower, a mo-

torway junction, a department store, a ﬁre station, an

ATM and an electric substation.

According to Table 3, there are sector nodes that

are providers of a large number of other nodes; some

of them are terminal nodes while others are, in turn,

providers of other nodes in other sectors. There-

fore, nodes with large outbound link degrees concern

highly relevant nodes that have strategic relevance,

since they are central to a large number of subnet-

works. A number of highly strategic nodes i.e. nodes

with high inbound degree centrality in a DRG, are nat-

ural “sinkhole” points of incoming dependency risk.

Such nodes appear to be suitable nodes to consider

when prioritizing mitigation controls so that multi-

ple cumulative dependency risk chains can be reduced

simultaneously. In Figure 5 we represent the net-

work resulting from the connection, according to the

adopted methodology, of all POIs of all sectors in Re-

gione Marche. Nodes with different colors represent

POI instances of different sectors; their sizes, further-

more, are proportional to their total degree (sum of

incoming and outgoing links).

4 CONCLUSIONS

The US CISR and the EU CER Directives have iden-

tiﬁed a large number of ”critical entities” grouped in

different sectors. As a further relevant information,

the CISR Directive has also inspired the further de-

livery of 16 Sector-Speciﬁc Plan (SSP) documents

that identify a dependency map enabling to link ele-

ments of the different different sectors on the bases

of their mutual functionalities as producer or con-

sumer nodes. The combination of open-data consist-

ing of the Points of Interest (POI) and the dependency

map proposed by the SSP documents has allowed

us to build a large regional network connecting ele-

ments of different sectors with oriented links, describ-

ing the functional dependencies among the different

nodes. With the described prescriptions, we have

analysed real data with infrastructure elements of Re-

gione Marche taken from publicly available reposito-

ries based on OpenStreetMap.

In a ﬁrst order analysis, the method allows the

identiﬁcation of topological properties of the net-

work and the localization of elements which, accord-

ing to their degree (outgoing and incoming degree),

should be considered as strategic. Aside to that, the

MARIS methodology, if implemented with the use of

weighted links among the nodes, opens up the pos-

sibility of performing network partitioning by recog-

nizing subnetworks slightly coupled with the rest of

the networks (i.e. whose all boundary links, if any,

COMPLEXIS 2024 - 9th International Conference on Complexity, Future Information Systems and Risk

Figure 3: Distribution of the outgoing node’s degrees in several sectors (abscissa, in log scale, represents the node degree; in

the ordinate the number of nodes with the given outgoing degree. Up-left: nodes of the Energy sector. Up-right: node of the

Emergency sector. Bottom-left: nodes of the Financial sector. Bottom-right: nodes of the Transportation sector. Lines are

drawn as a guide for the eye.

Figure 4: A GIS map of the area under analysis (Regione Marche, Italy) with the different sectors nodes highlighted in

different colors, according to the CISR sector to which they belong.

Modeling Networks of Interdependent Infrastructure in Complex Urban Environments Using Open-Data

Figure 5: A representation of the graph of the network under analysis (Marche region) where node’s size is proportional to

the out-degree value.

are weak) and by performing, on those networks, dy-

namical fault simulations ((Rosato et al., 2021a)) in

order to reproduce the cascading failures which issue

upon the fault of one (or more) nodes of the networks.

This approach, if appropriately exploited by us-

ing the largest and most realistic network of local

or regional elements, can be of invaluable support in

approaching the realization of ”educated” emergency

plans aiming at reducing the consequences od cascad-

ing failures.

ACKNOWLEDGEMENTS

This research has been partly supported by project

DRIVERS (”Combined data-driven and experience-

driven approach in the analysis of the systemic risk”)

funded by INAIL for a Contract provided to one of us

(VR).

REFERENCES

Abdelgawad, A., S. M. . G. J. (2019). Analyzing cascading

effects in interdependent critical infrastructures. In

Information Technology in Disaster Risk Reduction:

Third IFIP TC 5 DCITDRR International Conference,

ITDRR 2018, Poznan, Poland, 2018, Springer Inter-

national Publishing, pages 50–65.

Aida, A. and Katsikas, S. (2021). Identifying and analyzing

dependencies in and among complex cyber physical

systems. Sensors, 21(5):1685.

Almoghathawi, Y., . B. K. (2019). Component impor-

tance measures for interdependent infrastructure net-

work resilience. Computers and Industrial Engineer-

ing, 133:153–164.

CIPedia(c), (n.d.). (2024).

http://www.cipedia.eu [Accessed: February 1, 2024].

Council of European Union (2022). Directive (eu)

2022/2557 of the european parliament and of the

council of 14 december 2022 on the resilience

of critical entities and repealing council directive

2008/114/ec.

Di Pietro, A., Calabrese, A., De Nicola, A., Ferneti, D.,

Franchina, L., Mart

ı, J., and Ruocco, T. (2023). An

open-data-based methodology for the creation of a

graph of critical infrastructure dependencies at an ur-

ban scale. In Critical Infrastructure - Modern Ap-

proach and New Developments. IntechOpen, Rijeka.

Franchina, L., Carbonelli, M., Gratta, L., Crisci, M., and

Perucchini, D. (2011). An impact-based approach for

the analysis of cascading effects in critical infrastruc-

tures. International Journal of Critical Infrastruc-

tures, 7(1):73–90.

Kotzanikolaou, P., T. M. . G. D. (2013). Assessing n-order

dependencies between critical infrastructures. In-

ternational Journal of Critical Infrastructures, 6:93–

110.

Laug

e, A., Hernantes, J., and Sarriegi, J. M. (2015). Critical

infrastructure dependencies:a holistic, dynamic and

quantitative approach. International Journal of Criti-

cal Infrastructure Protection, 8(January 2015):16–23.

Marashi, K., Sarvestani, S. S., and Hurson, A. R. (2021).

Identiﬁcation of interdependencies and prediction of

fault propagation for cyber–physical systems. Relia-

bility Engineering & System Safety, 215:107787.

Mbanaso, U. M., . M. J. A. (2021). Conceptual modelling

of criticality of critical infrastructure nth order depen-

dency effect using neural networks. In 2020 IEEE

2nd International Conference on Cyberspac (CYBER

NIGERIA) IEEE Publishing, pages 127–131.

NIPP (2013). National Infrastructure Protection Plan.

https://www.cisa.gov/resources-tools/resources/

2013-national-infrastructure-protection-plan.

COMPLEXIS 2024 - 9th International Conference on Complexity, Future Information Systems and Risk

Presidential Policy Directive (2013). Presidential policy

directive 21 on critical infrastructure security and re-

silience (cisr/ppd-21).

R., W. (2019). Dependencies of elements recognized as

critical infrastructure of the state. Transportation Re-

search Procedia, 40:1625–1632.

Rosato, V., Di Pietro, A., Kotzanikolaou, P., Stergiopoulos,

G., and Smedile, G. (2021a). Integrating resilience in

time-based dependency analysis: A large-scale case

study for urban critical infrastructures. In Issues on

Risk Analysis for Critical Infrastructure Protection.

Rosato, V., Di Pietro, A., Kotzanikolaou, P., Stergiopoulos,

G., and Smedile, G. (2021b). Integrating resilience in

time-based dependency analysis: A large-scale case

study for urban critical infrastructures. In Rosato, V.

and Pietro, A. D., editors, Issues on Risk Analysis for

Critical Infrastructure Protection. IntechOpen.

Stergiopoulos, G., Kotzanikolaou, P., Theocharidou, M.,

and Gritzalis, D. (2015). Risk mitigation strategies for

critical infrastructures based on graph centrality anal-

ysis. International Journal of Critical Infrastructure

Protection, 10:34–44.

Stergiopoulos, G., K. P. T. M. . G. D. (2015). Risk mit-

igation strategies for critical infrastructures based on

graph centrality analysis. International Journal of

Critical Infrastructure Protection, 10:33–44.

Zhou, D. and Bashan, A. (2020). Dependency-based tar-

geted attacks in interdependent networks. Phys. Rev.

E, 102:022301.

Modeling Networks of Interdependent Infrastructure in Complex Urban Environments Using Open-Data

APPENDIX

Table 3: POI description. Source: (OpenStreetMap, 2024).

Sector POI type Description

Government

Post ofﬁce Post ofﬁce building with postal services.

Public building An ofﬁce of a (supra)national, regional or local government entity.

Townhall Building where the administration of a village, town or city.

Prison A prison or jail where people are incarcerated.

Diplomatic An embassy, diplomatic mission, consulate or liaison ofﬁce.

Courthouse A building home to a court of law, where justice is dispensed.

Telephone Public telephone.

Community centre A place mostly used for local events, festivities and group activities.

Nursing home A structure for disabled or elderly persons who need permanent care.

Social facility A facility that provides social services e.g. homeless shelters, etc.

Grave yard A (smaller) place of burial, often you’ll ﬁnd a church nearby.

University An university campus: an institute of higher education.

School An university campus: an institute of higher education.

Nursery school School and grounds - primary, middle and seconday schools.

College Campus or buildings of an institute of Further Education.

Healthcare

Hospital A hospital providing in-patient medical treatment

Clinic A medium-sized medical facility or health centre.

Doctors A doctor’s practice / surgery.

Dentist A dentist practice / surgery.

Physiotherapist Someone who practices physical therapy but is not a Physician.

Psychotherapist Someone who practices psychotherapy but is not a Physician.

Veterinary ofﬁce A place where a veterinary surgeon practices.

Medical laboratory Is a place that analyses body ﬂuids such as blood, urine, faeces etc.

Pharmacy A shop where a pharmacist sells medications.

Emergency Services

Police A police station where police ofﬁcers patrol from.

Fire station A station of a ﬁre brigade.

Phone Emergency telephone.

Siren A loud noise maker, such as an air raid siren or a tornado siren.

Ambulance station Structure set aside for storage of ambulance vehicles, equipment, etc.

Communication

Comm. tower A huge tower for transmitting radio applications.

Internet conn. point Last point of telecom local loops allowing connections to households

Water

Water tap Publicly usable water tap.

Fountain A fountain for cultural / decorational / recreational purposes.

Fire hydrant An active ﬁre protection measure.

Water tower Structure with a water tank at a high altitude to increase pressure.

Water well A structural facility to access ground water.

Water works A facility where water is treated to make it suitable for human use.

Wastewater plant A wastewater plant is a facility used to treat wastewater.

Commercial

facilities

Mall A shopping mall – multiple stores under one roof.

Department store A store – often multiple storeys high – selling a large variety of goods.

Generic items shop A store that carries a general line of merchandise.

Clothes shop Shop focused on selling clothes and/or underwear.

Fashion accessories Shop focused on selling fashion accessories.

Shoes shop Shop focused on selling shoes

Bicycle rental Shop where to rent a bicycle

Florist Shop focused on selling bouquets of ﬂowers

Sports centre A building that was built as a sports centre.

Pitch An area designed for practising a particular sport.

Swimming pool A swimming pool.

Tennis A tennis court.

Hotel A building with separate rooms available for overnight accommodation.

Motel Short term accommodation, particularly for people travelling by car.

Guest house Accommodation smaller than a hotel and typically owner-operated.

Hostel Cheap accommodation with shared bedrooms.

Bed and breakfast A ﬂat with cooking and bathroom facilities that can be rented.

Chalet A holiday cottage with self-contained cooking and bathroom facilities.

Continued on next page

COMPLEXIS 2024 - 9th International Conference on Complexity, Future Information Systems and Risk

Table 3: continued from previous page.

Sector POI type Description

Caravan site A place where you can stay in a caravan overnight or for longer periods.

Nightclub A place to drink and dance (nightclub).

Park A park, usually in an urban setting, created for recreation and relaxation.

Playground An area designed for children to play.

Arts centre A venue where a variety of arts are performed or conducted

Cinema A place where ﬁlms are shown

Refugee site A human settlement sheltering refugees or internally displaced persons

Museum A building which was designed as a museum.

Library A public library to borrow books from.

Shelter A small shelter against bad weather conditions.

Stadium A building constructed to be a stadium building.

Theatre A place where live performances occur, (e.g., musicals, concerts).

Water park An amusement park with features like water slides.

Fairground A site where a fair takes place.

Running area An area for running (jogging) routes.

Property manag. Ofﬁce of a property rental company for commercial or residential use.

Apartment A building arranged into individual dwellings, often on separate ﬂoors.

Detached house A free-standing residential building usually housing a single family.

House A dwelling unit inhabited by a single household.

Residential building A general tag for a building used primarily for residential purposes.

Place of worship A church, mosque, or temple, etc.

Semidetached house A residential house that shares a common wall with another on one side.

Casino A gambling venue with at least one table game that takes bets

Gambling A place for gambling.

Food and

Agriculture

Restaurant A normal restaurant that sells full sit-down meals.

Fast food A place for very fast counter-only service and take-away food.

Cafe An informal place that offers casual meals and beverages (coffee or tea).

Pub A place selling beer and other alcoholic drinks

Bar A place that sells alcoholic drinks to be consumed on the premises.

Biergarten An open-air area where alcoholic beverages along with food is served.

Food court An area with several restaurant food counters and a shared eating area.

Supermarket a large store with groceries and other items.

Bakery Shop focused on selling bread.

Kiosk A small shop sells magazines, tobacco, newspapers, etc.

Greengrocer Shop focused on selling vegetables and fruits.

Agrarian Shop that sells agrarian products, e.g. seeds, agricultural machinery.

Barn An agricultural building that can be used for storage.

Silo A storage container for bulk material, often grains e.g. corn, wheat.

Cowshed A cow barn is a building for housing cows, usually found on farms.

Distillery An establishment for distilling, especially alcoholic liquors.

Brewery A dedicated building for the making of beer.

Winery A place where wine is produced.

Stable for horses A building constructed as a stable for horses.

Windmill A traditional windmill, historically used to mill grain with wind power.

Greenhouse A greenhouse is a glass or plastic covered building used to grow plants.

Farm A residential building on a farm (farmhouse).

Oil mill A mill to crush or bruise oil-bearing seeds, e.g. linseed or peanuts.

Allotments A piece of land given for growing vegetables and ﬂowers.

Farmyard An area of land with farm buildings e.g. farmhouse, dwellings.

Farmland An area used for tillage (cereals, vegetables, oil plants, ﬂowers).

Energy Substation A power facility with transformers, switchgear or compensators.

Transportation

Public transport stop The position on the street or rails where a public transport vehicle stops.

Port Acoastal industrial area where commercial trafﬁc is handled.

Motorway junction A type of road junction linking highway facility to another.

Aerodrome An aerodrome, airport or airﬁeld.

Financial Services

Payment terminal Self-service payment kiosk/terminal.

o Bank A ﬁnancial establishment to make ﬁnanxial operations.

ATM Cash point

Modeling Networks of Interdependent Infrastructure in Complex Urban Environments Using Open-Data