BRAIN-IoT Architecture and Platform for Building IoT Systems

Salim Chehida

1 a

, Saddek Bensalem

1 b

, Davide Conzon

2 c

, Enrico Ferrera

2 d

and Xu Tao

CNRS, VERIMAG, University of Grenoble Alpes, Grenoble, France

IoT and Robotics Area, LINKS Foundation, Turin, Italy

Computer Science Department, University of Kentucky, U.S.A.

Keywords:

IoT, Architecture, Model-based Design, Interoperability, Distributed Execution, Security and Privacy,

Monitoring, Resiliency, Cloud, Edge, Decentralized IoT Applications.

Abstract:

The integration of Internet of Things (IoT) for building complex and critical systems requires powerful plat-

forms enabling to deal with multiple issues, including modeling, monitoring, control, maintaining and man-

agement of IoT applications. In this work, the authors propose a new platform based on layered architecture

that integrates a set of assets for model-based development of IoT systems. This platform named BRAIN-IoT

aims to meet the new challenges of IoT applications and to reduce the effort for building and managing these

applications. It consists of three frameworks that allow building decentralized IoT applications with comput-

ing capacity at the edge in a computing continuum with the cloud. The modeling and validation framework

is used to design, develop, and validate IoT applications logic. The distributed execution framework provides

an autonomic distributed infrastructure for the dynamic deployment and execution of IoT services on a mixed

cloud-edge environment. The security framework enables access control, end-to-end security and privacy of

data collected using IoT devices. The BRAIN-IoT platform is mapped to a well-established IoT reference

architecture and experimented on two industrial use cases.

1 INTRODUCTION

In recent years, IoT has demonstrated to be able to of-

fer signiﬁcant improvements to various domains, e.g.,

health, energy, hydraulics, and transport. Several ar-

chitectures and platforms (Adolphs and Epple, 2015;

Lin et al., 2019; R

omer et al., 2020; Bauer et al., 2013;

IEEE, 2020) have been proposed for building IoT sys-

tems. However, the complexity and the dynamic of

these systems highlights the need for new solutions

that allow promoting automation, managing complex-

ity and increasing trust. To cope with these demand-

ing requirements, a multitude of novel technologies

such as Edge Computing, Artiﬁcial Intelligence (AI),

and Analytics, as well as Security, Privacy and Trust

schemes are being investigated to be adopted in cur-

rent IoT architectures standards.

The 3D IoT architecture (Vermesan and Bacquet,

2018) speciﬁed by Alliance for the Internet of Things

Innovation (AIOTI) (AIOTI, 2020) is high-level ar-

https://orcid.org/0000-0002-5070-2591

https://orcid.org/0000-0002-5753-2126

https://orcid.org/0000-0002-2962-8702

https://orcid.org/0000-0002-4671-3861

chitecture proposed for supporting the novel require-

ments that generic IoT applications have. It aims to

be one pioneer of the next-generation IoT paradigm,

establishing a novel generic reference for different

IoT/Industrial Internet of Things (IIoT) applications

from different domains. In this work, the authors

present a new platform founded on a layered concrete

architecture based on the 3D reference model. The

BRAIN-IoT platform integrates novel technologies to

address the following challenges that arise in recent

IoT applications:

C1) Facilitate IoT systems modeling through multi-

ple abstractions and enabling the integration of

Validation and Veriﬁcation (V&V) techniques

and tools.

C2) Enable designing IoT applications involving sev-

eral heterogeneous platforms and Smart Things

interconnected to each other in a distributed en-

vironment.

C3) Support the analysis and exploitation of large

amount of raw data to extract and predict infor-

mation supporting system automaticity and au-

tonomicity.

Chehida, S., Bensalem, S., Conzon, D., Ferrera, E. and Tao, X.

BRAIN-IoT Architecture and Platform for Building IoT Systems.

DOI: 10.5220/0011086000003194

In Proceedings of the 7th International Conference on Internet of Things, Big Data and Security (IoTBDS 2022), pages 67-77

ISBN: 978-989-758-564-7; ISSN: 2184-4976

C4) Ensure the deployment of services in a dis-

tributed environment, dynamic resources alloca-

tion in response to environmental changes, and

autonomous dependency management to reduce

the system operational complexity.

C5) Ensure the security, privacy, and resiliency in

resource-constraint and distributed IoT environ-

ments.

In this work, the authors validate the BRAIN-IoT

platform through two use cases for warehouse logis-

tics and for water distribution management. The solu-

tion is evaluated using a standard well-known usabil-

ity questionnaire, the User Experience Questionnaire

(UEQ)

, which allows the end-users to evaluate their

feeling with the product in question.

Section 2 presents the existing IoT architectures

and platforms. Section 3 and 4 show the BRAIN-IoT

Architecture and its integrated components. In Sec-

tion 5, the authors map BRAIN-IoT components with

the 3D IoT reference architecture. Section 6 presents

the applications leveraged to validate the architecture.

Section 7 brieﬂy explains the validation results. Fi-

nally, the authors draw conclusions in Section 8.

2 RELATED WORK

Nowadays, it is still not available an unique IoT refer-

ence architecture. Several different organizations and

consortia have tried to deﬁne it, but the scenario is still

fragmented. This section presents the available IoT

reference architectures, describing their main charac-

teristics.

The Reference Architecture Model Industrie 4.0

(RAMI 4.0) (Adolphs and Epple, 2015) is a refer-

ence architecture focused on the smart industry do-

main. The effort has been started in Germany, but

today is driven by several companies and foundations

in relevant industry sectors, i.e., the ones associated

in the Platform Industrie 4.0

. RAMI 4.0 is a Service

Oriented Architecture (SOA), which combines all el-

ements and Information technology (IT) components

in a layered and life cycle model. RAMI 4.0 breaks

down complex processes into easy-to-grasp packages,

including data privacy and IT security. Since the ar-

chitecture is focused on the industry domain, it does

not cover the requirements of generic IoT applica-

tions.

The Industrial Internet Reference Architecture

(IIRA) (Lin et al., 2019) is a common architecture

https://www.ueq-online.org/

https://www.plattform-i40.de/IP/Navigation/DE/

Home/home.html

framework deﬁned by the Industry Internet of Things

Consortium (IIC), to develop interoperable IIoT sys-

tems for applications across a broad spectrum of in-

dustrial verticals, both public and private, to achieve

the IIoT goals. Also in this case, the architecture is in-

teresting, but it is focused only on industrial domain,

so it is not suitable to be applied in a more generic IoT

scenarios.

The Eclipse Arroehead (R

omer et al., 2020) is a

SOA with a reference implementation for IoT interop-

erability that was originally developed as part of the

Arrowhead Tools European research project

, aligned

with the concept of RAMI 4.0 and IIRA. Also in this

case, the architecture is mainly focused on industry

4.0 applications, but it can be also adapted to smart

cities, e-mobility, energy, and buildings domains.

The Internet of Things Architecture (IoT-A)

(Bauer et al., 2013) is a reference architecture for IoT

applications, which has been deﬁned as an outcome

of the IoT-A European Union (EU) project

, which

had the main objective to develop an architectural ref-

erence model for the interoperability of IoT systems.

The last version of the architecture has been deﬁned in

2013 at the end of the project, but its use as reference

model is currently limited.

The Standard for an Architectural Framework for

the Internet of Things (IoT) (IEEE, 2020) has deﬁned

an architecture framework description for IoT, which

conforms to the international standard International

Organization for Standardization (ISO)/International

Electrotechnical Commission (IEC)/Institute of Elec-

trical and Electronics Engineers (IEEE) 42010:2011.

The standard provides an architectural blueprint

for Smart City implementation, considering cross-

domain interaction and enabling semantic interoper-

ability among various domains and components of a

Smart City (e.g., mobility, healthcare). For this rea-

son, also if not specialized for a single domain, the

standard is limited to Smart City related applications.

The results of the newest research projects, es-

pecially in the EU, brought to the harmonization of

several standard approaches, which brought to the

3D IoT Architecture model(Vermesan and Bacquet,

2018) as shown in Figure 1. The AIOTI High Level

Architecture (HLA) (AIOTI, 2020) has been deﬁned

by the AIOTI Working Group (WG) Standardization

for IoT to be applied by Large Scale Pilots (LSPs)

projects

. Capturing the commonalities shared among

Reference Architectures of the LSPs, the 3D Refer-

ence Architecture has been produced from the EU

https://www.eclipse.org/org/research/project/

https://cordis.europa.eu/project/id/257521/it

https://european-iot-pilots.eu/

IoTBDS 2022 - 7th International Conference on Internet of Things, Big Data and Security

Figure 1: 3D IoT architecture (Vermesan and Bacquet, 2018).

project CREATE-IoT

. The 3D architecture is generic

and offers a representation that can include the differ-

ent IoT/IIoT applications across different sector do-

mains (e.g., automated/autonomous vehicles, smart

farming, smart cities, energy, manufacturing, health).

The architecture includes the function by design con-

cept with end-to-end functions addressed across the 8

layers. This allows addressing the heterogeneous ap-

plications including different IoT platforms and pro-

cessing at the edge, fog, and cloud. The LSPs projects

are reﬂecting such a model in their architectures,

which completely represents this new level of com-

plexity.

In this work, we present a new concrete architec-

ture platform based on the 3D model. The BRAIN-

IoT platform is developed in an EU project

. It pro-

vides a set of integrated assets to build IoT applica-

tions from different domains. Assets can be used in

the different stages of the IoT development process,

starting from modeling to deployment and mainte-

nance. They provide the key functions deﬁned by the

3D reference model to address next-generation IoT

challenges. The next two sections will introduce the

BRAIN-IoT architecture, which then will be mapped

in the 3D reference model.

https://european-iot-pilots.eu/project/create-iot/

https://www.brain-iot.eu/

3 BRAIN-IoT ARCHITECTURE

Figure 2 shows the BRAIN-IoT architecture, which

includes several assets structured in six layers.

The application layer describes the applications

that are built. As mentioned earlier, the BRAIN-IoT

platform can be applied for generic IoT applications.

In this work, two scenarios will be presented : Service

Robotics for Warehouse Logistics (SRWL) and Criti-

cal Infrastructure for Water Distribution Management

(CIWDM). SRWL is a special logistic service involv-

ing several robotic platforms from Robotnik Automa-

tion company

, which need to collaborate to scan a

given warehouse and to assist humans in a logistic

domain. A ﬂeet of robots supports the movement of

different loads in a warehouse. CIWDM is a scenario

related to the management of water distribution net-

work in collaboration with EMALCSA company

. It

focuses on monitoring and control the management of

the water urban cycle in metropolitan environment of

the city of la Coruna in Spain. More details on the

applications realized by the BRAIN-IoT platform are

described in Section 6.

The modeling and validation layer provides sev-

eral functions, e.g., design of complex IoT systems

considering several abstraction levels, simulation and

https://robotnik.eu/

https://www.emalcsa.es/index.php/es/

BRAIN-IoT Architecture and Platform for Building IoT Systems

Figure 2: BRAIN-IoT architecture.

validation of systems behavior before their real de-

ployment, automatic code generation from the mod-

els, ensuring online intelligent services for prediction

and anomaly detection based on data, monitoring and

controlling IoT applications, modeling IoT devices

and checking the correctness of systems before its de-

ployment in the physical world.

The distributed execution layer is responsible of

dynamic deployment and execution of IoT applica-

tions. It allows assembling, conﬁguring and maintain-

ing runtime applications. It contains required IoT ser-

vices as well as functionalities for discovery, look-up,

and name resolution of IoT services. This layer is also

responsible for operational management and monitor-

ing of speciﬁc devices via the gateways deployed to

each BRAIN-IoT Edge Node.

The security layer provides multiple services for

protecting IoT systems such as security risk assess-

ment of IoT systems (Chehida et al., 2021), ensuring

end-to-end security from IoT devices/Cyber Physical

System (CPS) to application (Maillet-Contoz et al.,

2020), ensuring identity and distributed access con-

trol management for IoT devices/CPS and users, and

the implementation of privacy features for the IoT ap-

plications (Rashid et al., 2019).

The bottom layers represent connectivity proto-

cols and Application Programming Interface (API)

such as Long Range (LoRa) and Robot Operating

System (ROS), and physical IoT devices, e.g., actu-

ators, sensors.

4 BRAIN-IoT INTEGRATED

PLATFORM

Figure 3 shows the BRAIN-IoT integrated assets

available in the Eclipse Research Lab

, which re-

sponds to challenges C1, C2, C3, C4 and C5 pre-

sented in Section 1.

The BRAIN-IoT Services Development Toolkit

adopts a methodology and toolset for modeling dif-

ferent abstraction levels of IoT systems (C1). It is re-

sponsible for designing the logic of the IoT Applica-

tion, which composes the services provided by avail-

able devices (i.e., sensors, actuators, CPSs) and exter-

nal services (e.g., weather forecast, open data, third-

party IoT platforms, databases) based on their inter-

actions. The application logic is modelled along with

the relevant IoT environment using IoT-ML through

BRAIN-IoT Modeling Tool. The IoT-ML Modeling

Tool deﬁnes a Domain-Speciﬁc Modelling Language

(DSL), based on Uniﬁed Modeling Language (UML)

(UML2, 2017), to describe system-level components

choices and their dependencies, also IoT devices ca-

pabilities. Then, the IoT-ML model is reﬁned to Be-

havior, Interaction, and Priority (BIP) model repre-

senting formally the components behavior and their

possible interactions using automata-based models

expressed in the BIP language (Basu et al., 2011).

The BIP language integrated in BRAIN-IoT platform

https://github.com/eclipse-researchlabs/brain-iot

IoTBDS 2022 - 7th International Conference on Internet of Things, Big Data and Security

Figure 3: BRAIN-IoT Integrated Assets.

gives a precise semantics to the system models ex-

pressed by IoT-ML. Also, the BIP Modelling and

Veriﬁcation Tool provides efﬁcient tool for veriﬁca-

tion and analysis of the BIP models based on Statis-

tical Model Checking (SMC) technique. SMC uses

simulation-based approach to reason about formal re-

quirements expressed in temporal logic properties.

After simulation and veriﬁcation, the BIP model is

converted in Java source code as application arte-

facts through the BIP Code Generator. The gener-

ated artifacts are then released and stored in BRAIN-

IoT Services Repository, to be deployed and executed

by the distributed execution framework. The BRAIN-

IoT modeling and validation framework also offers

the ability to use development-time models to super-

vise running execution platform states using sensi-

Nact Studio. This solution enables monitoring of

BRAIN-IoT services and starting or stopping them

from the BRAIN-IoT Services Development Toolkit.

In addition, the generated IoT applications, named

as BRAIN-IoT services, can be validated leverag-

ing IoT physical device models with Physical Layer

Modelling Tool based on SystemC-TLM language

(Ghenassia, 2006) to validate the correctness of the

system behavior, before deploying it in the physical

world. Thanks to the Physical Layer Modelling Tool,

the system reliability is increased, as the validation

strategy of the end-device is strengthened by adding

scenarios focusing on device robustness considering

its interaction with system environment. Neverthe-

less, at the system level modelling, it is able to de-

scribe the services provided by the existing IoT de-

vices with Web of Things (WoT) Thing Description

so that enforces the services composition in the sys-

https://www.w3.org/TR/wot-thing-description/

tem models and provides the inputs to the edge node

at runtime to communicate with the external IoT de-

vice using the communication protocol it supports.

Finally, an AI approach supported by the sOnar tool is

provided to model and implement AI modules, which

can be composed with the system behavior models

(C3). The sOnar tool provides the online data analy-

sis service for anomaly detection and prediction func-

tionality using machine learning techniques. It acts

as an intelligent computing unit to enable autonomic

functionalities for the Runtime Infrastructure through

the deployment and execution of reusable software

AI/ML algorithms. It sends notiﬁcations to the execu-

tion framework to trigger a prompt reaction that miti-

gates or avoids negative consequences to the IoT/CPS

systems.

The BRAIN-IoT distributed execution frame-

work provides an autonomic distributed infrastructure

(BRAIN-IoT Fabric) for the dynamic deployment, dis-

covery and execution of BRAIN-IoT services on a

mixed cloud-edge environment (C4). The BRAIN-

IoT Fabric is combined with the Paremus Service

Fabric to manage communications between the devel-

oped applications from BRAIN-IoT Services Reposi-

tory through asynchronous events using the BRAIN-

IoT EventBus. By using BRAIN-IoT EventBus, the

multiple deployed applications are completely decou-

pled and the failure of one application has no impact

on other running applications. The distributed exe-

cution framework also provides the interoperability

for the external IoT devices/platforms, using WoT-

enabled Edge Nodes based on a World Wide Web

Consortium (W3C) WoT Thing Description of the

communications interface and sensiNact Edge Nodes

urgen et al., 2016) based on Eclipse sensiNact gate-

BRAIN-IoT Architecture and Platform for Building IoT Systems

way

, which can be deployed in a distributed and

bulk manner (C2). The sensiNact-enabled Edge Node

provides connectivity, interoperability, and data pro-

cessing to various IoT devices by its capability to in-

teract with a wide variety of equipment and protocols,

as well as its extensibility mechanisms. The WoT-

enabled Edge Node implements an approach to en-

able the interoperability between ROS-based CPS ap-

plications and other heterogeneous IoT platforms in a

sophisticated IoT software ecosystem.

The BRAIN-IoT security framework provides a

Security Module and Gateway, a Message Integrity

Service (MIS), and a distributed Authentication, Au-

thorization, and Accounting (AAA) Server to ensure

data conﬁdentiality, integrity, availability, and authen-

tication for the BRAIN-IoT platform (C5). The se-

curity module authenticates and encrypts data sent

over the network at the application level with reduced

energy consumption for IoT sensors and actuators.

The security gateway checks the sender’s authentica-

tion before decrypting the data. The distributed AAA

server manages identity and rights from users and IoT

devices. The MIS signs the data event before sending

it and veriﬁes the message integrity and sender au-

thentication when the event is received by a node. The

Attack-Defense Strategies Exploration Tool (Chehida

et al., 2020) is a decision-supporting tool which gives

the suggestions for the security considerations by pro-

viding insightful information that allows the security

manager to evaluate system vulnerabilities and to de-

sign reliable security policies. The tool identiﬁes the

potential attack actions that are most likely to succeed

such as network attacks and data manipulation known

as False Data Injection Attacks (FDIAs). It also se-

lects high impactful defense actions that make the sys-

tem harder to attack while ﬁnding a balance between

the attack cost and its probability of success. Further-

more, the Privacy Control System provides the policy-

based mechanism for the IoT users to protect the per-

sonal data collected using IoT devices. It is based on a

Policy Enforcement Point (PEP) that applies the poli-

cies and controls access to the data by available ser-

vices. It attaches the policies to the data event and

delivers it along with the data over the EventBus. The

Privacy Control System aims allowing a Privacy-as-a-

Service approach and facilitating the adoption of pri-

vacy policies control in IoT environments based on

SOA.

https://projects.eclipse.org/projects/technology.

sensinact

5 MAPPING BRAIN-IoT ASSETS

WITHIN THE 3D MODEL

The BRAIN-IoT architecture previously presented is

based on the 3D architecture introduced by AIOTI. To

better explain the link between the two architectures,

this section presents the mapping of the BRAIN-IoT

main components in the 3D architecture as shown in

Figure 4.

Speciﬁcally, the Physical layer includes actuators,

sensors and CPSs providing the connectivity cross-

cutting function and useful to support integrability.

On the Network Connectivity Layer there are the End-

to-end Security Module and Gateway providing secu-

rity and integrability as well as a set of different APIs

providing connectivity and integrability. On the Pro-

cessing Layer there are the Privacy Control System

for privacy control and dependability, as well as the

sensiNact Edge Nodes and the WoT Enabled Edge

nodes that provide connectibity and interoperability;

at the same layer the AAA server and the MIS provid-

ing reliability and dependability. The Storage Layer

includes the BRAIN-IoT Services Repository provid-

ing resilience and availability. On the Service Layer

there are BRAIN-IoT Fabric that provides connec-

tivity and composability; the BRAIN-IoT Moelling

Tool to provide manageability; the IoT-ML that pro-

vides interoperability; the BIP Code Generator, which

provides dependability; sOnar providing intelligence;

Brain-IoT Physical Layer Modeling Language to pro-

vide connectivity and integrability. Finally on the

Application Layer there are the BRAIN-IoT Attack-

Defense Strategies Exploration Tool providing secu-

rity and availability as well as the sensiNact Studio

that provides connectivity and manageability.

This mapping shows that the BRAIN-IoT compo-

nents are able to provide the main IoT cross-cutting

functions and IoT system properties needed to satisfy

the requirements of next-generation IoT challenges.

6 APPLICATIONS

As said in Section 3, the BRAIN-IoT platform has

been validated through two main scenarios: Service

Robotics for Warehouse Logistics (SRWL) and Criti-

cal Infrastructure for Water Distribution Management

(CIWDM).

In the ﬁrst scenario, the BRAIN-IoT platform sup-

ported the implementation of a multi-agent system for

the distributed self-organized management of a ﬂeet

of robots to move loads among the different areas of

a warehouse. The movement of these loads does not

require any operator to control the ﬂeet. Robots are

IoTBDS 2022 - 7th International Conference on Internet of Things, Big Data and Security

Figure 4: Mapping BRAIN-IoT assets with 3D IoT architecture.

expected to empty continuously the “inbound area”

of the warehouse, where the loads are temporarily po-

sitioned when items are delivered in for storage, mov-

ing them to a speciﬁc place in the ”storage area”.

When triggered by the human operator, each robot

autonomously moves from their base station to the in-

bound area and patrols looking for the loads and iden-

tifying them through their barcode. Consequently, the

robot establishes a communication with the backend

system of the warehouse to ask which cell of the ”stor-

age area” the load must be placed in. When the back-

end system return back the coordinates of the storage

cell, the robot picks the load up and moves it through

the warehouse to the right position in the storage area.

The navigation of the robots across the warehouse is

designed to be safe, avoiding obstacles along the path,

and adaptable to the speciﬁc environment where the

ﬂeet of robots is deployed. For instance, the same

warehouse logistics application has been deployed in

a scenario similar to a supermarket, where the items

to store can also be perishable, e.g. like fruits and

vegetables or ﬁsh and meat. In such scenario, the

storage area must be kept at a speciﬁc environment

temperature to not ruin the food, and it is separated

from the the other warehouse areas by a controllable

door. In such scenario, the robots are able to dis-

cover the door as a new controllable service available

on the network. Consequently, through the BRAIN-

IoT platform it is able to trigger the deployment at

run time of a new driver which enables the interac-

tion with the automated door. Afterwards, the robots

capabilities are then augmented allowing them to nav-

igate across the warehouse also opening and closing

the door when needed. In this scenario, the main ob-

jectives of the application of the BRAIN-IoT platform

were to reduce the development time for multi-agent

robot applications, establishing a secure end-to-end

interaction with robotic platforms (more speciﬁcally

ROS-based

robots) as well as legacy backend sys-

tems and external devices within the surrounding en-

vironment. Moreover, the use-case allowed also to

demonstrate and evaluate the dynamic autonomous

reconﬁguration of the system for self-adaptation to

the environment layout and anomaly situations.

In the second scenario, the BRAIN-IoT platform

is adopted to implement a safe autonomous control

for the management of a critical infrastructure like

the one distributing the water in the city of A Coru

na,

in Spain. The objective is to collect and exploit data

coming from geographically distributed and hetero-

https://www.ros.org/

BRAIN-IoT Architecture and Platform for Building IoT Systems

geneous devices and platforms, used for water level

monitoring and treatment, with the aim of early de-

tecting abnormal and critical situations that may hin-

der the correct functioning of the infrastructure or

even damage it. The good quality of the data is

fundamental for creating accurate indicators for deci-

sion making and for real-time adaptive control pro-

cedures. For this reason, the protection of the nu-

merous resource-constrained sensors and meters is

extremely important to guarantee the dependability

of the measured data. Also, it is very important to

guarantee that all the different data sources are au-

thorized to provide data, and no malicious injection

of fake information are present. More speciﬁcally,

the needed control consists in the following: when

the level of water reaches a speciﬁc threshold, the

spillgates shall be opened. Normally, the spillgates

opening is manually performed. In this use case,

BRAIN-IoT platform controls automatically the spill-

gates opening, based on the measures of the water

ﬂows. Such control strategies are modelled and devel-

oped with the BRAIN-IoT Modelling Tool. A model-

based approach allows to modify in an agile way the

control strategies whenever, in the future, modiﬁca-

tions or improvements will be needed. Furthermore,

the BRAIN-IoT platform is used for collecting data

from different water ﬂow meters placed in different

locations and analyzing them to detect possible de-

viations w.r.t. the normal behaviour. In fact, possible

data outliers (e.g. due to device obsolescence, miscal-

ibration, external attacks) may be extremely risky for

the water infrastructure because data are used to con-

trol the spillgates for let the water ﬂowing correctly

in the pipes and deliver the water correctly to the cus-

tomers. According to the outcomes of the analysis,

the BRAIN-IoT platform must autonomously recon-

ﬁgure the spillgates control strategy in such a way to

cut off the ones which are directly affected by the me-

ter’s misbehaviors. This approach allows to: enhance

the resiliency of the water infrastructure from failures

in order to provide a service which has reduced possi-

bilities of discontinuity; identify in real-time the pos-

sible problems with the meters and sensors, allowing

a prompt reaction for mitigating the issues and pro-

vide a better service experience to the customers; col-

lect as a whole the data coming from heterogeneous

meters and sensors which are spread all over the city.

7 VALIDATION RESULTS

The BRAIN-IoT platform has been evaluated through

workshops conducted carrying out a usability assess-

ment performing several teleconferences with rele-

vant stakeholders and were oriented to the compila-

tion of a standard well-known usability questionnaire,

which allows end-user of a product to evaluate their

feeling with the product in question. The results of

the inquiry demonstrated to be very valuable to mea-

sure the interest from the stakeholders for the pro-

vided functionalities and their degree of satisfaction

with the overall BRAIN-IoT Platform.

Using the UEQ, the items are scaled from -3 to

+3, where -3 represents the most negative answer, 0

a neutral answer, and +3 the most positive answer.

The items are categorised into six dimensions each

of which consists of 4-6 items on the UEQ which

thus describes a distinct quality aspect of an interac-

tive product. Attractiveness: general impression to-

wards the product. Efﬁciency: is it possible to use

the product fast and efﬁcient? Perspicuity: is it easy

to understand how to use the product? Dependabil-

ity: does the user feel in control of the interaction?

Stimulation: is it interesting and exciting to use the

product? Novelty: Is the design of the product in-

novative and creative? The results are also evaluated

using a benchmark. For this analysis, the measured

scale means are set in relation to existing values from

a benchmark data set, which contains data from 20190

persons from 452 studies concerning different prod-

ucts (business software, web pages, web shops, social

networks). The benchmark classiﬁes a product into 5

categories (per scale). Excellent: in the range of the

10% best results. Good: 10% of the results in the

benchmark data set are better and 75% of the results

are worse. Above average: 25% of the results in the

benchmark are better than the result for the evaluated

product, 50% of the results are worse. Below aver-

age: 50% of the results in the benchmark are better

than the result for the evaluated product, 25% of the

results are worse. Bad: in the range of the 25% worst

results.

The results of the evaluation of the BRAIN-IoT

platform show that the stakeholders perceived it as an

innovative solution. The discussion during the work-

shops let the self-administrative capabilities emerged

as the main innovative functionality, along with the

capacity of distribute edge intelligence in a decentral-

ized environment. The good score for the depend-

ability of the platform is due to the good comments

about the security perspective, more speciﬁcally for

the capability of protecting data sourced by resource

constrained devices, as well as the capacity to pro-

vide the data owner the control of the access poli-

cies of their data. Despite one of the objectives of

BRAIN-IoT is to relief the IoT developers and sys-

tem operators from the burden of implementing and

operating IoT applications involving multiple hetero-

IoTBDS 2022 - 7th International Conference on Internet of Things, Big Data and Security

Figure 5: Validation results of the BRAIN-IoT platform.

Figure 6: Validation results of the BRAIN-IoT platform applied to the SRWL scenario.

Figure 7: Validation results of the BRAIN-IoT platform applied to the CIWDM scenario.

geneous existing platforms, the perspicuity score is

not very high. The comments have been that the plat-

form is considered very helpful once it has been con-

ﬁgured and executed in the production environment,

but the conﬁguration phase seems to be harder than

expected, because of the difﬁcult of the modelling

phase and the setup of the BRAIN-IoT Fabric. The

low score for the perspicuity also driven the low at-

tractiveness, which is affected by the perceived dif-

ﬁculty for conﬁguration and setup of the whole plat-

form. It is however important to notice the most part

of the BRAIN-IoT Platform components have been

developed from scratch or starting from low Technol-

ogy Readiness Level (TRL) background assets. The

BRAIN-IoT Architecture and Platform for Building IoT Systems

consequence is that some more effort will need to be

invested in order to improve the usability in terms of

system setup. The good achievement is that the main

novel characteristics of the platform have been well

received and considered relevant and useful by the ex-

perts.

Figure 5 shows the positioning of the BRAIN-IoT

Platform relative to the benchmark. The results show

the good feeling w.r.t. the Novelty and Dependabil-

ity, for which the score is Excellent. Considering the

benchmark refers to solutions in every Information

and Communication Technologies (ICT) ﬁeld, meant

to be commercial products, the curve of the BRAIN-

IoT platform is in line with the expectation of a re-

search and innovation action, where the maturity of

the developed solution is still about around TRL 6,

but the innovative aspects are well evident.

A similar evaluation has been performed to evalu-

ate the software within the scope of the two applica-

tion scenarios. When applied to the Service Robotics

scenario, despite the perspicuity remains in line with

the values determined before, the attractiveness is

much higher. This is because the perception may

change a lot when you think about the technology ap-

plied to a speciﬁc context, and the beneﬁts that the

platform could bring are much more evident. Basi-

cally, the stakeholder but himself in the shoes of the

person who will beneﬁt from the execution of the plat-

form, instead of the IT manager who is supposed to

install and setup the system. This means that, despite

the BRAIN-IoT Platform continues to be evaluated as

complicate to setup, it is considered good for manag-

ing swarm robotics applications.

Figure 6 shows the positioning of the BRAIN-IoT

Platform relative to the benchmark. It shows even

clearly how the BRAIN-IoT Platform is considered

above the average in a context of SRWL. Instead,

the dependability is decreased because of the secu-

rity perspective in this context is more in line with

the state of the art technologies. As for the SRWL

scenario, a similar analysis could be made for the CI-

WDM. However, differently from the other scenario,

here the perspicuity is higher: while for the SRWL

scenario, the robotic applications developer cannot ig-

nore the system setup phase, in the case of the man-

agement of the CIWDM, the platform is evaluated

from the perspective of the pure end-user. Figure

7 shows the positioning of the BRAIN-IoT Platform

relative to the benchmark. In this case, the platform

has been evaluated as beneﬁcial for managing a criti-

cal infrastructure, especially considering the security

features and the resiliency capabilities.

8 CONCLUSION

BRAIN-IoT platform aims to pave the way of the

research around the strict requirements of the next-

generation IoT paradigm(AIOTI, 2020). In other

words BRAIN-IoT platform is a ﬁrst implementa-

tion of a meta operating system for the IoT domain,

which facilitates the implementation of secure and

self-adaptive IoT applications. The platform has been

tested and applied to robotics scenarios in which

BRAIN-IoT enables the self-controlled robots inter-

act with the environment and adopt their behaviors

correspondingly. In addition, BRAIN-IoT platform

demonstrated its advanced features in industry and

agile critical infrastructures management especially

for the physical infrastructure monitoring, and abnor-

mal behavior detection and prediction. Such meta op-

erating system maps toward the 3D architecture de-

ﬁned by AIOTI but some functionalities still need to

be either covered or enhanced, such as safety, trust-

worthiness. As part of the future works, the authors

plan to extend the BRAIN-IoT platform implement-

ing these functionalities with the aim to implement a

full featured meta operating system for the IoT.

ACKNOWLEDGMENT

The research leading to these results has been sup-

ported by the European Union through the BRAIN-

IoT project H2020-EU.2.1.1. Grant agreement ID:

780089.

REFERENCES

Adolphs, P. and Epple, U. (2015). Reference architecture

model industrie 4.0 (rami4.0). Standard, ZVEI – Ger-

man Electrical and Electronic Manufacturers’ Associ-

ation.

AIOTI (2020). High Level Architecture (HLA).

Standard, AIOTI WG Standardisation.

https://aioti.eu/wp-content/uploads/2020/12/

AIOTI HLA R5 201221 Published.pdf.

Basu, A., Bensalem, S., Bozga, M., Combaz, J., Jaber,

M., Nguyen, T.-H., and Sifakis, J. (2011). Rigor-

ous Component-Based System Design Using the BIP

Framework. IEEE Software, 28(3):41–48.

Bauer, M., Boussard, M., Bui, N., Carrez, F., Jardak, C.,

Loof, D., Magerkurth, C., Meissner, S., Nettstr

ater,

A., Olivereau, A., Thoma, M., Walewski, Stefa, J.,

and Salinas, A. (2013). Internet of things – architec-

ture iot-a deliverable d1.5 – ﬁnal architectural refer-

ence model for the iot v3.0. Technical report.

Chehida, S., Baouya, A., Alonso, D. F., Brun, P.-E., Mas-

sot, G., Bozga, M., and Bensalem, S. (2021). Asset-

IoTBDS 2022 - 7th International Conference on Internet of Things, Big Data and Security

driven approach for security risk assessment in iot sys-

tems. In Garcia-Alfaro, J., Leneutre, J., Cuppens, N.,

and Yaich, R., editors, Risks and Security of Internet

and Systems, pages 149–163, Cham. Springer Interna-

tional Publishing.

Chehida, S., Baouya, A., Bozga, M., and Bensalem, S.

(2020). Exploration of impactful countermeasures on

iot attacks. In 2020 9th Mediterranean Conference on

Embedded Computing (MECO), pages 1–4.

Ghenassia, F. (2006). Transaction-Level Modeling with Sys-

temc: Tlm Concepts and Applications for Embedded

Systems. Springer-Verlag, Berlin, Heidelberg.

urgen, L., Munilla, C., Druilhe, R., Gandrille, E., and

Nascimento, J. (2016). sensiNact IoT Platform as a

Service, pages 127–147.

IEEE (2020). Ieee standard for an architectural framework

for the internet of things (iot). IEEE Std 2413-2019,

pages 1–269.

Lin, S.-W., Miller, B., Durand, J., Bleakley, G., Chigani,

A., Martin, R., Murphy, B., and Crawford, M. (2019).

The industrial internet of things volume g1: Refer-

ence architecture. Standard, Industry IoT Consortium.

https://www.iiconsortium.org/pdf/IIRA-v1.9.pdf.

Maillet-Contoz, L., Michel, E., Nava, M. D., Brun, P.-E.,

Lepr

etre, K., and Massot, G. (2020). End-to-end secu-

rity validation of iot systems based on digital twins of

end-devices. In 2020 Global Internet of Things Sum-

mit (GIoTS), pages 1–6.

Rashid, M. R. A., Conzon, D., Tao, X., and Ferrera,

E. (2019). Privacy Awareness for IoT Platforms:

BRAIN-IoT Approach. In Ramos, J. L. H. and

Skarmeta, A. F., editors, Security and Privacy in the

Internet of Things: Challenges and Solutions, vol-

ume 27 of Ambient Intelligence and Smart Environ-

ments, pages 24–43. IOS Press.

omer, L., Jeroschewski, S. E., and Kristan, J. (2020).

Leveraging eclipse iot in the arrowhead framework.

In NOMS 2020 - 2020 IEEE/IFIP Network Operations

and Management Symposium, pages 1–6.

UML2 (2017). Uniﬁed Modeling Language (Version 2.5.1).

Object Management Group.

Vermesan, O. and Bacquet, J. (2018). The Next Genera-

tion Internet of Things – Hyperconnectivity and Em-

bedded Intelligence at the Edge, pages 19–102. River

Publisher.

BRAIN-IoT Architecture and Platform for Building IoT Systems