A Software Framework for Context-aware Secure Intelligent

Applications of Distributed Systems

Soumoud Fkaier

1,2,3 a

, Mohamed Khalgui

2,3 b

and Georg Frey

1 c

Chair of Automation and Energy Systems, Saarland University, Saarbruecken, Germany

Tunisia Polytechnic School, Carthage University, Tunis, Tunisia

INSAT LISI Lab, Carthage University, Tunis, Tunisia

Keywords:

Framework, Context-awareness, Reconﬁguration, Intelligence, Coordination, Security.

Abstract:

Future distributed reconﬁgurable systems need to provide smarter services. Therefore the used software need

to include advanced mechanisms such as the context-awareness, artiﬁcial intelligence, collaboration between

distributed parts of the system, as well as the secure data exchange. Most of the existing context-aware

frameworks are restricted to a part of the mentioned scopes and are generally not suitable to reconﬁgurable

systems. Hence, there is a need for a software engineering solution that reconciles all the said requirements. In

this paper, we propose a software framework for developing collaborative, intelligent, and secure applications

of distributed systems. This paper extends an existing framework with the mentioned features and shows its

new structure and design. A software tool developing the proposed contributions is implemented using Java

programming language. An example of microgrids software applications is used to show the suitability of the

contributions.

1 INTRODUCTION

Smart behaviour is the main objective of current and

future context-aware reconﬁgurable systems. These

latter have to meet the challenge of providing more

sophisticated behaviour. Therefore, its software ap-

plications need to be strong enough to offer such fea-

tures. The advanced features can not only be reached

by means of context-awareness computing, but also

by means of different other concepts. Artiﬁcial in-

telligence concepts can play a major role to support

smartness since they allow to develop knowledge rea-

soning for multiple use cases (e. g. the prediction,

planning, learning). Besides, distribution paradigm is

becoming more and more substantial and most of the

current systems are considering it thanks to its assets

when compared to the centralized one. In fact, cen-

tralized approach has the limitation of single-point-

of-load and single-point-of-failure, which is less efﬁ-

cient for many cases. Furthermore, secure informa-

tion exchange between the distributed parts is an un-

deniable condition for reliable system behaviour. In

https://orcid.org/0000-0003-2903-0943

https://orcid.org/0000-0001-6311-3588

https://orcid.org/0000-0001-8483-0295

addition to all that, common requirements such as the

real-time as well as functional constraints, need also

to be satisﬁed.

The development of context-awareness software

applications of such systems is a complicated task to

perform. For this, different context-awareness soft-

ware frameworks have been proposed at the aim of

facilitating their development. However, most of

them handle limited features due to the complexity

of their orchestration. It is hard to provide a generic

software solution that can cover intelligence, coor-

dination and security for different case studies in a

same software support. In fact, a design challenge

rises in the deﬁnition of the structure of a framework

that needs to cover miscellaneous requirements. Few

are the frameworks that consider the coordination re-

quirement of distributed systems along with intelli-

gence or security. The work reported in (Fkaier et al.,

2016a) introduces and important framework that al-

lows to implement context-aware reconﬁgurable ap-

plications running under real-time and functional con-

straints but despite its importance, it is still missing

the ability to handle advanced intelligence issues, to

operate smoothly in a distributed system, and to guar-

antee security of exchanged data.

To overcome the aforementioned limitations, the

Fkaier, S., Khalgui, M. and Frey, G.

A Software Framework for Context-aware Secure Intelligent Applications of Distributed Systems.

DOI: 10.5220/0010604701110121

In Proceedings of the 16th International Conference on Software Technologies (ICSOFT 2021), pages 111-121

ISBN: 978-989-758-523-4

111

current paper proposes an extension to the framework

reported in (Fkaier et al., 2016a). We mainly enhance

the ﬁrst and second layers (the Reconﬁguration Layer

and Context Control Layer) by adding new pools and

reﬁning the deﬁnition of existing ones. We propose a

generic artiﬁcial intelligence mechanism that can be

used for different purposes. We also introduce a co-

ordination mechanism that helps to achieve coherent

reconﬁgurations. Moreover, we deﬁne a module to be

a container of security mechanisms. Hence, besides

the existing techniques, the new enhanced framework

allows to implement more features promoting appli-

cations smartness and autonomy. Furthermore, we

implemented the proposed framework in a software

tool using Java programming language.

In order to show the suitability of the new frame-

work, an example of microgrids software applica-

tions is addressed. In fact, smart microgrids are

a type of systems that are knowing an evolution

through integrating the Information and Communica-

tion Technologies (ICT). Despite the numerous chal-

lenges identiﬁed in this system, most of works tackle

the electric power level or the automation and control,

and the software side is generally omitted although

it is a key responsible for system smartness (Anvaari

et al., 2012), (Ghosh et al., 2020).

The outline of this paper is organized as follows.

Section 2 presents the state of the art. Section 3 intro-

duces the deﬁnition of the new framework. Section

4 shows how the new concepts are suitable through a

microgrids software application case study. Section 5

evaluates the performances of the contributions. Fi-

nally, Section 6 concludes the paper and opens per-

spectives.

2 STATE OF THE ART

In this section, we present in Section 2.1 an overview

of the existing context-awareness frameworks. Then,

we show in Section 2.2 the importance of context-

awareness, intelligence, security, and coordination for

smart grids.

2.1 Context-awareness Frameworks

Since its appearance, context-awareness computing

has been an important enabler of systems smart be-

havior. However, alone, this concept cannot provide

expected results especially with the never-ending ex-

pectations of systems autonomy. This is why, soft-

ware frameworks have been intensively developed.

The authors of (Sikder et al., 2019) have proposed a

framework for the detection of threats on smart de-

vices, but despite its importance this work do neither

consider the real-time reconﬁgurations nor the coor-

dination between distributed agents. In (Aid and Ras-

soul, 2017) a context-aware framework for the dis-

aster management is proposed. This work also does

not treat distribution, security and real-time reconﬁg-

urations. The work reported in (Tang et al., 2014)

presents a development environment that supports the

pervasive applications. Distribution needs are taken

into consideration by this work but requirements like

secure coordination and real-time reconﬁguration are

not covered. The work reported in (Alhamid et al.,

2016) proposes a collaborative context-aware frame-

work for the development of applications in ambi-

ent intelligence environment. This framework also

does not consider distribution, security, and recon-

ﬁguration. In (Bucchiarone et al., 2017) a context-

aware framework for the development of processes in

the internet of services is proposed. This latter does

not handle the coordination and real-time reconﬁg-

urations. In (Fkaier et al., 2016a) the authors have

proposed a context-awareness framework for the de-

velopment of reconﬁgurable systems having real-time

and functional constraints. The latter is an important

support of the development of future smart systems,

however it does not support the distribution, security,

and advanced intelligence features. As it can be seen,

most of the existing context-awareness frameworks

are dedicated to speciﬁc features, and the majority do

not tackle coordination and security.

2.2 Smart Grids and Intelligence,

Security, Distribution

Smart grids are new electricity grids promoting the

integration of ICTs into the traditional grid in order

to enable enhanced functionalities through the ex-

change of data between the grid components (users,

utility, and control infrastructure). These grids need

to use advanced software concepts to achieve the re-

quired functionalities. In (Fkaier. et al., 2020a) and

(Fkaier. et al., 2020b) context-awareness comput-

ing was used to enable the dynamic interaction with

the changing environment in real-time. Concerning

the intelligence, the authors of (Henri et al., 2020)

have proposed a simulator of artiﬁcial intelligence

concepts for microgrids. In (Khan et al., 2018) also

discusses the adoption of artiﬁcial intelligence tech-

niques in the microgrids development. The research

reported in (Hubana, 2020) uses the artiﬁcial intel-

ligence concepts for faults location and recovery in

microgrids. The security have been an inherent re-

quirement to all computerized systems. Ensuring se-

ICSOFT 2021 - 16th International Conference on Software Technologies

112

curity is an important condition to maintain a good

reliability level. The work reported in (Elgamal et al.,

2020) studies the microgrids proﬁt maximisation all

with sustaining its security. In (Hosseinimoghadam

et al., 2020) the authors study the security of micro-

grids based on neuro-fuzzy inference model. The au-

thors of (Dabbaghjamanesh et al., 2019) have used the

blockchain technology to enhance the privacy of net-

worked microgrids. The research in (Ferdowsi et al.,

2019) has also studied the reconﬁguration under se-

curity constraints. With the increasing adoption

of distributed paradigm in the operational tasks of

microgrids, coordination becomes necessary for the

logic coherence. The authors of (Mbungu et al., 2019)

have surveyed the optimal smart energy coordination

for microgrids. In (El-Naily et al., 2020) a coordina-

tion approach is used to resolve the microgrids faults,

while in (Wu et al., 2019) a distributed hierarchical

coordination strategy for parallel inverters is proposed

to improve the system ﬂexibility. As it can be seen,

most of the presented works about smart grids soft-

ware are restricted to one or two of the system re-

quirements. Software infrastructures that enables to

develop multiple requirements using one software so-

lutions are still needed.

3 ENHANCED FRAMEWORK

In this section, we introduce the enhanced framework,

where Section 3.1 shows the extension of RL and Sec-

tion 3.2 shows the extension of CCL.

The selected framework, was introduced in

(Fkaier et al., 2016a) and its concepts are based on a

meta-model reported in (Fkaier et al., 2017). A soft-

ware tool implementing the framework semantics is

reported in (Fkaier et al., 2016b).

The framework has a layered architecture com-

posed of four layers: (i) Reconﬁguration Layer (RL):

responsible for the interaction of the framework logic

with the environment through two modules of inputs

and outputs, (ii) Context Control Layer (CCL): con-

tains a set of pools where each of which is responsible

for the checking of well-deﬁned constraints, (iii) Ser-

vices Layer (SL): holds the functionalities of applica-

tions in the form of services, and (iv) Communication

Layer (CL): represents of the framework’s services.

The second layer, CCL, was composed of three

pools: controller (C), functional (FP), and real-time

(RTP) pools. The real-time pool provides the tech-

niques to analyze the timing behavior of applica-

tions through checking the timing constraints, e. g.,

the operational deadlines. The functional pool of-

fers the ability to check functional constraints related

to dependencies between the services (e.g. prece-

dence constraints) and especially the coherence rela-

tionships (i.e., the inclusion/exclusion relationships).

We come ﬁnally to the last pool, the controller, that

ensures the orchestration of the operation of the pools

and layers of the whole architecture, and holds the ap-

plication’s logic.

This framework is proved to be efﬁcient in the de-

velopment of software applications of complex sys-

tems such as the airports baggage handling systems as

reported in (Fkaier et al., 2016a). However, despite its

undeniable assets, it still have some limitations espe-

cially for the development of distributed, intelligent,

and secure systems. To satisfy these needs, we extend

the deﬁnition of the framework concepts by updating

the deﬁnition of the ﬁrst and second layers (RL and

CCL). We reﬁne the mechanisms of the inputs pool by

integrating a data processing mechanism using more

sophisticated context-awareness reasoning technique

of RL. We deﬁne also three new pools in the CCL: (1)

intelligence pool, (2) coordination pool, and (3) se-

curity pool. The new schematic representation of the

framework structure is depicted in Figure 1. In ad-

dition, the extension includes updating the controller

capabilities in a way to support the interaction with

the new pools. The internal dynamics of each new

pool are deﬁned in the following.

Figure 1: Framework architecture.

3.1 RL Extension

Input data are values coming from the sensors and

measurements infrastructure, then structured thanks

to a hardware control layer, and ﬁnally handed-over to

the software layer, speciﬁcally to the IP of the frame-

work. Figure 2 shows the abstraction of the informa-

tion ﬂow from ﬁeld level until the Inputs Pool.

The Inputs Pool receives input data (denoted by

A Software Framework for Context-aware Secure Intelligent Applications of Distributed Systems

113

Figure 2: Schematic presentation of the inputs pool.

id), processes it, then creates a Context Entry CE and

sends it to the controller pool of the upper layer (i.e.,

CCL). Input data are classiﬁed into two types: criti-

cal and uncritical. In order to specify which data are

the critical and which are the uncritical, developers

have to create the context ontology as mentioned in

(Fkaier. et al., 2020a). The context ontology con-

sists in a basis to generate Context Rows CR where

CR = {i

,...,i

} with i

is a context item. A con-

text item is deﬁned as a tuple i =< t,l,v > where t is

the type of the item expressed through its root in the

ontology model (i.e., the ontology o

, entity e

, and

attributes a

), l is the label of the item, and v is the

value of the item. If the input data are classiﬁed as

critical, then the Context Entry Creator (see Figure 2)

creates the context entry as

CE =



(ts||id) if id is critical

(ts||id||r) if id is uncritical

(1)

with ts is a time stamp and r is the set of recommenda-

tions resulted from the context reasoning. If the input

data are classiﬁed as uncritical then a context aware-

ness reasoning is performed according to reasoning

method deﬁned in (Fkaier. et al., 2020a). The pseu-

docode of the input module is then deﬁned as men-

tioned in Algorithm 1.

3.2 CCL Extension

3.2.1 Controller

After receiving CE from the inputs pool, the con-

troller parses it and decides whether a reconﬁguration

should take place. This logic is use-case-dependent

and must be deﬁned by the system owners. In case

a reconﬁguration is required, the controller starts to

search for the appropriate conﬁguration. For this, the

Algorithm 1: Inputs Pool Logic.

Inputs: id

Outputs: CE

CE =

while (true)

critical

=ListenForCriticalData();

id=id

critical

;

CE = CreateCE(ts, id);

end

foreach (period) do

id= ReadData();

r = ContextReasoning(id);

CE = CreateCE(ts, id,r);

end

if (input demand) then

id= ReadData();

CE = CreateCE(ts, id);

end

return CE;

controller need to analyze the convenience of a certain

conﬁguration with regard to some constraints. Pre-

viously, only functional and timing constraints can

be analyzed, but now more advanced requirements

can be checked such as the intelligence, coordination

with other peers in the system, and security of criti-

cal transactions. By the addition of the new pools, the

controller code is also extended in order to support

the new components. New communication requests

and replies are deﬁned and the structure of the con-

troller code becomes as depicted in the UML (Uni-

ﬁed Modeling Language) class diagram of Figure 3.

The code is organized with respect to the separation of

concerns as well as the single responsibility principles

into three packages Handling Layers, Using Pools,

and Control Main Logic: 1) The package Handling

Layers contains the classes that enable the controller

ICSOFT 2021 - 16th International Conference on Software Technologies

114

to supervise the work of the RL, SL, and CL. 2) The

package Using Pools contains the classes enabling the

use of one or many pools (developers are free to use

the pools they need according to the use case and not

necessarily all pools). 3) The package Control Main

Logic as its name indicates holds the main logic of the

application.

Figure 3: Controller class diagram.

3.2.2 Artiﬁcial Intelligence Pool (AIP)

The aim behind adding this pool in CCL is to make

applications more intelligent. Artiﬁcial intelligence

concepts can be used for many purposes such as the

optimization tasks of a system, prediction tasks, etc.

For this, we propose to create a generalized artiﬁcial

intelligence technique that can be adapted/adjusted

according to the use case. This pool analyzes an in-

telligence request created by the controller in order to

process it using an Expert System (ExpSys). Gen-

erally, existing expert systems consist of a knowl-

edge base and an inference engine, but for more ef-

ﬁciency in terms of computational power (used re-

sources such as the memory, CPU time, energy, etc.)

we add a system history that helps to avoid trigger-

ing known scenarios. Hence the ExpSys is deﬁned

by, ExpSys = (I,K,H), where I denotes the inference

engine, K denotes the knowledge base and H denotes

the intelligence pool history (see Figure 4).

Figure 4: Intelligence Pool.

The knowledge base, K, accomplishes its task us-

ing two bases K = (F,R): (1) the facts base, given by

F = { f

, j = (1,..., n)}, and it contains facts about the

system where a fact generally reﬂects an evidence. (2)

The rules base, given by R = {r

,i = (1,...,m)}, stor-

ing rules which are logical forms enabling the deduc-

tion of new facts. The rules must be deﬁned by human

experts of a speciﬁc domain that muster the consid-

ered case and know how the system should behave. A

rule has the form of r

=“if premise then conclusion”.

The system history, H, stores the history of the sys-

tem behavior deﬁned by the set: {h

,q = (1,..,k)},

where h

represents a previous demarche executed by

the pool. The inference engine, I, uses K and H in

order to deduct a new knowledge. The method of rea-

soning of I can be the forward or backward chaining.

3.2.3 Coordination Pool (CP)

In distributed approach, the operational tasks are per-

formed based on geographically distributed peers.

One important challenge that has often been faced is

how to make a real coherent distributed and collabo-

rative strategy.

The coordination between the distributed units is

not an easy task to perform especially that the need to

other peers can be manifested during the preparation

of reconﬁgurations. This is why, we propose a coor-

dination pool, denoted by CP, in the CCL that will be

able to satisfy this requirement.

CM[i][ j] =

system peers

z }| {







... c

.. .. .. ..

.. c







)

con f igurations

(2)

To accomplish its task, CP uses a matrix of conﬁgura-

tions called Coordination Matrix (CM) of size (n,m)

in which the columns represent all the existing peers

in the system, and the lines represent the different

conﬁgurations. Whenever a coordination with other

system peers is needed, a speciﬁc peer must consult

its CM and depending on the context decides which

(re)conﬁguration to process.

3.2.4 Security Pool (SP)

Coordination between the distributed peers of a sys-

tem requires an exchange of data where some of them

can be critical. Hence, securing them is important

to guarantee successful applications deployment. In

particular, it is required to ensure data integrity and

conﬁdentiality of some functionalities and/or recon-

ﬁgurations. For these reasons, we add a security pool

A Software Framework for Context-aware Secure Intelligent Applications of Distributed Systems

115

to be a container of security mechanisms such as the

algorithms necessary for encryption/decryption (e.g.

RSA, ElGamal, ECIES), user access supervision, er-

ror detection software, blockchain-related algorithms

(e.g. mining, consensus, validation). This pool pro-

vides the base to use such techniques and developers

can extend it with any required technique. Hence, this

pool is deﬁned as SP={st

,...,st

}, where st

stands for

a security technique.

4 CASE STUDY

This section presents an example of microgrids soft-

ware application development using the framework.

Section 4.1 presents the case study and Section 4.2

depicts the software development.

4.1 Case Study Presentation

Microgrids are electricity grids promoted with infor-

mation systems to make smarter electricity genera-

tion, distribution, and consumption. Microgrids are

also characterised by the integration of the renewable

energy resources and the electricity storage systems

as detailed in (Fkaier. et al., 2020a). Hence, complex

and diversiﬁed functionalities need to be included in

the software applications in order to provide efﬁcient

operation. In this paper, we demonstrate how the pro-

posed framework facilitates the development of such

functionalities. We consider the microgrids system

depicted in Figure 5 where the proposed framework

is used to the development of software applications

of the software level. The considered grid consists of

three microgrids mg

, mg

, where each has its

loads, renewable sources, and storage system. In this

paper, the conﬁgurable components set of a microgrid

is deﬁned as C

= {B,U,RES,ROS}, where B is

the set of batteries, U is the connection to utility, RES

is the set of renewable energy sources, and ROS is the

set of remotely operating switches that connects to the

other microgrids.

Every microgrid mg

has a set of conﬁgurations,

denoted by CFG = {c f g

,c f g

,...,c f g

}, with n is

the number of all conﬁgurations. A conﬁguration

c f g

determines the combination of components ar-

rangements (i.e., the mode of use). A c f g

is com-

posed of two main sections: (i) internal conﬁguration,

c f g

int

: determining the way of use of the microgrid

components such as the batteries, the switch of the

utility, and the RES for internal functionalities, (ii)

external conﬁguration, c f g

ext

: determining the way

of use of the microgrid components for external func-

tionalities. These external functionalities are used, in

Figure 5: Considered microgrids system.

this paper, in the trading or emergency cases.

4.2 Software Applications Development

The software applications of the three microgrids

mentioned in Figure 5 are implemented with the

proposed framework. For simplicity reasons, let

us consider only three services that are performed

by the application as follows: trading electric-

ity with other microgrids, storage management,

and the renewable energy integration. Hence,

the deﬁnition of the services layer is given by

SL = {s

Trading

BatteryScheduling

RESControl

}. Figure 6

shows the structure of the microgrid application ac-

cording to the proposed framework.

Figure 6: Microgrids application strcuture.

Before showing the details of the proposed scenarios,

let us ﬁrst deﬁne the settings of the framework that

are parts of the deﬁned pools, such as possible con-

ﬁgurations CFG, context model with Ontology Web

Language (OWL), context rules store CRS, rules base

ICSOFT 2021 - 16th International Conference on Software Technologies

116

RB of the AIP, the coordination matrix CM of the CP.

Let us consider the possible conﬁgurations of mg

shown in Table 1. This table is used by the controller

module of CCL to determine reconﬁgurations and it

is saved in an XML format.

Table 1: Conﬁgurations of mg

Internal External

ID B U RES ROS2

Cfg 1 MC On TA PS

Cfg 2 HC On PA TS

Cfg 3 MD On TA PS

Cfg 4 HD On PA TS

In this table MC: Medium Charging, HC: High Charging, MD: Medium Dis-

charging, HD: High Discharging, TA: Total Activation, PA: Partial Activa-

tion, TS: Total Supply, PS: Partial Supply.

In this paper, we deﬁne the microgrid context model

with OWL language as shown in Figure 7. The con-

text modeling is created with the aim of analyzing real

measurements that help to decide whether to charge or

discharge batteries and also whether to turn on or off

the connection to the main utility.

Figure 7: Microgrids context model with OWL language.

Using the context model of Figure 7, the context rules

store is deﬁned as depicted in Table 2. The role of

the AIP is to decide whether it is required to initiate a

trading session to buy electricity, to analyze more the

recommendations of the RL especially in case of rec-

ommendation to switch off from the main grid. It is

needed to avoid blackouts and to decide how to efﬁ-

ciently use the battery (when to charge/discharge, and

with which intensity medium or high). The rules used

by the AIP are given in Table 3.

Let us consider a time driven scenario to manifest

the need to context-awareness, reconﬁguration, AIP,

CP, and SP. The inputs pool, as mentioned in the pre-

vious section, sends a CE in a periodic way in case of

uncritical data and in sporadic way (i.e., event driven)

in case of critical data. Initially, the application uses

the conﬁguration C f g

mentioned in Table 1 which

means that the microgrid is highly charging the bat-

teries while it is not activating all RES. Also, it has

the context row CR

mentioned in Table 4.

Scenario 1: At t

= 12 : 00, 15

,July. The inputs

pool collect data in order to create a new context row

. Let us assume the newly values are as men-

tioned in Table 5.

According to the context reasoning process

shown in Figure 2, the difference between the

current and new rows is calculate as follows:

Di f f erence =

∑

j=1

(|x

− v

|/[(x

+ v

)/2] × 100)/n,

which gives 24,76%. Assuming that the ﬁxed

difference threshold is 10%, we need to proceed

to the next step and trigger rules of CRS. This

means that the context is changed, mainly timing

ontology, and new recommendations need to be

found to achieve better efﬁciency. From the CRS of

Table 2, the conditions of R2 are met, then R2 is ﬁred

and recommendations to “Discharge batteries” is

obtained. Therefore, the IP creates the following con-

text entry CE

= (12 : 00,15/07||PeriodO f Day =

Midday,LevelO fCharge = 90% ||(Discharge

batteries). The controller of CCL receives CE

and

must now ﬁnd whether this recommendation is efﬁ-

cient. Since such a decision is hard to take and human

expertise is required to determine which can preserve

batteries efﬁciency, the controller send a request to

the AIP for analysis. The controller module, with

the help of the IP, provides the information of actual

value of loads (L) which is high since it is midday,

the actual level of solar panels generation (RESG)

which is also high, the actual market electricity price

(MEP) which is medium, and the predicted solar

generation of tomorrow (TSG) this is assumed to be

fed by the weather forecast agency, and it is predicted

to be low. Having these facts, the AIP parses the RB

and collects triggerable rules. In this case, Rule 4

can be triggered, which results in “medium battery

discharge”. Figure 8 depicts some of the methods

developed as part of the AIP.

Figure 8: Some of the methods of AIP.

The result is given back to the controller and this latter

searches in the possible conﬁgurations (see Table 1)

A Software Framework for Context-aware Secure Intelligent Applications of Distributed Systems

117

Table 2: Context Rules Store (CRS).

R1 if (Weather is Sunny) and (PeriodOfDay is Morning) and (Season is Summer) → charge batteries

R2 if (Weather is Sunny) and (PeriodOfDay is Midday) and (Season is Summer) → (discharge batteries)

R3 if (Weather is Cloudy) and (PeriodOfDay is Morning) and (Season is Summer) → switch on grid

R4 if (Weather is Cloudy) and (PeriodOfDay is Midday) and (Season is Summer) → discharge batteries

Table 3: Rules Base (RB) for mg

Rule1 if (BC is high) and (L is low) and (RESG is high) and (MEP is high) and (TSG is high)→ high battery charge

Rule2 if (BC is medium) and (L is low) and (RESG is low) and (MEP is high) and (TSG is low)→ medium battery charge

Rule3 if (BC is high) and (L is medium) and (RESG is high) and (MEP is high) and (TSG is low)→ high battery discharge

Rule4 if (BC is high) and (L is high) and (RESG is high) and (MEP is medium) and (TSG is low)→ medium battery discharge

In this table BC: battery level of charge, L: loads, RESG: level of solar panels generation, MEP: market electricity price, TSG: tomorrow solar generation.

Table 4: Initial context row CR

< o

utility

ConnectionMode

,OnMode,1 >

< o

Battery

LevelO fCharge

,medium, 50% >

< o

Weather

type

,Sunny,2 >

< o

Time

Season

name

,Summer,1 >

< o

Time

PeriodO f Day

name

,Morning,1 >

Table 5: Measured context row CR

< o

utility

ConnectionMode

,OnMode,1 >

< o

Battery

LevelO fCharge

,high,90% >

< o

Weather

type

,Sunny,2 >

< o

Time

Season

name

,Summer,1 >

< o

Time

PeriodO f Day

name

,Midday,2 >

which conﬁguration matches the need. C f g

is the

conﬁguration that suits the requirements of the con-

text. Hence, the controller starts to change from c f g

to c f g

by loading the proper methods developed as

part of the service s

BatteryScheduling

which contains the

right methods enabling the new battery and RES op-

eration style. The execution of the changes must re-

spect functional and timing constraints as detailed in

(Fkaier et al., 2016a).

As it can be seen from Scenario 1, the enhanced

inputs pools served to provide awareness about the

system environment where a reconﬁguration possibil-

ity is then analyzed by the controller module. Also,

thanks to the AIP the battery scheduling measures are

consolidated using the experts knowledge.

Scenario 2: Later on, at t

= 16 : 00, 15

,July.

receives an emergency call published by mg

af-

ter blackouts are occurring due to rocks and trees fall

on electric lines, which requires some time to be re-

paired. The emergency call includes the information

of the needed electricity quantity Q

and the re-

quired time of supply T

. In this case, a coordina-

tion session starts between the three microgrids based

on the coordination matrix (see Table 6) as follows:

• mg

publishes an emergency request R

) to the rest of the grid.

• mg

(resp. mg

) checks the current and planned

resources, and decide if it is possible to help mg

• mg

(resp. mg

) parses the coordination matrix

CM and selects the possible conﬁgurations.

• mg

(resp. mg

) sends the found possible conﬁg-

uration to mg

• mg

selects a conﬁguration based on this logic:

– if received conﬁgurations contain a conﬁgura-

tion where only one microgrid can provide the

electricity, then select this conﬁguration. In

case more than one conﬁguration is found, then

select the one related to the nearest microgrid

geographically.

– else select the shared conﬁguration between

and mg

. If more than one conﬁguration

are shared, than select the one with nearest mi-

crogrids geographically.

con f ig

, con f ig

, and con f ig

are the subject of dis-

cussion between mg

and mg

, where any decision

is made based on the current conﬁguration of a mi-

crogrid (i.e., c f g

for mg

). Assuming that mg

can

only provide a partial supply since it does not have

big beneﬁt from the water dams as RES in the sum-

mer, con f ig

is the matching conﬁguration. In case

a partial supply is required by mg

and mg

, internal

reconﬁgurations (see Table 1) need to be conducted

as mentioned in Scenario 1. In order to transfer elec-

tricity to mg

, mg

needs to activate the remotely op-

erating switches binding the two microgrids.

We can see from this scenario, that the use of the

coordination pools has helped microgrids to resolve

the fault through a collaboration based on known con-

ﬁgurations.

Scenario 3: at t

= 09 : 00, 18

,July. mg

pub-

lishes a request R

to buy the quantity of electricity

ICSOFT 2021 - 16th International Conference on Software Technologies

118

Table 6: Coordination Matrix (CM).

con f ig

F TS I

con f ig

F I TS

con f ig

F PS PS

con f ig

PS F PS

con f ig

TS F I

con f ig

I F TS

con f ig

PS PS F

con f ig

I TS F

con f ig

TS I F

In this table I: idle, F: Faulty, PS: Partial Supply, TS: Total Supply.

during the period T

. Here in order to par-

ticipate in the trading process, each microgrid should

check its resources and ﬁnd the proper conﬁguration

to deploy in case of its selection as a seller. Details

of context analysis and intelligent storage handling

should be conducted as demonstrated in the ﬁrst sce-

nario. In this scenario, we focus on another important

side of microgrids communication, which is the secu-

rity of exchanged data.

In order to properly conduct a trading process

that preserves the data immutability and integrity,

blockchain technology is used as a secure storage

medium. Hence, each microgrid must exchange data

through the blockchain (i.e., requests, bids, decisions

are added to the chain). In order to be able to pub-

lish any transaction into the blockchain, a microgrid

must create a block and add it to the blockchain. The

techniques used to create and add blocks are deﬁned

in the security pool.

creates the following bid Bid

(id

bid

||price

bid

). Then, it creates a block hav-

ing this bid as follows, ﬁrst it must get the last block

added to the blockchain, calculate the hash code of

the new block based on the previous one, add the bid

(the argument tx in Figure 9) and mine the block.

Hence, the security pool serves as a container of

security methods necessary to maintain reliable and

secure trading process between the distributed micro-

grids.

5 PERFORMANCE EVALUATION

Since the contribution of this paper is to add artiﬁ-

cial intelligence, coordination and security pools to

the control aspects of the logic of software applica-

tions, we focus on evaluating these aspects (Section

5.1). In Section 5.2, a comparison with related works

is conducted.

Figure 9: Some of the blockchain methods required in the

trading.

5.1 Discussions

Coordination Pool: In order to show the efﬁciency

of the coordination pool, let us consider an increasing

number of peers (i.e., microgrids) in the system and

see how the pool behaves.

Thanks to creating a predeﬁned coordination ma-

trix that speciﬁes the possible conﬁgurations, it is not

needed that peers exchange data with each other in

order to ﬁnd a valid system state. If each peer needs

to negotiate with each other peer in the system at run-

time, then there will be a continuously increasing co-

ordination time and data with the increasing number

of peers and number of conﬁgurations (see Figure 10).

Figure 10: Number of transactions with CM versus without

CM.

In case a coordination matrix is predeﬁned, for n

peers it is required to exchange 2 ∗ (n − 1) transac-

tion. However, in case peers need to ﬁnd a conﬁgura-

tion at run-time, the number of transaction can reach

(n − 1) + (n − 1) ∗ (n − 2). Let us assume that in av-

erage, a transaction takes 2 minutes. Figure 11 shows

the time taken to ﬁnd the proper conﬁguration in min-

utes.

A Software Framework for Context-aware Secure Intelligent Applications of Distributed Systems

119

Table 7: Comparison of our framework with other context-aware frameworks.

Coordination AI Security Reconﬁguration

(Sikder et al., 2019) - - X -

(Aid and Rassoul, 2017) - - X -

(Bucchiarone et al., 2017) - X - -

Our framework X X X X

In this table: Xmeans addressed, - means not addressed.

Figure 11: Time to ﬁnd proper conﬁguration when scaling

the system.

Figure 11 depicts that in our approach, the coordina-

tion time is very slightly increasing by the increasing

number of peers since they need only to parse the co-

ordination matrix, select possible conﬁgurations and

send it to the concerned peer.

Artiﬁcial Intelligence Pool: The artiﬁcial intelli-

gence pool (AIP) builds upon an enhanced expert sys-

tem. Adopting expert systems to take decision is very

efﬁcient thanks to its: (1) speed: automated rules re-

duce the complexity of work and ﬁnd decisions for

repetitive problems, (2) culmination: sum of diverse

human expertise, (3) clearness: it provides reasonable

explanations of the made decisions. However, despite

these assets, intelligence techniques that builds upon

a rule base have some drawbacks, mainly a mainte-

nance one. In fact, rules need to be updated from time

to time in order to upgrade the logic or to include new

knowledge. Further, it is sometimes complex to ex-

tend existing rules, a lot of effort might be required.

5.2 Comparison to Related Frameworks

We ﬁnish this section with a comparative study in

which we highlight the advantages of the proposed

framework with respect to other works. We summa-

rize the study in Table 7.

Compared with the existing context-awareness

frameworks, the proposed one provides the opportu-

nity to tackle numerous requirements of today’s smart

systems. Particularly, the framework allows to de-

velop security, coordination and intelligence logic.

Most of the works address one speciﬁc feature not a

combination of them. More importantly, reconﬁgura-

tion which a key enabler of system’s behavior adap-

tation is often not addressed. The frameworks pre-

sented in (Sikder et al., 2019) and (Aid and Rassoul,

2017) consider security but important requirements

such as reconﬁguration and coordination are not sat-

isﬁed. The framework reported in (Bucchiarone et al.,

2017) uses the artiﬁcial intelligence techniques to pro-

vide knowledge management, however it does not ad-

dress security, coordination, nor reconﬁguration.

6 CONCLUSIONS

This paper has introduced a new software framework

that overcomes the limitations of the existing frame-

works. The proposed framework enables the develop-

ment of context-aware, distributed, intelligent, and se-

cure software applications of reconﬁgurable systems.

The proposed framework introduces a solution

to the coordination of peers in a distributed system.

Hence, collaborative behavior can be achieved efﬁ-

ciently and seamlessly. It also introduces an enhanced

expert system to be the promoter of intelligence oper-

ations. In order to keep a trustful data exchange for

speciﬁc transactions, the framework includes a secu-

rity pool having the role of holding the security tech-

niques.

Thanks to the abstract and clear framework archi-

tecture, precisely the deﬁnition of the pools and lay-

ers, it becomes easy to arrange the requirements of

applications in a loosely coupled way. The separation

of the miscellaneous logic aspects into pools facili-

tates the applications development. More importantly,

the concepts of the framework are implemented with

Java programming language. The availability of the

software tool of the framework helps to increase the

developers productivity.

We have used the proposed framework to develop

the application of microgrid’s system. The case study

helped to show the suitability of the intelligent and

collaborative requirements development as well as the

secure data exchange in distributed systems.

In future works, we aim to apply the concepts of

the framework to more case studies also we plan to

ICSOFT 2021 - 16th International Conference on Software Technologies

120

improve the quality of code of the software tool. An-

other concern is to integrate more advanced analytic

techniques to cover more logic aspects.

REFERENCES

Aid, A. and Rassoul, I. (2017). Context-aware framework to

support situation-awareness for disaster management.

International Journal of Ad Hoc and Ubiquitous Com-

puting, 25(3):120–132.

Alhamid, M. F., Rawashdeh, M., Dong, H., Hossain, M. A.,

Alelaiwi, A., and El Saddik, A. (2016). Recam: a

collaborative context-aware framework for multime-

dia recommendations in an ambient intelligence envi-

ronment. Multimedia Systems, 22(5):587–601.

Anvaari, M., Cruzes, D. S., and Conradi, R. (2012). Smart

grid software applications as an ultra-large-scale sys-

tem: Challenges for evolution. In 2012 IEEE PES In-

novative Smart Grid Technologies (ISGT), pages 1–6.

IEEE.

Bucchiarone, A., Marconi, A., Pistore, M., and Raik, H.

(2017). A context-aware framework for dynamic com-

position of process fragments in the internet of ser-

vices. Journal of Internet Services and Applications,

8(1):6.

Dabbaghjamanesh, M., Wang, B., Mehraeen, S., Zhang, J.,

and Kavousi-Fard, A. (2019). Networked microgrid

security and privacy enhancement by the blockchain-

enabled internet of things approach. In 2019 IEEE

Green Technologies Conference (GreenTech), pages

1–5. IEEE.

El-Naily, N., Saad, S. M., and Mohamed, F. A. (2020).

Novel approach for optimum coordination of overcur-

rent relays to enhance microgrid earth fault protection

scheme. Sustainable Cities and Society, 54:102006.

Elgamal, M., Korovkin, N., Elmitwally, A., and Chen,

Z. (2020). Robust multi-agent system for efﬁ-

cient online energy management and security enforce-

ment in a grid-connected microgrid with hybrid re-

sources. IET Generation, Transmission & Distribu-

tion, 14(9):1726–1737.

Ferdowsi, F., Dabbaghjamanesh, M., Mehraeen, S., and

Rastegar, M. (2019). Optimal scheduling of reconﬁg-

urable hybrid ac/dc microgrid under dlr security con-

straint. In 2019 IEEE Green Technologies Conference

(GreenTech), pages 1–5. IEEE.

Fkaier., S., Khalgui., M., and Frey., G. (2020a). Hybrid

context-awareness modelling and reasoning approach

for microgrid’s intelligent control. In Proceedings of

the 15th International Conference on Software Tech-

nologies - Volume 1: ICSOFT,, pages 116–127. IN-

STICC, SciTePress.

Fkaier., S., Khalgui., M., and Frey., G. (2020b). Meta-

model for control applications of microgrids. In 2020

6th IEEE International Energy Conference (ENERGY-

Con), pages 945–950. IEEE.

Fkaier, S., Romdhani, M., Khalgui, M., and Frey, G.

(2016a). Enabling reconﬁguration of adaptive control

systems using real-time context-aware framework. In

2016 IEEE/ACS 13th International Conference of

Computer Systems and Applications (AICCSA), pages

1–8. IEEE.

Fkaier, S., Romdhani, M., Khalgui, M., and Frey,

G. (2016b). R2tca: New tool for develop-

ing reconﬁgurable real-time context-aware frame-

work—application to baggage handling systems. In

Proc. Int. Conf. Mobile Ubiquitous Comput., Syst.,

Services Technol.(UBICOMM), pages 113–119.

Fkaier, S., Romdhani, M., Khalgui, M., and Frey, G.

(2017). Context-awareness meta-model for recon-

ﬁgurable control systems. In Proceedings of the

12th International Conference on Evaluation of Novel

Approaches to Software Engineering - Volume 1:

ENASE,, pages 226–234. INSTICC, SciTePress.

Ghosh, P., Eisele, S., Dubey, A., Metelko, M., Madari,

I., Volgyesi, P., and Karsai, G. (2020). Designing

a decentralized fault-tolerant software framework for

smart grids and its applications. Journal of Systems

Architecture, 109:101759.

Henri, G., Levent, T., Halev, A., Alami, R., and Cordier, P.

(2020). pymgrid: An open-source python microgrid

simulator for applied artiﬁcial intelligence research.

arXiv preprint arXiv:2011.08004.

Hosseinimoghadam, S. M. S., Dashtdar, M., Dashtdar, M.,

and Roghanian, H. (2020). Security control of is-

landed micro-grid based on adaptive neuro-fuzzy in-

ference system. Scientiﬁc Bulletin”: Series C Electri-

cal Engineering and Computer Science, (1):189–204.

Hubana, T. (2020). Artiﬁcial intelligence based station

protection concept for medium voltage microgrids.

In 2020 19th International Symposium INFOTEH-

JAHORINA (INFOTEH), pages 1–6. IEEE.

Khan, S., Paul, D., Momtahan, P., and Aloqaily, M. (2018).

Artiﬁcial intelligence framework for smart city mi-

crogrids: State of the art, challenges, and opportuni-

ties. In 2018 Third International Conference on Fog

and Mobile Edge Computing (FMEC), pages 283–

288. IEEE.

Mbungu, N. T., Naidoo, R. M., Bansal, R. C., and Vahidi-

nasab, V. (2019). Overview of the optimal smart en-

ergy coordination for microgrid applications. IEEE

Access, 7:163063–163084.

Sikder, A. K., Aksu, H., and Uluagac, A. S. (2019). A

context-aware framework for detecting sensor-based

threats on smart devices. IEEE Transactions on Mo-

bile Computing, 19(2):245–261.

Tang, L., Yu, Z., Wang, H., Zhou, X., and Duan, Z. (2014).

Methodology and tools for pervasive application de-

velopment. International Journal of Distributed Sen-

sor Networks, 10(4):516432.

Wu, Y., Guerrero, J. M., and Wu, Y. (2019). Distributed co-

ordination control for suppressing circulating current

in parallel inverters of islanded microgrid. IET Gen-

eration, Transmission & Distribution, 13(7):968–975.

A Software Framework for Context-aware Secure Intelligent Applications of Distributed Systems

121