AI-DRIVEN PERFORMANCE TESTING FRAMEWORK FOR MOBILE APPLICATIONS

International Journal of Software Engineering & Applications (IJSEA), Vol.15, No.6, November 2024
DOI: 10.5121/ijsea.2024.15601 1
TOWARDS A GENERIC SOFTWARE ARCHITECTURE
FOR IOT SYSTEMS
Yousef Abuseta
Department of Internet Technologies, IT Faculty, University of Tripoli, Tripoli, Libya
ABSTRACT
The complexity of IoT, owing to the inherent distributed and dynamic nature of such systems, brings more
challenges to the software development process. A vast number of devices with different communication
protocols and data formats is involved and needs to be connected and exchange data with each other in a
seamless manner. Traditional software architectures fall short of addressing the requirements of IoT
systems and, therefore, a new approach to software architecture is required. This paper presents an
attempt to lay out the foundation for a quality attribute driven software architecture for the development of
IoT systems. This architecture accommodates the appropriate architectural styles and design patterns
necessary for the development of a robust IoT system. These include edge computing, microservices and
event driven architectures. The proposed architecture treats IoT systems as autonomic systems which
require a closed control loop to regulate and orchestrate the operational aspect of the IoT system.
KEYWORDS
IoT software architecture, Quality Attributes, Architectural Styles, Edge Computing, Cloud Computing.
1. INTRODUCTION
The Internet of Things (IoT) is the natural evolution of the Internet revolution that serves as a
global architecture interconnecting machines and humans. The IoT paradigm revolves around
designing software systems out of uniquely addressable things provided by sensors, actuators and
processing capabilities to communicate and exchange data to accomplish a specific task [1]. This
kind of connectivity enables a large number of applications in different domains such as
healthcare, transportation, manufacturing, smart homes, etc. [2]. IoT systems are composed of
heterogenous devices and applications in addition to communication networks for connecting
them. It is estimated that by 2025, there will be over 75 billion connected devices worldwide [3],
leading to more interactive and smarter environments.
Despite being widely adopted and highly promising, IoT systems still pose crucial challenges
owing to the integrated heterogeneous software and hardware resources, communication
protocols as well as data from various resources. A more challenging issue is the uncertainty that
stems from the dynamic nature of the environment in which the IoT applications operate. Such a
dynamicity takes the form of important events which include a device joining/leaving,
introducing a new service or a change in system requirements.
These challenges may be addressed by a well-designed software architecture that accommodates
the necessary components and appropriate design decisions. Therefore, designing an architecture
that exhibits the qualities of supporting the development of a scalable, reliable, maintainable and
extensible system is of paramount importance to successful IoT systems.

2
A great deal of experiences of software development for IoT systems have been reported in
literature. However, most proposed architectures for these systems have not emphasized enough
the importance of the rationale for the choice of software components and how they relate to each
other with regard to devising a robust IoT software architecture.
In this paper, we attempt to design a generic software architecture that supports the realization of
the necessary quality attributes that are crucial for developing resilient IoT applications. This
architecture consists of a set of appropriate structures, the relations between them as well as any
constraints that control these relations. Also, the architectural styles, design patterns employed
for achieving such an architecture are proposed here. The rest of this paper is organized as
follows. Section 2 reviews related topics and concepts to the work conducted. In this paper
Section 3 discusses the software architecture for IoT systems its characteristics and the important
architectural and design patterns. Section 4 presents the proposed architecture. Section 5
discusses some related works. The paper is concluded in section 6 with some suggestions for
future work.
2. BACKGROUND
2.1. A High Level IoT Layered Architecture
This section presents the IoT architecture layers along with the building blocks that constitute
almost any IoT system. The IoT is built around the idea of connecting physical devices that are
capable of exchanging information and can be monitored and controlled with an external system,
usually an IoT platform. One of the main ideas of the IoT systems is the interaction between
relevant users and physical devices, where sensors provide information to the user about the
environmental conditions, and users can perform desired actions via actuators over other physical
devices based on that information. Another key point of IoT systems is the possibility of the
development of smart applications for specific domains. Examples of tasks such applications may
carry out include data analytics, prediction, monitoring and user notification [4].
It is worth mentioning that there is no single agreed upon IoT architecture. Nevertheless, the key
building blocks of any IoT system are almost the same which include smart things, networks and
gateways, IoT middleware and applications.
These building blocks constitute the backbone of IoT systems on which an effective layered
architecture can be developed. Such layers are described below and shown in Figure 1.
The Sensing layer. This layer represents the IoT system’s point of contact with the physical
world. Its main task is to translate physical information into data streams, or vice versa. The
primary component of this layer is the physical device or thing. Physical devices include personal
objects such as smart watches, and mobile phones. They also include objects in our environment
(e.g. home, vehicle, etc.) and industry (e.g. machines, robots, motors). Devices are equipped with
sensors and actuators. Sensors transmit the captured contextual environmental information using
electrical signals to devices, to which they are attached.
This connection can be established either by wire or wirelessly [5]. Sensors may come in various
forms such as motion sensors, location sensors, and temperature sensors or embedded sensors in
industrial machines to gauge the machine’s state [6]. An Actuator, on the other hand, is a
hardware component which is used to apply necessary actions to the physical environment. It
receives commands from its connected device and translates the received signals into some kind
of physical actions [5].

3
The Networking layer. The main objective of this layer is to seamlessly connect IoT devices in
the sensing layer with upper layers. It often contains a number of local gateways that can
establish a connection with IoT devices using several network technologies such as Wi-Fi,
Bluetooth, Zigbee and cellular networks. Data received from IoT devices is then converted from
the device's protocol and forwarded to the Internet. An IoT gateway works as a bridge between
the sensing layer and the Internet ensuring that data collected by IoT devices find its way to the
Internet infrastructure. In addition to forwarding, gateways may also be tasked with performing
some important operations, such as data aggregation and protocol translation [6].
The Middleware layer. The middleware layer is one of the primary components of the IoT
platform as it serves as an integration layer for heterogeneous IoT devices and ICT systems in the
platform. It facilitates application development by providing a uniform way of interaction with
different hardware, protocols, and data formats. With a middleware layer in place, developers are
relieved from the underlying hardware or network protocols and thus they concentrate fully on
the application business logic. The main tasks of the middleware include: 1) Receiving data from
the connected devices, 2) Processing the received data, 3) Providing the received data to
connected Applications, and 4) Controlling devices [5]. Typically, an IoT Middleware can be
accessed using APIs, for example, HTTP-based REST APIs.
The Application layer. An IoT application is a software component that carries out some tasks
by reading data collected by IoT devices as well as and controlling these devices. IoT
applications are used by end users to achieve data insights, control IoT devices and in general
interact with the physical world using computers and smartphones. Applications can be built on
top of the IoT platform or integrate with it through APIs. An IoT application can vary from a
simple analytics dashboard where all IoT events are shown, to a complex application that
processes data and extract information from IoT events and combine it with other information
sources [6].
Figure 1. A high level IoT layered Architecture [7]
2.2. IoT Messaging Protocols
According to the IoT architecture depicted in Figure 1, multiple layers of communication can be
recognized. In the first layer, communication occurs from IoT device to gateway, using protocols

4
and radio technologies like Zigbee, Bluetooth, 802.15.4 and Wi-Fi. In the second layer,
communication takes place from gateway to middleware. These include HTTP (REST), MQTT,
AMQP, XMPP and CoAP. Event-based protocols such as MQTT are preferred, to prevent
unnecessary costly polling. REST and MQTT are the most used protocols [8]. And finally,
communication is conducted from middleware to applications, often using REST APIs [9].
The following are the most common protocols that are of great relevance to IoT systems. The
discussion is based on a survey presented in [4].
 The Hypertext Transfer Protocol (HTTP): it is the primary the communications protocol
forweb based applications. It is a generic stateless protocol which adopts the client–server
model. It can be employed for many different tasks in addition to its use in the Web. In the
IoT paradigm, HTTP can be used to exchange data with a publish or polling model.
 The Constrained Application Protocol (CoAP): it is a special protocol that is primely used in
constrained environments (e.g., low-power devices). It is specifically designed for machine to
machine (M2M) applications such as smart energy and building automation. Similar to
HTTP, CoAP adopts the client–server model and it can be easily configured to communicate
with HTTP endpoints.
 Message Queuing Telemetry Transport (MQTT): it is a lightweight publisher-subscriber
protocol. MQTT is ideal for M2M and IoT contexts where a small code footprint is required
and network bandwidth is at a premium.
3. SOFTWARE ARCHITECTURE FOR IOT SYSTEMS
3.1. Quality Attribute Driven Architecture
In [10], the software architecture is defined as: "The software architecture of a system is the set of
structures needed to reason about the system. These structures comprise software elements,
relations among them, and properties of both". Another definition in [11], defines software
architecture as follows: "software architecture reflects fundamental organization of a software
system, its components and their relationships, as well as relations of a software system with its
environment". The key point in designing a resilient software architecture is to achieve loose
coupling by reducing dependencies between involved structures. This simplifies software
evolution as changes are often local and do not propagate through the architecture which highly
contributes to the reduction of the cost and complexity of software maintenance [11]. In modern
software systems, the software architecture is driven by quality attributes. The quality attributes
are often referred to interchangeably as the non-functional requirements of the system under
consideration. Such requirements are concerned with how well the system performs as opposed
to the functional requirements which are related to the core functions the system in question must
provide. In [12] the following definition of quality attributes was introduced: "Quality attributes
are properties of a software system and a subset of its non-functional requirements. Like other
requirements, they should be measurable and testable. Software quality attributes are
benchmarks that describe the software system’s quality and measure the fitness of the system". In
another definition by [10], the quality attribute is defined as: "A quality attribute (QA) is a
measurable or testable property of a system that is used to indicate how well the system satisfies
the needs of its stakeholders beyond the basic function of the system". Examples of quality
attributes include scalability, portability, security, availability and performance.
The quality attributes serve as a support tool for the evaluation of proposed software architecture.
Various quality attributes are considered during the software design process. The satisfaction
level of quality attributes impacts the overall performance of the system in consideration. [11].

5
Authors of [12, 13] stressed the importance of considering quality attributes in the early phase of
eliciting software requirements. However, they stated that their testing and evaluation should be
carried out during the whole system life cycle.
Quality attributes can be classified into two categories: the first category is concerned with those
attributes that describe the system in question at runtime such as availability and performance
while the second includes those that describe some property of the development of the system,
such as modifiability, testability, or deployability [10].
A description of some important quality attributes is presented below.
Interoperability. This quality attribute can be defined as the ability of different technologies,
software components or applications to seamlessly communicate with each other and interpret
properly the exchanged data. In IoT, interoperability represents a serious issue and real challenge
due to the heterogeneity of IoT systems and the lack of reference standards. [14].
Performance. It is about time and the software system’s ability to meet timing requirements. In
the context of IoT, this quality attribute is crucial as most IoT applications need to meet some
desirable response time and throughput.
Modifiability. It is an attribute that enables a system to introduce a change. Changes may be
required to: 1) introduce new features or modify existing ones, 2) fix defects, increase security
and improve performance, 3) adopt new technology or protocols and 4) integrate different
systems that were never designed to work together. [10].
Energy Efficiency. It is about managing energy consumption efficiently. This attribute is highly
important as the majority of IoT devices, such as mobile phones and tablets, are energy
constrained. Also, cloud providers are increasingly concerned with the energy efficiency of their
server farms.
Scalability. It is about accommodating more of something Scalability is often needed to improve
the system performance. Two kinds of scalability can be defined: horizontal and vertical. The
horizontal scalability is concerned with accommodating more resources to logical entities, such
as adding a new server to a server farm, whereas the vertical scalability refers to adding more
resources to a physical entity, such as adding more memory to a computer.
Based on literature review conducted in [11], the following quality attributes were identified as
the most important for IoT architecture: scalability, security, interoperability, and performance.
3.2. Architectural Styles for IoT Systems
The following discussion revolves around the realization of the above discussed quality attributes
using some specific architectural styles and design patterns.
3.2.1. Edge computing
Over the last decade, the trend has been to use the centralized cloud computing infrastructure to
provide data processing and storage in large data centres. This is due to its elasticity and cost-
effectiveness. However, IoT systems pose significant challenges as many of these systems have
devices (things) at the network edge that generate a tremendous amount of data that needs to be
stored and processed using heavy resources. Current cloud computing is not efficient to process
very large data in a timely manner to satisfy the end user and system requirements. Any failure

6
by the cloud to comply with the timing requirement will likely have a negative impact on the
quality of service (QoS) and affect the IoT applications and network performance in general.
[15].
The edge computing has been proposed to address this issue. It is designed to bridge the gap
between data generation and data processing. By processing data close to the source of its
generation, edge computing significantly reduces latency, conserves bandwidth, and improves
response times. It ensures that only relevant, processed data is sent to the cloud, optimizing
bandwidth and processing power. It complements the cloud computing by offering a balanced
approach to data processing. Edge computing addresses latency and bandwidth constraints for
real-time applications, whereas cloud computing provides extensive computational power,
storage, and advanced analytics.
Examples of smart IoT applications that benefit from edge computing include smart homes, smart
vehicles, smart grid, smart cities, smart healthcare, etc. [15]. With regard to the quality attributes,
edge computing can contribute to the realization of performance, privacy, energy consumption
and bandwidth utilization. Promising features of edge computing include location awareness,
mobility support and being close to the user [16]. These features enable edge computing to play a
crucial role in addressing different IoT applications such as smart traffic monitoring, industrial
automation, virtual reality and smart home. [17].
3.2.2. Microservices Architecture
For the last few years, the traditional monolithic approach to the software architecture has fallen
out of favour. Its failure to meet the quality requirements of modern systems (e.g. scalability,
interoperability and modifiability) has paved the way for the modular architecture approach. This
approach has been widely embraced in many domains and become the driver for the development
of modern software systems. In this context, Microservices represent a promising architectural
model especially in domains where distributed and pervasive computing is highly important, as in
Internet of Things [18].
Microservices architecture is built on SOA concepts, focusing on refactoring large monolithic
applications into small, highly decoupled, independently deployable and scalable services. These
services carry out a single, well defined and domain-oriented task, and communicate among
themselves by means of a simple protocol, for instance the HTTP [18] [19].
Quality attributes such as availability, reliability, interoperability, latency, delay, are important
characteristics of any distributed system like IoT. Yet, Satisfying and addressing these quality
attributes is not a trivial task. The characteristics of microservices make it suitable for addressing
this issue at each layer of the IoT architecture [20].
3.2.3. Event Driven Architecture
Currently, the development of software systems has taken place in increasingly changing
business environments leading to the need for high flexibility to support rapid system adaptations
[21]. The real issues of concern in modern software systems development include dynamic
software requirements, technology change, a quick response to market, and systems with tightly
coupled modules. [22]. Therefore, companies have been searching for better approaches to
minimize problems in the development and maintenance of such systems, for instance, the high
cost of maintenance and complexity of the systems. Event driven architecture has been largely
embraced in the software development, emerging as a promising approach to modular
development aiming at supporting the required rapid adaptations of constantly evolving systems

7
[23]. The adoption of event driven architectures aims mainly to improve maintenance by
breaking down monolithic applications into independent modules that communicate via events.
In the context of IoT systems, the event driven architecture is primarily employed to minimize
the coupling between the different components engaged in this ecosystem. It eases the exchange
of data between IoT devices while making each one unaware of the others. This facilitates the
integration of a new component or device once a change has been required. It also helps in
reducing the energy consumption since devices are not constantly sending data to interested
parties but instead data are sent only when a significant event is triggered. This is crucial as most
IoT devices are resource constrained with regard to energy and capacity. In terms of quality
attributes, adopting the event driven architecture helps IoT systems realize the integrability and
modifiability. As a messaging protocol, this architectural style uses the publish-subscribe pattern.
4. PROPOSED IOT ARCHITECTURE
This section is dedicated to describe the proposed architecture for IoT systems. It includes a meta
model that serves as a high-level description of IoT systems. Such a meta model contains the
fundamental elements and concepts that form the IoT ecosystem that spread over the different
layers of the IoT architecture presented in Figure 1.
4.1. A Metamodel for IoT Systems
As a first crucial step towards the design of an effective IoT software architecture, a metamodel is
developed. This metamodel defines the main concepts and their relationships and serves as
common understanding of the IoT systems. A UML class diagram is employed here to document
this metamodel. Following the "divide and conquer" design principle, the metamodel is divided
into smaller metamodels, namely the edge computing, the cloud platform and the application.
4.1.1. The Edge Computing Metamodel
Figure 2 depicts the concepts that exist in this metamodel as well as the relationships between
them.

8
Figure 2. Edge computing metamodel.
These concepts are presented and described as follows:
Processing Node. this concept plays the role of the middleware layer of the edge computing. It is
responsible for processing data generated and transmitted by a physical device through a
gateway. In addition, It also provides a unified APIs to be used by IoT application developers.
Application Enabler. this provides the necessary and required APIs for developing IoT
applications. Developers can use it to develop
Rule engine. This contains a set of rules used by the processing node to control and regulate the
operational aspect of IoT applications.
Data analytics. It is responsible for analyzing data received from physical devices using some
specific analytical tool in order to infer some insights.
Device manager. This is responsible for managing IoT devices through neutral and unified
interfaces.
Gateway. The gateway is the communications hub for the IoT network. It performs important
functions, such as data aggregation, translation between communication protocols, and data
processing before the gateway forwards the data to the cloud or on-premises network.
Protocol translator. Devices send data via non-standard protocols, so they transform it into a
common format that the cloud platform can analyze and comprehend with ease.
Thing. The "thing" in the IoT vision is very wide concept which includes many forms of physical
elements. These include personal objects such as smart phones, tablets and digital cameras as
well as elements in our environments (e.g. room, building, vehicle) and industries (e.g.,

9
machines, robots). Each thing has attributes that describe it as well as a state that is relevant from
a user or an application perspective.
Interesting property. It is an observable property of a physical entity. The temperature of a
human body is an example of this concept. The property can be single or composed of a number
of properties.
Sensor. It is a type of resource that has the ability to detect changes in the environment.
Thermometer, humidity and motion sensors are some examples of sensor.
Actuator. It is a resource that has the ability to make changes in the environment through an
action. For example, turn the lights on or the air conditioning system in response to some certain
data obtained.
Device. It is an entity that hosts sensor and actuator resources and gives them the ability to
communicate with other entities. Examples include mobile phones and personal computers.
4.1.2. The Cloud Metamodel
As stated earlier in this paper, the edge computing isn’t a replacement for cloud computing.
Instead, they complement each other and each one has its own use cases and its viable
contribution to the delivery of a resilient IoT system. Cloud computing should be used when
there is need for heavy data processing and big data need to be stored and analysed using
complex data analytics algorithms.
In the proposed architecture, the cloud computing is used as a global controller whose main
mission is to regulate the operational aspect of the IoT system in question. This controller
accomplishes its task through a set of processing nodes distributed throughout the edge
computing platform. The fundamental concepts and elements of cloud metamodel include global
controller, rule engine, data analytics, data storage and processing node. Such concepts were
described in the edge computing metamodel. They have similar tasks except that they operate and
coordinate to accomplish the work of the global controller.
Figure 3. Cloud Metamodel.

10
4.1.3. The Application Metamodel
An IoT application is composed of a set of services to be used by end users. Such services reflect
the software requirements of the system and are used to address business values for
organizations. Services are provided here to accomplish a number of tasks such as monitoring
and controlling devices at the sensing layer, visualizing insights derived from processed data and
building an appropriate dashboard to monitor and observe the performance of the whole system.
Also, the application user should be brought into the loop by the provision of a user management
system that enables the user to create an account which can be used later by the system to
accomplish important task such as sending notifications and personalising the user interface of
the IoT application. The fundamental concepts and elements of the application metamodel
include device controller, user, user manager, subscriber, dashboard and notifications.
These concepts are described below.
Dashboard. It is the user interface in the IoT platform through which the user monitors and
controls connected devices at the sensing layer. It also enables the user to monitor and visualise
the application performance and observe any significant changes in the operational aspect either
at the device level or the whole system level.
Notification Manager. Notifications are sent out to interested users who have subscribed to be
notified of some significant events. Therefore, this concept provides the necessary APIs for an
interested user to subscribe for a specific event.
User. The user is a broad concept that represents the application beneficiary which varies
depending on the application domain being developed. An account is created for each user which
contains the appropriate information needed for sending out customized notifications and
personalising the user interface.
User Manager. It is responsible for providing the necessary APIs for creating and managing user
accounts.
Subscriber. The subscriber is any interested user that registers with the notification manager to
receive notifications of some significant events.
Device controller. It is used to offer the necessary APIs for controlling a set of available physical
devices. Remotely turning on lights at home is an example of using this controller.

11
Figure 4. IoT application layer metamodel.
4.2. IoT application business logic
In addition to the user interface with which the end user interacts with the IoT system, there is the
business logic which constitutes the components of the core system as well as the orchestration
system. The latter is responsible for controlling and regulating the operational aspect of the core
system.
4.2.1. The Core System
In order to model the core system, this paper identifies a number of concepts which are described
as follows.
Domain. The domain here represents the system in question which comprises a set of tasks.
Examples of domain include the healthcare, smart agriculture, home automation, smart metering
and smart building.
Task. A task is a high-level goal that is addressed in order to realize the overall system
requirements. Each task, in turn, encompasses a set of services responsible for achieving that
task. A task in a healthcare system is, for example, monitor remotely blood sugar level for a
diabetic patient.
Service. A service is an abstract representation of a software or hardware entity that plays a role
in addressing the ultimate goal of the task at hand. These services are then mapped to software
components such as RESTful microservices. A central heating thermostat is an example of
service. In our approach, each device or a set of devices involved in IoT applications can be
treated as a service.
Business process. A Business process represents a specific ordering of services and activities
across time and place to realize a high-level business goal. The services of a particular task
interact and coordinate with each other to address the purpose of that task. Such coordination is
encapsulated in an entity called a business process. A Business process might consist of only one
service. However, a useful business process is often composed of a number of services.

12
4.2.2.IoT Orchestration System
To orchestrate the work of the different components of the core IoT system, an appropriate
orchestration system must be put in place. The complexity induced by the massive number of
connected IoT devices rapidly outperforms the human capability to handle it. As a consequence,
the autonomy characteristic is crucial with regard to the design of the orchestration system.
Autonomic computing is a field that studies how software systems can
achieve desirable behaviours without the human intervention [24]. Such behaviours take the
form of self-healing, self-configuring, self-protecting and self-optimising [25].
To design the orchestration system, the IBM autonomic computing model [26] is adopted here.
This model uses the MAPE-K feedback control loop which stands for Monitor, Analyse, Plan,
Execute and Knowledge. The MAPE-K feedback control loop performs MAPE phases using a
shared Knowledge base. The Monitor (M) phase acquires data from the system and its
environment through a set of sensors. The Analyse (A) phase involves activities such as filtering
and transforming of the monitored data and then analysing the clean data. In the Plan (P) phase, a
set of actions are composed taking into account the outcome of the analysis as well as the
knowledge of the system. Finally, the Execute (E) of the planned actions should be performed,
often through a dedicated number of actuators. A MAPE-K loop stores the Knowledge (K)
required for decision making in the Knowledge Base (KB).
The orchestration system is distributed where the local orchestration is performed at the edge
computing while the global orchestration is conducted at the cloud platform.
The local orchestration is composed of a set of MAPE-K loops where each loop is in charge of
orchestrating a specific region at the edge computing platform.
The local orchestration system is offered as a MAPEaaService in the edge computing platform
while the global orchestration is provided as APaaService in the cloud platform since it contains
only, in addition to the knowledge component, the analysis and planning activities.
The orchestration system is offered in two modes: centralized and decentralized. In the
centralized mode, a central control loop is deployed either on the edge computing or cloud
platform (depends on the application scale) to orchestrate the operating of the different control
loops that are located at the same level. The master-slave model is employed here to establish
such a relationship as shown in Figure 5.
In the decentralized mode, a set of MAPE loops of the same level is coordinated to accomplish
the four activities (monitoring, analysis, planning and execution). For instance, the execute
components of involved control loops communicate and coordinate to perform the corrective
actions in the absence of a central controller. Self-organising systems are a popular example of
systems operating and functioning in the decentralised mode.

13
Figure 5. The centralized mode of the orchestration system.
4.3. Proposed Architectural and Design Patterns
In this section, two important architectural and design patterns are presented to demonstrate the
realization of some of the important quality attributes highlighted earlier in this paper. A
description of these patterns is presented as follows.
 The Master Slave pattern: it is a design model in computing where one central entity, called
the master, controls and directs the operation of multiple subordinate entities, known as
slaves. In the proposed architecture, this pattern was employed to establish the architectural
relation between the global orchestration system and a set of local orchestration systems. This
architectural pattern enables parallel processing and load balancing, thereby improving
system performance and scalability.
 The Publisher-Subscriber pattern: it is a messaging pattern where publishers send
messages without knowing who will receive them, and subscribers receive messages without
knowing who sent them. The communication between publishers and subscribers is
facilitated by a third party known as message broker. This decoupling allows for more
flexible and scalable systems, as components can evolve independently.
In the proposed architecture, this pattern may be used to model the notification system where
interested users subscribe to interesting events categorized as topics in the broker. The
publisher here can be any event source such as a senor of a specific device or a monitor in the
MAPE-K loop that publishes a new data to the knowledge base after which the analyse
component gets notified. These two scenarios are depicted in Figures 6 and 7.

14
Figure 6. Publisher-Subscriber patten at the IoT application layer.
Figure 7. Publisher-Subscriber patten at the local orchestration System.
The Publisher-Subscriber model is a powerful pattern that is adopted in modern software
development to address many challenges. By decoupling components, it enables scalable,
maintainable, and flexible systems suitable for IoT applications.
5. RELATED WORK
This section presents some works related to the research conducted in this paper. Therefore,
works on software architecture, architectural patterns and design patterns for IoT systems are
considered.
Stojanov et al. [11] presented discussion of specific quality attributes during the design of layered
sensor-based IoT system for monitoring the industrial environmental conditions in manufacturing
sector. They emphasized on the importance of considering quality attributes of software
architecture in order to ensure proper functioning of the whole systems and minimizing risks and
costs in the system life cycle. Kim [28] proposed a quality model of IoT applications based on the
following characteristics of IoT systems: participation of hardware devices, collaboration of
software and hardware components, mobility and connectivity of devices, monitoring of devices,
and limited resources. Temkar et al. [29] conducted a study to evaluate the quality of a system for
agriculture field monitoring using Wireless Sensor Network and IoT. The system architecture
consists of components that enable data collection and monitoring, data processing, execution,
and feedback.
Guth et al. [7] proposed an abstract IoT reference architecture in an attempt to enable a uniform
terminology and to ease the comparison of existing platforms. Porruvecchio et al. [27] introduced
an IoT platform called CMC-IoT. Such a platform is based on the microservices architecture
aiming at providing tools for developing scalable and robust IoT applications.
In [30], The IoT-A project proposed an architectural reference model and a preliminary set of
buildings blocks to promote a fully interoperable and scalable vision of IoT. The foundation of
the IoT-A Reference Model is the IoT Domain Model, which introduces the main concepts of the
Internet of Things. Important concepts include Devices, IoT Services, and Virtual Entities (VE),
as well as the relations between these concepts. Qanbari et al. [31] proposed design patterns for

15
edge computing based applications. They include: 1) edge computing pattern to handle the
provision of all edge devices automatically; 2) source code deployment pattern for edge devices
to handle the deployment of the code to all devices connected to the IoT system; 3) edge
orchestration pattern to handle the automation of creation, monitoring, and deployment of
resources in the IoT environment.
Despite the significant contribution made by the papers presented above, most of them
overlooked the importance of building software architectures driven by quality attributes where
the focus is on the rationale choice of software components and the way they relate to each other.
6. CONCLUSIONS
This paper was an attempt to lay out the foundation for a generic software architecture for IoT
platform. The work presented here emphasizes the importance of a well designed software
architecture driven by some related quality attributes for the delivery of a resilient and future
proof IoT systems. Significant architectural styles and design patterns were proposed to address
these quality attributes. These include edge computing, microservices, event driven architectures,
and the publish-subscribe pattern (based on the observer design pattern). The emphasis was also
on treating IoT systems as autonomic systems which require a closed control loop to regulate and
orchestrate the operational aspect of the IoT system. The MAPE-K control loop proposed by the
IBM autonomic computing model was adopted in this paper. Future works may include a more
detailed and technical description of each concept and component involved in the metamodel of
the proposed architecture. Also, a dedicated case study should be conducted to clarify and
demonstrate the application of proposed architectural and design patterns.
REFERENCES
[1] Rebeca C. Motta and Káthia M. de Oliveira and Guilherme H. Travassos, (2023) “An Evidence-based
Roadmap for IoT Software Systems Engineering”, Journal of Systems and Software, Vol. 201.
[2] Hassaan Siddiqui, Ferhat Khendek, Maria Toeroe ( 2023) Microservices based architectures for IoT systems -
State-of-the-art review, Internet of Things, vol. 23.
[3] Li, C., Wang, J., Wang, S., & Zhang, Y. (2024). A review of IOT applications in healthcare.
Neurocomputing, 565, 127017.
[4] Domínguez-Bolaño, T., Campos, O., Barral, V., Escudero, C. J., & García-Naya, J. A. (2022). An
overview of IOT architectures, technologies, and existing open-source projects. Internet of Things, 20,
100626.
[5] Guth, J., Breitenbücher, U., Falkenthal, M., Fremantle, P., Kopp, O., Leymann, F., & Reinfurt, L.
(2017). A detailed analysis of IoT platform architectures: Concepts, similarities, and differences.
Internet of Things, 81–101.
[6] Court, M.A. de la (2020). Design of a model-driven platform for IoT event stream processing. M.SC
thesis. University of Twente, Netherlands.
[7] Guth, J., Breitenbücher, U., Falkenthal, M., Leymann, F. & Reinfurt, L. (2016). Comparison of IoT
platform architectures: A field study based on a reference architecture. 2016 Cloudification of the
Internet of Things (CIoT), Paris, France, pp. 1-6.
[8] Dizdarević, J., Carpio, F., Jukan, A., & Masip-Bruin, X. (2019). A survey of communication protocols
for internet of things and related challenges of fog and cloud computing integration. ACM Computing
Surveys, 51(6), 1–29.
[9] Hejazi, H., Rajab, H., Cinkler, T., & Lengyel, L. (2018). Survey of platforms for massive IoT. 2018
IEEE International Conference on Future IoT Technologies (Future IoT).
[10] Bass, L., Clements, P., & Kazman, R. (2022). Software architecture in practice. Addison-Wesley.
[11] Stojanov, Z & Dobrilovic, D. (2021). Software Architecture Quality Attributes of a Layered Sensor-
Based IoT System. Workshop Information Technologies: Algorithms, Models, Systems (ITAMS),
Russia.
[12] Ingeno, J. (2018). Software architect’s handbook. Packt Publishing. Birmingham, UK.

16
[13] Sommerville, I. (2015) Software Engineering, 10 ed., Addison Wesley, Boston, MA, USA.
[14] González-Usach, R.; Yacchirema-Vargas, DC.; Julián-Seguí, M.; Palau Salvador, CE. (2019).
Interoperability in IoT. Handbook of Research on Big Data and the IoT. 149-173.
[15] Laroui, M., Nour, B., Moungla, H., Cherif, M. A., Afifi, H., & Guizani, M. (2021). Edge and fog
computing for IOT: A survey on current research activities & future directions. Computer
Communications, 180, 210–231.
[16] Kozik, R., Choraś, M., Ficco, M., & Palmieri, F. (2018). A scalable distributed machine learning
approach for attack detection in edge computing environments. Journal of Parallel and Distributed
Computing, 119, 18–26.
[17] Khan, W. Z., Ahmed, E., Hakak, S., Yaqoob, I., & Ahmed, A. (2019). Edge computing: A survey.
Future Generation Computer Systems, 97, 219–235.
[18] Newman, S., & O’Brien, T. (2022). Building microservices. 1st ed. O’Reilly Media, Inc.,
[19] Fowler, M & Lewis, J (2014). “Microservices,”. [Online]. Available:
http://martinfowler.com/articles/microservices.html
[20] Razzaque, M. A., Milojevic-Jevric, M., Palade, A., & Clarke, S. (2016). Middleware for internet of
things: A survey. IEEE Internet of Things Journal, 3(1), 70–95.
[21] Oliveira, A., Bischoff, V., Gonçales, L. J., Farias, K., & Segalotto, M. (2018). BRCode: An
interpretive model-driven engineering approach for enterprise applications. Computers in Industry, 96,
86–97.
[22] Rubert, M., & Farias, K. (2022). On the effects of continuous delivery on code quality: A case study
in industry. Computer Standards & Interfaces, 81, 103588.
https://doi.org/10.1016/j.csi.2021.103588
[23] Lazzari, L., & Farias, K. (2023). Uncovering the Hidden Potential of Event-Driven Architecture: A
Research Agenda. ArXiv, abs/2308.05270.
[24] López-Huguet, S., Pérez, A., Calatrava, A., de Alfonso, C., Caballer, M., Moltó, G., & Blanquer, I.
(2019). A self-managed mesos cluster for data analytics with QoS guarantees. Future Generation
Computer Systems, 96, 449–461.
[25] Desai, F., Chowdhury, D., Kaur, R., Peeters, M., Arya, R. C., Wander, G. S., Gill, S. S., & Buyya, R.
(2022). HealthCloud: A system for monitoring health status of heart patients using machine learning
and cloud computing. Internet of Things, 17, 100485.
[26] ] IBM (2005). An architectural blueprint for autonomic computing, IBM, Tech. Rep.
[27] Porruvecchio, G., Romanino, A., Casari, C., & Sanna, R. (2021). A microservice-based platform for
IOT application development. 2021 IEEE 12th Annual Ubiquitous Computing, Electronics &
Mobile Communication Conference (UEMCON), 29, 0332–0336.
[28] Kim, M. (2016). A quality model for evaluating IOT applications. International Journal of Computer
and Electrical Engineering, 8(1), 66–76.
[29] Et. al., R. T. (2021). Quality assurance of IoT based systems using analytic hierarchy process. Turkish
Journal of Computer and Mathematics Education (TURCOMAT), 12(10).
[30] Bassi, A., Bauer, M., Fiedler, M., Kramp, T., Van Kranenburg, R., Lange, S., & Meissner, S. (2016).
Enabling things to talk designing IOT solutions with the IOT Architectural Reference Model. Springer
Berlin Springer.
[31] Qanbari, S., Pezeshki, S., Raisi, R., Mahdizadeh, S., Rahimzadeh, R., Behinaein, N., Mahmoudi, F., Ayoubzadeh,
S., Fazlali, P., Roshani, K., Yaghini, A., Amiri, M., Farivarmoheb, A., Zamani, A., & Dustdar, S. (2016). IOT
design patterns: Computational constructs to design, build and engineer edge applications. 2016 IEEE First
International Conference on Internet-of-Things Design and Implementation (IoTDI), 277–282.
AUTHOR
Yousef Abusetta is an associative professor at the University of Tripoli, Libya. He joined the Internet
Technologies department of the IT faculty in 2022. He is currently in charge of the supervision of the
postgraduate studies program at the Internet Technologies department. His research interests include
autonomic systems, IoT, software architecture and design patterns.

AI-DRIVEN PERFORMANCE TESTING FRAMEWORK FOR MOBILE APPLICATIONS

Recommended

More Related Content

Similar to AI-DRIVEN PERFORMANCE TESTING FRAMEWORK FOR MOBILE APPLICATIONS (20)

Recently uploaded (20)

AI-DRIVEN PERFORMANCE TESTING FRAMEWORK FOR MOBILE APPLICATIONS