IAES International Journal of Artificial Intelligence (IJ-AI)

Vol. 14, No. 5, October 2025, pp. 4308~4318

ISSN: 2252-8938, DOI: 10.11591/ijai.v14.i5.pp4308-4318  4308

Unveiling critical features for failure prediction in green

internet of things applications

Ouiam Khattach1, Omar Moussaoui1, Mohammed Hassine2

1
Laboratory of Mathematics, Signal and Image Processing, and Computing Research (MATSI), Higher School of Technology (ESTO),
Mohammed First University, Oujda, Morocco
2
Tisalabs Limited, Cork, Ireland

Article Info ABSTRACT

Article history: The rapid growth of the green internet of things (GIoT) in recent years
signifies a transformative shift in internet of things (IoT) solution
Received Jan 17, 2025 development. This evolution is driven by technological advancements,
Revised Jun 29, 2025 heightened environmental awareness, and a global imperative to combat
Accepted Jul 13, 2025 climate change. Ensuring the reliability of GIoT applications is crucial for
their success. This study identifies critical features for predicting IoT device
failures, enabling early detection and intervention. Using datasets from
Keywords: industry, energy, and agriculture sectors, we employ a feature selection
strategy to analyze extensive data from diverse GIoT deployments. Our
Failure prediction analysis identifies significant features and integrates key insights from
Features selection existing literature. Our findings support enhanced predictive maintenance
Green internet of things strategies, reduced downtime, and improved overall performance of
Internet of things sustainable IoT solutions.
Predictive maintenance
This is an open access article under the CC BY-SA license.

Corresponding Author:
Ouiam Khattach
Laboratory of Mathematics, Signal and Image Processing, and Computing Research (MATSI)
Higher School of Technology (ESTO), Mohammed First University
Oujda, Morocco
Email: [email protected]

1. INTRODUCTION
In recent years, the internet of things (IoT) has seen substantial growth and integration across various
facets of daily life [1]. As IoT continues to expand into numerous industries, the reliability and performance of
IoT devices have become increasingly crucial. The IoT concept aims to use microcontrollers, transceivers, and
protocol stacks for connectivity and combine everyday objects with the internet. By connecting everyday
objects and devices to the digital world, the IoT has an extensive number of uses in daily life, from precision
agriculture that optimizes farming practices to industrial IoT that revolutionizes manufacturing and production
processes. IoT has created an interconnected network of smart devices, embedded sensors, and cloud
computing, revolutionizing a variety of industries, including smart traffic [2], [3], healthcare [4], [5],
agriculture [6], [7], and Industry 4.0. The IoT assists the current energy sector [8]. The integration of IoT
technology with green energy applications marks a transformative leap toward a more sustainable future of
renewable energy sources [9]. IoT technology facilitates connecting all the components of energy production
and consumption, getting insight into the processes, and giving actual control at every stage of the energy
flow, from exploitation to delivery to end users.
Predicting failures in green internet of things (GioT) application offers several advantages. First, it
enhances system resilience by enabling rapid responses to potential disruptions, ensuring continuous
operation in critical applications like smart grids, environmental monitoring, and healthcare systems, where

Journal homepage: http://ijai.iaescore.com

Int J Artif Intell ISSN: 2252-8938  4309

downtime can have severe consequences. Early intervention improves system stability, allowing IoT
infrastructure to recover quickly from potential failures. Secondly, it facilitates proactive maintenance
strategies, reducing energy consumption and preventing unnecessary downtime [10]. Additionally, failure
prediction extends the lifecycle of IoT devices, decreasing the need for frequent replacements and supporting
sustainability efforts. By ascertaining key factors contributing to failures, researchers can further optimize
design and manufacturing processes, resulting in more durable and efficient IoT solutions.
This study has been structured into six sections. Section 1 is the introductory phase, which presents
the evolution of IoT and addresses the associated challenges. Section 2, sheds light on the background of IoT
and GIoT. In section 3, the focus has been only on the literature review related to the green energy field.
Section 4 highlights the challenges and techniques involved in feature selection for optimizing energy
efficiency and resource consumption in GIoT. Results, challenges, and future research directions of the
modern day have been discussed in section 5. Section 6 is the concluding portion of the research article. The
articles have been extracted based on the highest number of citations over the past few years.

2. BACKGROUND
2.1. Internet of things architecture and components
The IoT is a global network infrastructure consisting of various connected devices that rely on
sensors. It operates through a four-step architecture, as illustrated in Figure 1. Each stage in this process is
interconnected, enabling data captured or processed at one stage to provide value to the next [5].
The following is a simplified representation of a typical IoT workflow. End devices: this phase
involves the deployment of IoT devices or sensors at various locations to collect data from the physical
environment. These devices consist of sensors, actuators, cameras, or other hardware that gather relevant
data. They are responsible for collecting and transmitting data to the following phase.
Data preprocessing: the data collected by IoT devices is preprocessed before being effectively
analyzed. This step includes data cleaning, filtering, and normalization to improve data reliability and accuracy.
The data is modified or improved to render it suitable for analysis. Sensors or other devices frequently send
back analog data, which need to be integrated and converted to digital format for further processing.
Data storage: after preprocessing, the combined and digitalized data needs to be properly stored in
an appropriate repository for further analysis. This phase involves selecting a suitable storage solution, such
as databases or data lakes while considering scalability, and data reliability. Additionally, the standardized
data is transferred to the selected data center or cloud infrastructure for effective and secure storage.
Data analysis: in this stage, a variety of approaches are applied to the stored IoT data to extract
relevant insights and knowledge. The data is examined for patterns, trends, correlations, and anomalies
applying techniques including statistical analysis, data mining, artificial intelligence, and machine learning.
The objective is to generate practical understandings to promote informed decisions.

Figure 1. The four phases of IoT solutions

2.2. Green internet of things applications

GIoT applications have extensive effects across several areas, significantly advancing sustainability
and efficiency. In energy management, Farhan et al. [11] have examined energy efficiency strategies and
electric power systems for GIoT networks, highlighting the importance of sustainable practices. In smart
agriculture, IoT devices can optimize agricultural practices, increase crop yields, and reduce environmental
impact [12], [13]. In waste management, IoT technologies, such as smart bins equipped with sensors and
solar power [14], promote environmentally friendly practices. Mohammadi et al. [15] state that a holistic
approach integrates waste collection and sorting, reducing costs and minimizing social and environmental
impacts. GIoT also enhances sustainable transportation by implementing technologies that reduce emissions
and improve efficiency. Furthermore, environmental monitoring leverages IoT devices to provide real-time
data for managing and protecting natural resources, as shown in studies [16]. These applications underscore
the transformative potential of GIoT in fostering sustainable development.

Unveiling critical features for failure prediction in green internet of things applications (Ouiam Khattach)
4310  ISSN: 2252-8938

3. RELATED WORK
The number of survey papers on GIoT that have attracted attention in recent years has increased.
Alsharif et al. [17] advocates for the adoption of eco-friendly IoT solutions by thoroughly exploring energy-
efficient practices and strategies which presents four principles/frameworks to achieve that vision by tackling
the energy efficiency issues related to hardware such as machine-to-machine communication, radiofrequency
identification, microcontroller units, wireless sensor networks, integrated circuits, embedded systems, and
processors. The objective is to advance sustainable and energy-efficient IoT technologies, contributing to the
next generation of eco-friendly implementations. Albreem et al. [18] examined effective behavioral change
models to raise awareness about energy conservation among IoT users and service providers. This article
delves into the key elements driving the development of the GIoTs, emphasizing energy efficiency hardware
design, data-center strategies, and software-based data traffic management. Almalki et al. [19] are motivated
by pursuing a sustainable smart world and delves into various technologies and considerations related to
GIoTs to reduce energy consumption. The study systematically examines key green information and
communication technologies (ICTs), including green radio frequency identification, green wireless sensor
networks, green cloud computing, green machine-to-machine, and green data centers, while distilling general
principles for green ICT. Varjovi and Babaie [20] examines the necessary measures to implement GIoT
across various levels, including hardware, software, communication, and network architecture. Along with
highlighting the significance of GIoT for environmental preservation, it also examines the prospects,
difficulties, and uses of this technology. Leading IT organizations' business models are examined, and
unresolved problems including standardization, technical difficulties, security, and innovations are examined
to inform future studies. In order to lower energy usage and CO2 emissions, the study emphasizes the
necessity of solutions at every stage of the GIoT life cycle, from design and production to use and recycling.

4. METHODOLOGY
This study provides a comprehensive approach to addressing IoT-based predictive maintenance
challenges. The workflow focuses on the problem of predicting IoT device failures and extracting a list of
essential features. Figure 2 illustrates the proposed methodology comprises three essential steps. Each step is
elaborated in more detail as follows:
− Step 1: data collection. The first step involves collecting pertinent data from IoT devices, such as sensor
readings, device logs, and historical maintenance records. This data serves as the foundation for building
accurate predictive models. Preprocessing techniques are then applied to clean the data, handle missing
values, and normalize the features, ensuring the data is suitable for analysis. By actively monitoring and
analyzing this data, patterns, and anomalies can be detected, enabling the prediction of potential failures
or malfunctions in IoT systems.
− Step 2: domains extraction. In IoT failure prediction, it is essential to identify and analyze the relevant
domains or areas of focus within an IoT system that are prone to failures or malfunctions. Understanding
the specific domains affected by failures makes it possible to develop more targeted and accurate failure
prediction models. This step involves conducting domain-specific analysis and identifying the key factors
or variables contributing to each domain's failures.
− Step 3: feature selection. In this step, we select and extract the dataset’s most relevant and informative
features. Feature selection methods, such as principal component analysis (PCA) and correlation analysis,
are applied to identify the most influential features that significantly contribute to failure prediction. By
focusing on the most important features, we can optimize the predictive models and enhance the accuracy
of failure predictions.

Figure 2. IoT-based predictive maintenance workflow

Int J Artif Intell, Vol. 14, No. 5, October 2025: 4308-4318

Int J Artif Intell ISSN: 2252-8938  4311

4.1. Data collection

Data collection provides a pivotal role in predicting IoT device failures. Several e-resources were
identified, as presented in Table 1. Kaggle and IEEE DataPort emerged as the primary repositories employed
in this research, as illustrated in Figure 3. Targeted keywords were used across multiple domains to facilitate
comprehensive data collection. Examples of these keywords include "IoT failure in {domain}" and
"monitoring IoT device failures”.

Table 1. List of e-resources for dataset discovery

No. E-resources Content
1 https://www.kaggle.com/ Databases
2 https://ieee-dataport.org/subscribe
3 https://zenodo.org/
4 https://archive.ics.uci.edu/ml/datasets.php
5 https://github.com/

Figure 3. Distribution of e-resources used in the study, highlighting Kaggle and IEEE DataPort
as primary sources

The size of the dataset has a significant impact on the results of the implemented models. Thus, in
this study, we considered the datasets' sizes an important key feature. The dataset sizes varied significantly,
ranging from 6.11 kB to 6 GB. Figure 3 illustrates the use of Kaggle and IEEE DataPort as primary sources,
highlighting their importance in providing relevant and high-quality datasets for IoT failure prediction
research, followed by Zenodo, GitHub, UCI, and other similar sources. These sources offer a wide range of
datasets encompassing different domains, which enables us to examine failure patterns, identify key features,
and develop accurate predictive models across diverse IoT applications.

4.2. Domaine extraction

The exploration of IoT across various applications has revealed significant opportunities to develop
advanced systems in diverse domains. Predicting IoT device failures offers considerable benefits in these
areas. Figure 4 illustrates the distribution of datasets used for IoT failure prediction in various domain
applications. A total of 32 datasets, 19 from the industry sector, 5 from agriculture, and 6 from energy.
However, some data contain unclear labels and abbreviations, which can lead to confusing the context and
understanding of the dataset. A rigorous data preprocessing approach will be put in place to overcome these
challenges. These include the elimination of confusing datasets, maintaining well-defined features, and
renaming labels to enhance the dataset's comprehensibility. In the next section, a feature selection strategy
was applied to identify the key features of this study.

Unveiling critical features for failure prediction in green internet of things applications (Ouiam Khattach)
4312  ISSN: 2252-8938

Figure 4. The distribution of collection data by domain

4.3. Feature selection strategy

In this section, we explore a range of feature selection methods, each adapted to the specific
approach for identifying essential features. Feature selection plays a pivotal role in identifying the key
contributors to IoT device failures. The objectives are to understand the data, reduce computational demands,
mitigate the effects of the curse of dimensionality, and enhance the performance of predictive models [21].
Feature selection aims to choose a subset of relevant variables from the input data. This process involves
minimizing the influence of noise or irrelevant variables while ensuring that the selected features contribute
to accurate predictions and optimized computational efficiency. Effective feature selection can significantly
improve model performance, reduce overfitting, and lower the computational cost associated with processing
large datasets. Feature selection methods are generally classified into two categories: supervised and
unsupervised methods, as shown in Figure 5.

Figure 5. Overview of feature selection techniques

Supervised feature selection targets classification tasks by leveraging the relevance or correlation
between features and class labels. The objective is to identify an optimal subset of features that enhances
classification accuracy. Various supervised techniques have been developed for that purpose [22]. Filter
methods are preprocessing processes for evaluating features, selecting those with high relevance scores based
on mutual information and correlation measurements [23]. On the other hand, wrapper methods apply
sequential or heuristic search techniques to identify the feature subset that maximizes performance through
embedding the predictor within a search algorithm. Embedded methods include feature selection directly
within the training process, eliminating the need for splitting data into separate training and testing sets.
Additionally, hybrid approaches and ensemble techniques combine filter and wrapper models, typically
involving two stages: initially reducing the feature space using the filter, and then employing the wrapper to
determine the most effective subset among the remaining features [21].
Other approaches highlight the important features of lacking dataset labels for unsupervised
learning. Clustering is a typical approach that groups similar data points. Clustering techniques, such as

Int J Artif Intell, Vol. 14, No. 5, October 2025: 4308-4318

Int J Artif Intell ISSN: 2252-8938  4313

hierarchical clustering and density-based clustering, organize data into clusters based on their proximity,
revealing underlying patterns without predefined classifications [24]. By applying these advanced feature
selection methods, we aim to enhance the accuracy and reliability of predictive models, ultimately improving
the effectiveness of IoT device failure predictions. In this study, the dataset goes through several stages. After
the preprocessing phase, exploratory data analysis (EDA) is applied. EDA provides valuable details about the
dataset, exposing distribution patterns and correlations between important characteristics. EDA serves as a
crucial guide for the next stages, assisting in data selection and facilitating the execution of machine learning
tasks [25]. After preprocessing, a feature selection strategy is applied to create a representative subset of
features critical for predictive modeling with AI methods. Performing EDA prior to feature selection is
essential for understanding dataset characteristics and identifying potential relationships between input
features and target variables [26]. This comprehensive approach ensures the selection of meaningful features
for effective modeling.
The feature selection strategy begins with an EDA to assess the characteristics and correlations
among the original features. This is followed by the implementation of advanced selection techniques,
including feature importance, information gain, Chi-square test, Fisher’s score, mutual information gain,
recursive feature elimination, and PCA. These methods are used for supervised learning, else other feature
selection methods are adopted for unsupervised learning. Variance threshold, mean absolute difference
(MAD), clustering k-means, hierarchical clustering, density-based clustering, PAM, Gaussian mixture model
(GMM), and self-organized maps (SOM) were implemented to eliminate the irrelevant original features. The
study is implemented in Python along with the required libraries, such as scikit-learn, Matplotlib, and
Skfeature. The feature selection strategy eliminates irrelevant original features, revealing deeper insights
within the remaining dataset.

5. RESULTS AND DISCUSSION

5.1. Results of feature selection strategy
A rigorous feature selection strategy was implemented to identify critical variables in our research.
EDA and other methods were employed to select important features. Table 2 presents an overview of the results,
highlighting the contributions and application domains of each selected dataset. There is no unified feature
selection strategy that can select important features in the application, we need to select the most appropriate
methods among a range of different methods to achieve the best performance. Features selection is an important
step in failure prediction problems, especially in GIoT applications. By applying the feature selection strategy
outlined in the previous section, the number of selected features was reduced by more than half. Thus, combining
these selected features can help generate new, original features for training models in future work.

Table 2. Datasets features selection

ID Contribution Application Features selection results
1 Predictive maintenance dataset Industry Torque, tool wear, rotation speed, air temperature, process temperature
2 Machine failure dataset Leakage, risk_MM, max_temp, parameter1_speed, electricity,
evaporation, min_temp
3 Predictive maintenance using Pressure, rotate, voltage, vibration
Microsoft case study
4 Elevator predictive maintenance Vibration, revolutions
dataset
5 Preventive to predictive Dust_feed, differential_pressure, time
maintenance
6 AnoML-IoT Light, humidity, loudness, temperature
7 Predicting machine failures Temperature, humidity, hours since previous failure, date.day-of-month,
date.day-of-week, date. month, date. hour.
8 Distributed transformer Energy Oil temperature indicator, winding temperature indicator, ambient
monitoring temperature indicator, oil level indicator, oil temperature indicator alarm,
oil temperature indicator trip, magnetic oil gauge indicator, voltage
9 MetroPT: a benchmark dataset gpsSpeed, gpsLat, gpsLong, Tp2, oil_temperature, flow meter,
for predictive maintenance motor_current, gpsQuality, H1, Tp3, DV_pressure, COMP,
caudal_impulses, MPG
10 Smart home dataset with weather Wind Bearing, dew point, apparent temperature, temperature, well,
information pressure, wind speed, humidity, house overall, solar, visibility, furnace,
wine cellar, precipIntensity, precipProbability
11 Real-time pond water dataset for Agriculture pH, temperature
fish framing
12 Smart agricultural production Potassium, rainfall, temperature, humidity, phosphorus, nitrogen, pH
optimizing engine
13 Intelligent irrigation system Humidity, temperature, watering
14 Sensor-based aquaponics Temperature, turbidity, dissolved oxygen, pH, ammonia, nitrate

Unveiling critical features for failure prediction in green internet of things applications (Ouiam Khattach)
4314  ISSN: 2252-8938

5.2. Critical features for failure prediction

The primary goal of this study is to identify the critical features for predicting failures in GIoT
applications. Table 3 provides an overview of the extracted features from various research articles, including
their contributions and domains of application. An analysis of the data presented in Tables 2 and 3,
complemented by Figures 6 and 7, reveals significant patterns in feature usage across different datasets.

Table 3. Key features identified in the literature and their domains of application
References Contribution Application Sensors/actuators Features selection results
[27] Prediction of Industry - Voltage, pressure, vibration,
machine failure in rotation, machine age, error
Industry type, number of components,
model type and failure
[28] Predictive - Temperature, vibration, two
maintenance analytics current
of autoclave sterilizer
[29] Rail transit vehicles - vibration
[30] MEP components Temperature sensor, pressure sensor, flow Sensor name, sensor id, sensor
HVAC systems rate sensor value, sensor type
[31] Industrial equipment ADXL345, ACS712, temperature sensor, Vibration measurement,
monitoring: electrical MLX90614 Infra-red thermometer, SHT21 temperature, voltage
motors digital humidity and temperature sensor
[32] Wind turbines Energy - Wind speed, power output, oil
temperature, bearing
temperature
[33] Fault detection and Temperature sensor, humidity sensor, Temperature, irradiance,
power prediction of irradiance sensor, voltage sensor, current power, voltage, humidity,
photovoltaic plants sensor current
[34] IoT smart home Temperature sensors, humidity, leak, water, -
smoke, air sensor, light sensors, dry contact
sensors, smart plugs, current transformers,
AC/DC voltage sensors, power synching
sensors, smart home monitoring kits
[35] Smart framing Agriculture CO2 sensor, UV sensor, luminance sensor, -
soil sensor, barometric pressure sensor,
moisture, temperature, electrical
conductivity (EC), pH sensors
[36] Smart framing: Soil sensors, temperature, EC, moisture -
calibrationTalk sensors, humidity sensor, nitrogen,
phosphorus, potassium
[37] IoT-based Temperature: MLX90614(TS1ca), humidity: Temperature, environmental
monitoring system: HTU21D(HS1c), humidity of the temperature relative humidity,
aeroponics environment: HTU21D(HS1a), luminous luminosity, pH level, EC
greenhouse sensor: BH1750(IS1a), webcam: LC4, IP (electrical conductivity), level
camera UC4 and nutrient solution
temperature, RGB and
thermographic images
[38] Smart framing: Temperature/humidity: E+E elektronik Temperature, humidity,
greenhouse EE160, electrical connectivity: B&C electrical connectivity, pH,
electronics 2731312-31/3-017T, pH: B&C level controller, liquid counter,
electronics SZ 1093, level controller: flow meter, solar radiation,
Omron K8AK-LS1, liquid counter: ARAD CO2, nitrogen, sulphur,
SF 15, flow meter: Gems FT110 G3/8, solar phosphorus, calcium,
radiation: Apogee, Instruments Inc. SP110 potassium, iron, copper,
manganese, boron, zinc,
molybdenum
[39] Smart sensors in Leaf sensor, temperature sensor, crop -
agriculture sensor, disease sensor, pest sensor, nutrient
sensor, soil moisture sensor, acoustic-based
sensor, electromagnetic sensors,
electrochemical sensors: volatile organic
compound sensor, humidity sensor, nutrient
sensor, pesticide sensor, O3 sensor, NO2
sensor, light detection and ranging
(LiDAR), optical sensors, field
programmable gate array (FPGA) based
sensors, Eddy covariance (EC) based
sensors, Mechanical and mass flow sensors,
flexible and wearable sensors, battery-free
and self powered sensors

Int J Artif Intell, Vol. 14, No. 5, October 2025: 4308-4318

Int J Artif Intell ISSN: 2252-8938  4315

Figure 6. Study’s key features variety

Figure 7. Study’s features frequency

Figure 6 categorizes these features using standardized terms, highlighting their broad applications
and variations. For example, the term “temperature” is consistently represented in various forms such as “oil
temperature” and “air temperature”, demonstrating its fundamental role in different contexts. Similarly, the
term “Wind” appears in multiple forms, including “wind bearing”, “wind speed”, and “wind temperature
indicator”, indicating its importance in capturing diverse atmospheric conditions. The term “speed” also
shows considerable variation, with instances such as “rotation speed”, “GPS speed”, and “wind speed”,
reflecting its application across different measurement scenarios. Additionally, “Light” is denoted by various
descriptors including “luminosity”, “luminous sensor”, and “light sensor”, while “soil nutrient” covers
specific elements such as “zinc”, “nitrogen”, “CO2”, and “sulphur”. The term “solar” includes “solar
radiation” and “UV sensor”, highlighting its relevance to solar energy studies.
This analysis underscores the necessity for standardized feature terminology to improve data
consistency and comparability. By aligning terminology across datasets, researchers can ensure more
accurate and coherent data integration, facilitating better comparative analyses and enhancing the robustness
of predictive models. This approach streamlines data processing and improves the reliability of insights
derived from diverse research studies and application domains.

Unveiling critical features for failure prediction in green internet of things applications (Ouiam Khattach)
4316  ISSN: 2252-8938

The high frequency of certain terms in this study strongly indicates their importance.
This recurrence shows the significance of these features across various datasets and applications in predictive
analysis. The frequency of the most cited terms from Tables 2 and 3 is shown in Figure 7. For instance,
terminology such as “temperature”, “humidity”, “pressure”, “voltage”, “pH” and “vibration” appear
frequently. With over thirty citations “temperature” is referenced, followed by “humidity” which indicates
their importance in the datasets analyzed. The recurrence of these terminologies across multiple studies and
datasets substantiates their indispensability, which are key indicators for monitoring and predicting IoT
device performance.

6. CONCLUSION
The present study reviews the crucial features for predicting failures in GIoT applications in
significant sectors, including agriculture, industry, and energy. A variety of datasets and research papers
focus on these features' critical role in predictive maintenance. Features such as temperature, humidity,
voltage, vibration, and pH are highlighted through a workflow process of data collection, domain extraction,
and feature selection. In future work, we will focus on developing predictive models using the most crucial
features within each domain. We aim to implement a robust model capable of predicting failures before they
can occur, optimizing performance, and reducing downtime. This study will contribute to developing reliable
and sustainable GIoT technologies, supporting environmental sustainability, and advancing the capabilities of
IoT systems in various domains.

ACKNOWLEDGEMENTS
We express our sincere gratitude to YosoBox SARL Company and the University Mohammed First
in Oujda, Morrocco, for their invaluable support and provision of essential facilities for this research.

FUNDING INFORMATION
This research received no external funding.

AUTHOR CONTRIBUTIONS STATEMENT

This journal uses the Contributor Roles Taxonomy (CRediT) to recognize individual author
contributions, reduce authorship disputes, and facilitate collaboration.

Name of Author C M So Va Fo I R D O E Vi Su P Fu
Ouiam Khattach ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
Omar Moussaoui ✓ ✓ ✓ ✓ ✓
Mohammed Hassine ✓ ✓ ✓ ✓ ✓

C : Conceptualization I : Investigation Vi : Visualization

M : Methodology R : Resources Su : Supervision
So : Software D : Data Curation P : Project administration
Va : Validation O : Writing - Original Draft Fu : Funding acquisition
Fo : Formal analysis E : Writing - Review & Editing

CONFLICT OF INTEREST STATEMENT

Author Mohammed Hassine was employed by the company Tisalabs Limited. The remaining
authors declare that the research was conducted in the absence of any commercial or financial relationships
that could be construed as a potential conflict of interest.

INFORMED CONSENT
Not applicable.

ETHICAL APPROVAL
Not applicable.

Int J Artif Intell, Vol. 14, No. 5, October 2025: 4308-4318

Int J Artif Intell ISSN: 2252-8938  4317

DATA AVAILABILITY
Data availability is not applicable to this paper as no new data were created or analyzed in this study.

REFERENCES
[1] L. Atzori, A. Iera, and G. Morabito, “The internet of things: a survey,” Computer Networks, vol. 54, no. 15, pp. 2787-2805,
Oct. 2010, doi: 10.1016/j.comnet.2010.05.010.
[2] S. Shibu et al., “Anomaly detection using deep learning approach for IoT smart city applications,” Multimedia Tools and
Applications, vol. 84, no. 17, pp. 17929–17949, Jul. 2025, doi: 10.1007/s11042-024-19176-x.
[3] S. B. Atitallah, M. Driss, W. Boulila, and H. B. Ghezala, “Leveraging deep learning and IoT big data analytics to support the smart
cities development: review and future directions,” Computer Science Review, vol. 38, Nov. 2020, doi: 10.1016/j.cosrev.2020.100303.
[4] N. Alharbe and M. Almalki, “IoT-enabled healthcare transformation leveraging deep learning for advanced patient monitoring and
diagnosis,” Multimedia Tools and Applications, vol. 84, no. 19, pp. 21331–21344, Jul. 2024, doi: 10.1007/s11042-024-19919-w.
[5] C. Ejiyi et al., “The internet of medical things in healthcare management: a review,” Journal of Digital Health, pp. 30–62,
Jun. 2023, doi: 10.55976/jdh.22023116330-62.
[6] N. G. Rezk, E. E. D. Hemdan, A. F. Attia, A. E. -Sayed, and M. A. E. -Rashidy, “An efficient IoT based smart farming system
using machine learning algorithms,” Multimedia Tools and Applications, vol. 80, no. 1, pp. 773–797, Jan. 2021,
doi: 10.1007/s11042-020-09740-6.
[7] S. Rudrakar and P. Rughani, “IoT based agriculture (Ag-IoT): a detailed study on architecture, security and forensics,”
Information Processing in Agriculture, vol. 11, no. 4, pp. 524–541, Dec. 2024, doi: 10.1016/j.inpa.2023.09.002.
[8] W. Mao, Z. Zhao, Z. Chang, G. Min, and W. Gao, “Energy-efficient industrial internet of things: overview and open issues,”
IEEE Transactions on Industrial Informatics, vol. 17, no. 11, pp. 7225–7237, Nov. 2021, doi: 10.1109/TII.2021.3067026.
[9] X. Zhang, G. Manogaran, and B. A. Muthu, “IoT enabled integrated system for green energy into smart cities,” Sustainable
Energy Technologies and Assessments, vol. 46, Aug. 2021, doi: 10.1016/j.seta.2021.101208.
[10] M. A. Sami and T. A. Khan, “Forecasting failure rate of IoT devices: a deep learning way to predictive maintenance,” Computers
and Electrical Engineering, vol. 110, Sep. 2023, doi: 10.1016/j.compeleceng.2023.108829.
[11] L. Farhan et al., “Energy efficiency for green internet of things (IoT) networks: a survey,” Network, vol. 1, no. 3, pp. 279–314,
Nov. 2021, doi: 10.3390/network1030017.
[12] M. Dhanaraju, P. Chenniappan, K. Ramalingam, S. Pazhanivelan, and R. Kaliaperumal, “Smart farming: internet of things (IoT)-
based sustainable agriculture,” Agriculture, vol. 12, no. 10, Oct. 2022, doi: 10.3390/agriculture12101745.
[13] M. A. Ferrag, L. Shu, X. Yang, A. Derhab, and L. Maglaras, “Security and privacy for green IoT-based agriculture: review,
blockchain solutions, and challenges,” IEEE Access, vol. 8, pp. 32031–32053, 2020, doi: 10.1109/ACCESS.2020.2973178.
[14] I. Sosunova and J. Porras, “IoT-enabled smart waste management systems for smart cities: a systematic review,” IEEE Access,
vol. 10, pp. 73326–73363, 2022, doi: 10.1109/ACCESS.2022.3188308.
[15] M. Mohammadi, G. Rahmanifar, M. H. -Keshteli, G. Fusco, and C. Colombaroni, “Industry 4.0 in waste management: an
integrated IoT-based approach for facility location and green vehicle routing,” Journal of Industrial Information Integration,
vol. 36, Dec. 2023, doi: 10.1016/j.jii.2023.100535.
[16] P. Asha et al., “IoT enabled environmental toxicology for air pollution monitoring using AI techniques,” Environmental
Research, vol. 205, Apr. 2022, doi: 10.1016/j.envres.2021.112574.
[17] M. H. Alsharif, A. Jahid, A. H. Kelechi, and R. Kannadasan, “Green IoT: a review and future research directions,” Symmetry,
vol. 15, no. 3, Mar. 2023, doi: 10.3390/sym15030757.
[18] M. A. Albreem, A. M. Sheikh, M. H. Alsharif, M. Jusoh, and M. N. M. Yasin, “Green internet of things (GIoT): applications,
practices, awareness, and challenges,” IEEE Access, vol. 9, pp. 38833–38858, 2021, doi: 10.1109/ACCESS.2021.3061697.
[19] F. A. Almalki et al., “Green IoT for eco-friendly and sustainable smart cities: future directions and opportunities,” Mobile
Networks and Applications, vol. 28, no. 1, pp. 178–202, Feb. 2023, doi: 10.1007/s11036-021-01790-w.
[20] A. E. Varjovi and S. Babaie, “Green internet of things (GIoT): vision, applications and research challenges,” Sustainable
Computing: Informatics and Systems, vol. 28, Dec. 2020, doi: 10.1016/j.suscom.2020.100448.
[21] D. Theng and K. K. Bhoyar, “Feature selection techniques for machine learning: a survey of more than two decades of research,”
Knowledge and Information Systems, vol. 66, no. 3, pp. 1575–1637, Mar. 2024, doi: 10.1007/s10115-023-02010-5.
[22] P. Dhal and C. Azad, “A comprehensive survey on feature selection in the various fields of machine learning,” Applied
Intelligence, vol. 52, no. 4, pp. 4543–4581, Mar. 2022, doi: 10.1007/s10489-021-02550-9.
[23] M. Buyukkececi and M. C. Okur, “A comprehensive review of feature selection and feature selection stability in machine
learning,” Gazi University Journal of Science, vol. 36, no. 4, pp. 1506–1520, Dec. 2023, doi: 10.35378/gujs.993763.
[24] S. S. -Fernández, J. A. C. -Ochoa, and J. F. M. -Trinidad, “A review of unsupervised feature selection methods,” Artificial
Intelligence Review, vol. 53, no. 2, pp. 907–948, Feb. 2020, doi: 10.1007/s10462-019-09682-y.
[25] M. Komorowski, D. C. Marshall, J. D. Salciccioli, and Y. Crutain, “Exploratory data analysis,” in Secondary Analysis of
Electronic Health Records, Cham, Switzerland: Springer, pp. 185–203, 2016.
[26] K. Sahoo, A. K. Samal, J. Pramanik, and S. K. Pani, “Exploratory data analysis using Python,” International Journal of
Innovative Technology and Exploring Engineering, vol. 8, no. 12, pp. 4727–4735, Oct. 2019, doi: 10.35940/ijitee.L3591.1081219.
[27] A. Wahid, J. G. Breslin, and M. A. Intizar, “Prediction of machine failure in industry 4.0: a hybrid CNN-LSTM framework,”
Applied Sciences, vol. 12, no. 9, Apr. 2022, doi: 10.3390/app12094221.
[28] I. N. Mihigo, M. Zennaro, A. Uwitonze, J. Rwigema, and M. Rovai, “On-device IoT-based predictive maintenance analytics
model: comparing TinyLSTM and TinyModel from edge impulse,” Sensors, vol. 22, no. 14, Jul. 2022, doi: 10.3390/s22145174.
[29] D. Yang, E. Cui, H. Wang, and H. Zhang, “EH-edge-an energy harvesting-driven edge iot platform for online failure prediction of
rail transit vehicles: a case study of a cloud, edge, and end device collaborative computing paradigm,” IEEE Vehicular
Technology Magazine, vol. 16, no. 2, pp. 95–103, Jun. 2021, doi: 10.1109/MVT.2021.3053193.
[30] J. C. P. Cheng, W. Chen, K. Chen, and Q. Wang, “Data-driven predictive maintenance planning framework for MEP components
based on BIM and IoT using machine learning algorithms,” Automation in Construction, vol. 112, Apr. 2020,
doi: 10.1016/j.autcon.2020.103087.
[31] N. A. Mohammed, O. F. Abdulateef, and A. H. Hamad, “An IoT and machine learning-based predictive maintenance system for
electrical motors,” Journal Europeen des Systemes Automatises, vol. 56, no. 4, pp. 651–656, Aug. 2023, doi: 10.18280/jesa.560414.
[32] C. H. Yeh, M. H. Lin, C. H. Lin, C. E. Yu, and M. J. Chen, “Machine learning for long cycle maintenance prediction of wind
turbine,” Sensors, vol. 19, no. 7, Apr. 2019, doi: 10.3390/s19071671.

Unveiling critical features for failure prediction in green internet of things applications (Ouiam Khattach)
4318  ISSN: 2252-8938

[33] M. Emamian, A. Eskandari, M. Aghaei, A. Nedaei, A. M. Sizkouhi, and J. Milimonfared, “Cloud computing and IoT based
intelligent monitoring system for photovoltaic plants using machine learning techniques,” Energies, vol. 15, no. 9, Apr. 2022,
doi: 10.3390/en15093014.
[34] G. Bedi, G. K. Venayagamoorthy, R. Singh, R. R. Brooks, and K. C. Wang, “Review of internet of things (IoT) in electric power
and energy systems,” IEEE Internet of Things Journal, vol. 5, no. 2, pp. 847–870, Apr. 2018, doi: 10.1109/JIOT.2018.2802704.
[35] Y. B. Lin, Y. W. Lin, J. Y. Lin, and H. N. Hung, “SensorTalk: an IoT device failure detection and calibration mechanism for
smart farming,” Sensors, vol. 19, no. 21, Nov. 2019, doi: 10.3390/s19214788.
[36] Y. W. Lin, Y. B. Lin, and H. N. Hung, “CalibrationTalk: a farming sensor failure detection and calibration technique,” IEEE
Internet of Things Journal, vol. 8, no. 8, pp. 6893–6903, Apr. 2021, doi: 10.1109/JIOT.2020.3036859.
[37] H. A. M. -Guzmán et al., “IoT-based monitoring system applied to aeroponics greenhouse,” Sensors, vol. 22, no. 15, Jul. 2022,
doi: 10.3390/s22155646.
[38] M. A. Z.-Izquierdo, J. Santa, J. A. Martínez, V. Martínez, and A. F. Skarmeta, “Smart farming IoT platform based on edge and
cloud computing,” Biosystems Engineering, vol. 177, pp. 4–17, Jan. 2019, doi: 10.1016/j.biosystemseng.2018.10.014.
[39] P. Rajak, A. Ganguly, S. Adhikary, and S. Bhattacharya, “Internet of things and smart sensors in agriculture: scopes and
challenges,” Journal of Agriculture and Food Research, vol. 14, Dec. 2023, doi: 10.1016/j.jafr.2023.100776.

BIOGRAPHIES OF AUTHORS

Ouiam Khattach is a Ph.D. candidate in Computer Engineering at Mohammed

First University in Oujda, Morocco, where she is researching IoT failures using machine
learning approaches. Graduated and received an engineering degree in Big Data and Cloud
Computing from the school ENSET Mohammedia, Morocco, in 2021. She participates in
several scientific & organizing committees of national and international conferences.
Additionally, she holds several certifications in artificial intelligence, programming, and
networking. She can be contacted at email: [email protected].

Prof. Dr. Omar Moussaoui is an Associate Professor at the Higher School of

Technology (ESTO) of Mohammed First University, Oujda–Morocco. He has been a member
of the Department of Computer Science of ESTO since 2013. He is currently the Director of
the MATSI Research Laboratory. He completed his Ph.D. in Computer Science at the
University of Cergy-Pontoise, France, in 2006. His research interests lie in the fields of IoT,
wireless networks, and security. He has actively collaborated with researchers in several other
computer science disciplines. He participated in several scientific & organizing committees of
national and international conferences. He served as a reviewer for numerous international
journals. He has more than 20 publications in international journals and conferences, and he
has co-authored 2 book chapters. He is an instructor for CISCO Networking Academy on
CCNA Routing & Switching and CCNA Security. He can be contacted at email:
[email protected].

Mohammed Hassine is a distinguished professional with over 20 years of

expertise in product development, prominently contributing to the success of two startups,
Raidtec Corporation and MPSTOR. As a key member of the management team at MPSTOR,
he played a pivotal role in the company's triumphant exit, securing a multi-million $ deal in
2016. With a focus on cybersecurity and AI, he is executive role at MPSTOR involved leading
the development of intricate storage controllers, overseeing both hardware and software
aspects. His responsibilities spanned high-level product design, strategic marketing initiatives,
and the orchestration of all engineering processes necessary for transforming conceptual ideas
into fully industrialized products. His project management expertise was instrumental in
successfully delivering complex projects to Tier 1 customers, including industry giants such as
EMC, Intel, Quantum, Avid, and AIC. His multifaceted role showcased a blend of technical
proficiency, strategic vision, and leadership acumen, establishing him as a driving force
behind the achievements of the organizations he has been a part of, particularly in the realms
of cybersecurity and AI. He can be contacted at email: [email protected].

Int J Artif Intell, Vol. 14, No. 5, October 2025: 4308-4318