Amjad Gawanmeh
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Research Interests
Amjad Gawanmeh has a total of 231 co-authors. Here are the top co-authors by number of shared publications:
- Hussam Al Hamadi
College of Engineering and IT, University of Dubai, Dubai, UAE
13 shared publications Sazia Parvin
9 shared publications- Mahmoud Al-Qutayri
Department of Computer and Information Engineering, System on Chip Center, Khalifa University, Abu Dhabi, United Arab Emirates
7 shared publications Sarra Cherbal
7 shared publications- Shadi Atalla
College of Engineering and Information Technology, University of Dubai, Dubai 14143, United Arab Emirates
7 shared publications - Wathiq Mansoor
University of Dubai
7 shared publications
Latest Publications
- 15 October 2026
- Published by Springer Nature in Cluster Computing
- Vol. 29 (15) , 855
Blockchain–federated learning integration in the internet of vehicles: analysis, taxonomies, and future directions
- 15 October 2026
- Published by Springer Nature in Cluster Computing
- Vol. 29 (15) , 855
Blockchain–federated learning integration in the internet of vehicles: analysis, taxonomies, and future directions
- Open Access
- 25 September 2026
- Published by Elsevier in Procedia Computer Science
- Vol. 290, 467-474
Modeling Robotaxi and eVTOL Integration in Dubai’s Public Transport Network Using a Multi-Layer Graph Framework
- 25 September 2026
- Published by Elsevier in Procedia Computer Science
- Vol. 290, 467-474
Open AccessModeling Robotaxi and eVTOL Integration in Dubai’s Public Transport Network Using a Multi-Layer Graph Framework
- Open Access
- 1 August 2026
- Published by Institute of Advanced Engineering and Science in IAES International Journal of Artificial Intelligence (IJ-AI)
- Vol. 15 (4) , 3164-3175
Exploring cognitive patterns in children with autism spectrum disorder using correlation and cluster analysis
This paper investigates the relationships between age and five cognitive abilities in children diagnosed with autism spectrum disorder (ASD). Data from 210 children aged 6 to 12 years who completed measures of visual motor precision (VM), facial memory (FM), spatial vision (SV), drawing memory (DM), and orientation (OR) were analyzed. Data preparation, descriptive statistical analysis, correlation analysis, cluster analysis, and factor analysis were conducted. The results showed that age was not significantly associated with the cognitive measures. The clearest association was a moderate positive correlation between VM and FM (0.34). Cluster analysis identified three groups. However, the low silhouette score (0.1368) indicated weak separation and substantial overlap between cognitive profiles. Factor analysis suggested three latent dimensions: factor 1 was strongly associated with VM and FM; factor 2 was negatively related to DM and moderately positively related to OR; and factor 3 was primarily driven by SV. These findings suggest that cognitive abilities in children with ASD operate relatively independently, with some shared mechanisms between certain skills. It can be concluded that individualized approaches in assessment and intervention strategies are warranted.
- 1 August 2026
- Published by Institute of Advanced Engineering and Science in IAES International Journal of Artificial Intelligence (IJ-AI)
- Vol. 15 (4) , 3164-3175
Open AccessExploring cognitive patterns in children with autism spectrum disorder using correlation and cluster analysis
This paper investigates the relationships between age and five cognitive abilities in children diagnosed with autism spectrum disorder (ASD). Data from 210 children aged 6 to 12 years who completed measures of visual motor precision (VM), facial memory (FM), spatial vision (SV), drawing memory (DM), and orientation (OR) were analyzed. Data preparation, descriptive statistical analysis, correlation analysis, cluster analysis, and factor analysis were conducted. The results showed that age was not significantly associated with the cognitive measures. The clearest association was a moderate positive correlation between VM and FM (0.34). Cluster analysis identified three groups. However, the low silhouette score (0.1368) indicated weak separation and substantial overlap between cognitive profiles. Factor analysis suggested three latent dimensions: factor 1 was strongly associated with VM and FM; factor 2 was negatively related to DM and moderately positively related to OR; and factor 3 was primarily driven by SV. These findings suggest that cognitive abilities in children with ASD operate relatively independently, with some shared mechanisms between certain skills. It can be concluded that individualized approaches in assessment and intervention strategies are warranted.
- Open Access
Indoor Environmental Quality as an Incremental Signal in Residential Valuation Using Hedonic Modeling
This study presents an Indoor Environmental Quality (IEQ)-aware framework for residential valuation by integrating low-cost IoT sensing, transparent scoring, and hedonic price modeling. The analysis uses a dataset of 244 apartments across 12 districts in Tehran. It combines indicators of thermal comfort, particulate exposure, lighting, acoustics, stability, exceedance, and uncertainty with conventional housing covariates (area, age, bedrooms, floor level, renovation status, amenities, and accessibility proxies). Results show that pooled IEQ–price relationships are weak and confounded, whereas controlled specifications produce modest but consistent improvements in explanatory fit after IEQ features are introduced. Conventional location and structural attributes remain the dominant determinants of price per square meter. Still, IEQ contributes a non-redundant information layer that improves within-segment differentiation and interpretability for inspection and listing workflows. Methodologically, the framework extends beyond average comfort metrics by incorporating volatility, threshold exceedance duration, and sensor uncertainty, enabling uncertainty-aware reporting rather than single-point scoring. In practice, the workflow supports portable sensing, reproducible analytics, and privacy-preserving edge aggregation, suitable for PropTech deployment. The findings support a cautious but actionable conclusion: IEQ should be treated as an incremental valuation signal rather than a standalone pricing determinant. In this context, IEQ is conceptualized as a supplementary attribute block that may add explanatory value beyond conventional housing covariates rather than as a standalone pricing determinant.
Open AccessIndoor Environmental Quality as an Incremental Signal in Residential Valuation Using Hedonic Modeling
This study presents an Indoor Environmental Quality (IEQ)-aware framework for residential valuation by integrating low-cost IoT sensing, transparent scoring, and hedonic price modeling. The analysis uses a dataset of 244 apartments across 12 districts in Tehran. It combines indicators of thermal comfort, particulate exposure, lighting, acoustics, stability, exceedance, and uncertainty with conventional housing covariates (area, age, bedrooms, floor level, renovation status, amenities, and accessibility proxies). Results show that pooled IEQ–price relationships are weak and confounded, whereas controlled specifications produce modest but consistent improvements in explanatory fit after IEQ features are introduced. Conventional location and structural attributes remain the dominant determinants of price per square meter. Still, IEQ contributes a non-redundant information layer that improves within-segment differentiation and interpretability for inspection and listing workflows. Methodologically, the framework extends beyond average comfort metrics by incorporating volatility, threshold exceedance duration, and sensor uncertainty, enabling uncertainty-aware reporting rather than single-point scoring. In practice, the workflow supports portable sensing, reproducible analytics, and privacy-preserving edge aggregation, suitable for PropTech deployment. The findings support a cautious but actionable conclusion: IEQ should be treated as an incremental valuation signal rather than a standalone pricing determinant. In this context, IEQ is conceptualized as a supplementary attribute block that may add explanatory value beyond conventional housing covariates rather than as a standalone pricing determinant.
- Open Access
- 15 May 2026
- Published by The Scientific and Technological Research Council of Turkey (TUBITAK-ULAKBIM) - DIGITAL COMMONS JOURNALS in Turkish Journal of Electrical Engineering and Computer Sciences
- Vol. 34 (3) , 419-436
Multiobjective optimization framework for renewable energy IoT networks balancing security, energy efficiency, and communication reliability
The deployment of Internet of things (IoT) networks powered by renewable energy sources presents unique challenges in balancing security requirements, energy efficiency, and communication reliability. This paper presents a comprehensive multiobjective optimization framework for secure renewable energy IoT nodes that addresses fundamental trade-offs between these competing objectives. We develop a mathematical model incorporating energy harvesting dynamics, security protocols, and communication performance metrics across various environmental scenarios. The proposed framework employs a modified NSGA-II algorithm to identify Pareto-optimal configurations for different deployment contexts. Through extensive simulation analysis, we demonstrate that hybrid energy sources (solar-wind combinations) with lightweight security protocols achieve optimal balance across performance metrics. The results show that each security level increment reduces energy efficiency by approximately 7.5%, while hybrid configurations maintain 15-20% higher reliability compared to single-source alternatives across varied environmental conditions. The framework provides practical guidelines for IoT deployment strategies in sustainable applications.
- 15 May 2026
- Published by The Scientific and Technological Research Council of Turkey (TUBITAK-ULAKBIM) - DIGITAL COMMONS JOURNALS in Turkish Journal of Electrical Engineering and Computer Sciences
- Vol. 34 (3) , 419-436
Open AccessMultiobjective optimization framework for renewable energy IoT networks balancing security, energy efficiency, and communication reliability
The deployment of Internet of things (IoT) networks powered by renewable energy sources presents unique challenges in balancing security requirements, energy efficiency, and communication reliability. This paper presents a comprehensive multiobjective optimization framework for secure renewable energy IoT nodes that addresses fundamental trade-offs between these competing objectives. We develop a mathematical model incorporating energy harvesting dynamics, security protocols, and communication performance metrics across various environmental scenarios. The proposed framework employs a modified NSGA-II algorithm to identify Pareto-optimal configurations for different deployment contexts. Through extensive simulation analysis, we demonstrate that hybrid energy sources (solar-wind combinations) with lightweight security protocols achieve optimal balance across performance metrics. The results show that each security level increment reduces energy efficiency by approximately 7.5%, while hybrid configurations maintain 15-20% higher reliability compared to single-source alternatives across varied environmental conditions. The framework provides practical guidelines for IoT deployment strategies in sustainable applications.
- 12 May 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-6
Integrating Computational Intelligence for Automated Administration in Moodle Learning Environments
Increased complexity and size of virtual learning environments (VLEs) have rendered the old, manual administrative methods ineffective and this has led to burnout and poor optimum support of the students by the instructor. The paper describes a new system, which involves application of computational intelligence (CI) to make major administrative processes automated in Moodle learning management system. A tree-based ensemble and a Long Short-term Memory (LSTM) network are jointly applied in the recognition of a temporal pattern and feature analysis respectively. These models are driving an intelligent orchestrated engine that automates interventions i.e., personalized alert generation and resource prescription, on accepted policies. It has an administrator special dashboard with real time analytics and system health. The data pipeline was employed in this work to integrate user activity, assessment and forum data, which are further utilized to predict the identification of the at-risk students and time-on-task estimates. The effectiveness of the system shown by simulation experiments on a simulated dataset based on a typical university course shows a 92 percent accuracy rate at the first level of predicting at-risk students and an estimated 60 percent decrease in the volume of simulated administrative task. In the conclusion of the research, it is concluded that CI as an automated option of administrating the materials can be not only integrated into Moodle, it is necessary to scale the individual education and enhance the efficiency of the institutions.
- 12 May 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-6
Integrating Computational Intelligence for Automated Administration in Moodle Learning Environments
Increased complexity and size of virtual learning environments (VLEs) have rendered the old, manual administrative methods ineffective and this has led to burnout and poor optimum support of the students by the instructor. The paper describes a new system, which involves application of computational intelligence (CI) to make major administrative processes automated in Moodle learning management system. A tree-based ensemble and a Long Short-term Memory (LSTM) network are jointly applied in the recognition of a temporal pattern and feature analysis respectively. These models are driving an intelligent orchestrated engine that automates interventions i.e., personalized alert generation and resource prescription, on accepted policies. It has an administrator special dashboard with real time analytics and system health. The data pipeline was employed in this work to integrate user activity, assessment and forum data, which are further utilized to predict the identification of the at-risk students and time-on-task estimates. The effectiveness of the system shown by simulation experiments on a simulated dataset based on a typical university course shows a 92 percent accuracy rate at the first level of predicting at-risk students and an estimated 60 percent decrease in the volume of simulated administrative task. In the conclusion of the research, it is concluded that CI as an automated option of administrating the materials can be not only integrated into Moodle, it is necessary to scale the individual education and enhance the efficiency of the institutions.
- 12 May 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-6
QRphish: An Automated QR Code Phishing Detection Approach
QR codes have become the dominant physical-to-digital handoff for payments, tickets, menus, and login links. That convenience also makes them a clean delivery mechanism for phishing. Most scanners treat the symbol as a transport wrapper: decode the payload, show a URL, and stop. QR symbols carry structural choices (error correction level and masking pattern) that change how tolerant the code is to physical tampering and how much modification it can absorb while remaining scannable. We present QRphish, a detection framework that fuses URL lexical features with QR structural metadata—specifically Error Correction Level (ECL) and mask pattern—extracted from the format information bits. On a dataset of 10,000 QR codes (5,000 benign, 5,000 phishing), QRphish achieves a 98.5% detection rate with a 1.1% false-positive rate, outperforming URL-only and image-only baselines. The most actionable result is a strong behavioral skew: 78% of phishing QR codes in our corpus use ECL H, compared to 15% of benign codes. This "Level H anomaly" matches a practical attacker need. High redundancy (nominally up to 30% restoration) lets adversaries place logos, stickers, or visual clutter over the symbol to steer attention while it can still be decoded.
- 12 May 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-6
QRphish: An Automated QR Code Phishing Detection Approach
QR codes have become the dominant physical-to-digital handoff for payments, tickets, menus, and login links. That convenience also makes them a clean delivery mechanism for phishing. Most scanners treat the symbol as a transport wrapper: decode the payload, show a URL, and stop. QR symbols carry structural choices (error correction level and masking pattern) that change how tolerant the code is to physical tampering and how much modification it can absorb while remaining scannable. We present QRphish, a detection framework that fuses URL lexical features with QR structural metadata—specifically Error Correction Level (ECL) and mask pattern—extracted from the format information bits. On a dataset of 10,000 QR codes (5,000 benign, 5,000 phishing), QRphish achieves a 98.5% detection rate with a 1.1% false-positive rate, outperforming URL-only and image-only baselines. The most actionable result is a strong behavioral skew: 78% of phishing QR codes in our corpus use ECL H, compared to 15% of benign codes. This "Level H anomaly" matches a practical attacker need. High redundancy (nominally up to 30% restoration) lets adversaries place logos, stickers, or visual clutter over the symbol to steer attention while it can still be decoded.
- 12 May 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-6
Enhanced Moodle Analytics for Learning Management Optimization using Computational Intelligence
Learnning Management Systems (LMS) such as Moodle have been adopted in higher education which has created huge volumes of unexploited data that contain the secret of learning and can be used to enhance the process. Conventional analytics tend not to offer actionable and predictive information as a means to intervene in real time. To modify Moodle data to a strategic learning optimization asset, this paper suggests a new computational intelligence (CI) system. We combine machine learning (ML) to do predictive modeling, neural networks to do pattern recognition during multi-dimensional user interactions, and a fuzzy logic to deal with inherent inaccuracy in educational data. The proposed system takes in Moodle log data, does large-scale feature engineering, and uses a multimodel CI scheme to forecast student performance and learners at risk. The experimental results on a dataset of a real-world university Moodle deployment, consisting of more than 1,200 students and 15 courses, show that our ensemble model is able to predict the performance of students with a predictive accuracy of 91.3% and a precision of 89.5% which is significantly better than the baseline models. It is combined with the help of a business intelligence (BI) dashboard to visualize the insights, allowing making decisions based on data in order to instruct. The study adds to the scalable, intelligent analytics framework that will improve pedagogical performance and open the way to proactive, personalized learning.
- 12 May 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-6
Enhanced Moodle Analytics for Learning Management Optimization using Computational Intelligence
Learnning Management Systems (LMS) such as Moodle have been adopted in higher education which has created huge volumes of unexploited data that contain the secret of learning and can be used to enhance the process. Conventional analytics tend not to offer actionable and predictive information as a means to intervene in real time. To modify Moodle data to a strategic learning optimization asset, this paper suggests a new computational intelligence (CI) system. We combine machine learning (ML) to do predictive modeling, neural networks to do pattern recognition during multi-dimensional user interactions, and a fuzzy logic to deal with inherent inaccuracy in educational data. The proposed system takes in Moodle log data, does large-scale feature engineering, and uses a multimodel CI scheme to forecast student performance and learners at risk. The experimental results on a dataset of a real-world university Moodle deployment, consisting of more than 1,200 students and 15 courses, show that our ensemble model is able to predict the performance of students with a predictive accuracy of 91.3% and a precision of 89.5% which is significantly better than the baseline models. It is combined with the help of a business intelligence (BI) dashboard to visualize the insights, allowing making decisions based on data in order to instruct. The study adds to the scalable, intelligent analytics framework that will improve pedagogical performance and open the way to proactive, personalized learning.
- 12 May 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-6
Enhancing LMS Privacy and Performance through Computational Intelligence for Secure and Efficient Moodle Administration
Moodle has found extensive application in higher education, although there has been a growing pressure on both its administration and security, as well as on performance that can survive bursty academic loads. The paper describes a computational intelligence (CI) architecture, which consolidates. (i) multi-source telemetry web attack and misuse detection, (ii) privacy-preserving operational decision learning, and (iii) predictive performance management risk-conscious autoscaling and configuration guardrails. The framework is an amalgamation of signature-based controls with admonitored HTTP attack classification, log-sequence anomaly identification with pre-trained language models, and forecast-driven autoscaling policy informed by meta reinforcement learning. To address the privacy limitations, it fosters on-premise analytics, federated training across institutional silos, and differentiated privacy of aggregate indicators delivered to administrators. Predictive CI-driven scaling improves the 99th-percentile response time of 0.681 s (0.050 s) with 8 hours of Moodle-like workload simulation, eradicates SLA violations (0.81 to 0 percent), and decreases the compute cost (64.00 to 57.35 server-hours). The proposed supervised layer has F1 = 0.817 and AUC = 0.994 on a synthetic dataset of HTTP requests, against a baseline of F1 = 0.358 using the rule only. The findings help to support incremental implementation of CI elements having auditable policies and low operation load.
- 12 May 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-6
Enhancing LMS Privacy and Performance through Computational Intelligence for Secure and Efficient Moodle Administration
Moodle has found extensive application in higher education, although there has been a growing pressure on both its administration and security, as well as on performance that can survive bursty academic loads. The paper describes a computational intelligence (CI) architecture, which consolidates. (i) multi-source telemetry web attack and misuse detection, (ii) privacy-preserving operational decision learning, and (iii) predictive performance management risk-conscious autoscaling and configuration guardrails. The framework is an amalgamation of signature-based controls with admonitored HTTP attack classification, log-sequence anomaly identification with pre-trained language models, and forecast-driven autoscaling policy informed by meta reinforcement learning. To address the privacy limitations, it fosters on-premise analytics, federated training across institutional silos, and differentiated privacy of aggregate indicators delivered to administrators. Predictive CI-driven scaling improves the 99th-percentile response time of 0.681 s (0.050 s) with 8 hours of Moodle-like workload simulation, eradicates SLA violations (0.81 to 0 percent), and decreases the compute cost (64.00 to 57.35 server-hours). The proposed supervised layer has F1 = 0.817 and AUC = 0.994 on a synthetic dataset of HTTP requests, against a baseline of F1 = 0.358 using the rule only. The findings help to support incremental implementation of CI elements having auditable policies and low operation load.