Selina Sharmin
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Selina Sharmin has a total of 33 co-authors. Here are the top co-authors by number of shared publications:
- Alamin Talukder
Department of Computer Science and Engineering, International University of Business Agriculture and Technology, Dhaka, Bangladesh
4 shared publications - Ashraf Uddin
School of Information Technology, Deakin University, Waurn Ponds Campus, Geelong, Australia
4 shared publications - Manowarul Islam
Department of Computer Science and Engineering Jagannath University Dhaka Bangladesh
3 shared publications - Kari-Jouko Räihä
University of Tampere
2 shared publications - Oleg Špakov
Tampere University
2 shared publications - Sajeeb Saha
Jagannath University
2 shared publications
Latest Publications
- Open Access
- 1 July 2026
- Published by Institute of Advanced Engineering and Science in Indonesian Journal of Electrical Engineering and Computer Science
- Vol. 43 (1) , 179-191
REHA:real-time IoT-based energy efficient home automation system using ESP8266 and PIR motion sensors
Technological developments have improved living standards, leading to greater demand for home automation based on the internet of things (IoT) concept. This study addresses the limitations of existing home automation systems, which are often costly, complex, and lack energy-saving features. A low-cost IoT-based home automation system is developed using the ESP8266 NodeMCU and PIR motion sensors to enable both remote control and automatic operation of house hold appliances. It uses the Blynk platform for cloud-based monitoring with smart motion-based logic to minimize unnecessary power consumption. The evaluation is conducted through various controlled scenarios and an estimate based energy analysis derived from standard appliance ratings. Experimental results showed a reduction in household energy consumption of about 13–15%. The proposed system offers a cost-effective, practical approach to smart home energy management by combining affordability, automation, and measurable efficiency improvements.
- 1 July 2026
- Published by Institute of Advanced Engineering and Science in Indonesian Journal of Electrical Engineering and Computer Science
- Vol. 43 (1) , 179-191
Open AccessREHA:real-time IoT-based energy efficient home automation system using ESP8266 and PIR motion sensors
Technological developments have improved living standards, leading to greater demand for home automation based on the internet of things (IoT) concept. This study addresses the limitations of existing home automation systems, which are often costly, complex, and lack energy-saving features. A low-cost IoT-based home automation system is developed using the ESP8266 NodeMCU and PIR motion sensors to enable both remote control and automatic operation of house hold appliances. It uses the Blynk platform for cloud-based monitoring with smart motion-based logic to minimize unnecessary power consumption. The evaluation is conducted through various controlled scenarios and an estimate based energy analysis derived from standard appliance ratings. Experimental results showed a reduction in household energy consumption of about 13–15%. The proposed system offers a cost-effective, practical approach to smart home energy management by combining affordability, automation, and measurable efficiency improvements.
- Open Access
- 6 May 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 14, 70027-70045
A Scalable Spatiotemporal Framework for Multi-Hazard Climate Risk Assessment Using Remote Sensing in Bangladesh
Bangladesh is vulnerable to climatic disasters such as floods, droughts, and vegetation degradation. While previous research has predominantly addressed single hazards and not adopted integrated methodologies, there is also a lack of studies combining multiple satellite-derived indicators with consistent annotations. This limitation constrains the scope of long-term and district-level analyses. To address these gaps, this research introduces a framework for multi-hazard environmental risk assessment in Bangladesh, utilizing satellite-derived indicators such as NDVI, NDWI, land surface temperature (LST), and rainfall over a 23-year period. The framework employs a hybrid annotation strategy that integrates literature-informed rule-based thresholds with K-means clustering. As a result, a reproducible benchmark dataset representing flood, drought, and vegetation risk conditions across 64 districts is produced. Subsequently, the resulting labels are evaluated using ensemble machine learning models, including Gradient Boosting, Bagging, and Stacking, with the primary objective of assessing dataset consistency rather than optimizing model performance. Validation against independently verified meteorological records and historical events demonstrates substantial agreement. Moreover, the annotated dataset achieves accuracies of 0.895 and 0.939 for flood and drought risks, respectively, indicating a dependable reflection of real-world environmental conditions. Additionally, the framework is methodologically scalable and supports the integration of more satellite data through a standardized processing pipeline. Spatiotemporal analysis further identifies regional patterns: coastal and northeastern districts exhibit stronger flood-related signals, whereas western and southwestern regions display conditions associated with drought and vegetation stress. Overall, the proposed dataset and framework establish a basis for long-term environmental hazard analysis and climate risk monitoring in Bangladesh.
- 6 May 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 14, 70027-70045
Open AccessA Scalable Spatiotemporal Framework for Multi-Hazard Climate Risk Assessment Using Remote Sensing in Bangladesh
Bangladesh is vulnerable to climatic disasters such as floods, droughts, and vegetation degradation. While previous research has predominantly addressed single hazards and not adopted integrated methodologies, there is also a lack of studies combining multiple satellite-derived indicators with consistent annotations. This limitation constrains the scope of long-term and district-level analyses. To address these gaps, this research introduces a framework for multi-hazard environmental risk assessment in Bangladesh, utilizing satellite-derived indicators such as NDVI, NDWI, land surface temperature (LST), and rainfall over a 23-year period. The framework employs a hybrid annotation strategy that integrates literature-informed rule-based thresholds with K-means clustering. As a result, a reproducible benchmark dataset representing flood, drought, and vegetation risk conditions across 64 districts is produced. Subsequently, the resulting labels are evaluated using ensemble machine learning models, including Gradient Boosting, Bagging, and Stacking, with the primary objective of assessing dataset consistency rather than optimizing model performance. Validation against independently verified meteorological records and historical events demonstrates substantial agreement. Moreover, the annotated dataset achieves accuracies of 0.895 and 0.939 for flood and drought risks, respectively, indicating a dependable reflection of real-world environmental conditions. Additionally, the framework is methodologically scalable and supports the integration of more satellite data through a standardized processing pipeline. Spatiotemporal analysis further identifies regional patterns: coastal and northeastern districts exhibit stronger flood-related signals, whereas western and southwestern regions display conditions associated with drought and vegetation stress. Overall, the proposed dataset and framework establish a basis for long-term environmental hazard analysis and climate risk monitoring in Bangladesh.
- Open Access
- 10 May 2025
- Published by Elsevier in Transportation Engineering
- Vol. 20
Quality-aware bus-driver scheduling for intelligent transportation system
- 10 May 2025
- Published by Elsevier in Transportation Engineering
- Vol. 20
Open AccessQuality-aware bus-driver scheduling for intelligent transportation system
- Open Access
- 17 March 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 13, 50996-51011
Q-DMAC: Quality-Aware Distributed Channel Selection MAC Protocol for Enhancing Multi-Channel Communication in Directional Sensor Networks
The performance of wireless networks has continuously improved using Directional Wireless Sensor Networks (DSNs). DSN achieves this by enhancing data sensing accuracy, maximizing spectrum use, decreasing interference and collisions, and lengthening the network’s operating lifetime.Within the domain of Directional Sensor Networks (DSNs), the primary difficulties are coordinating transmission and reception possibilities for a variety of sender-receiver pairs and offering traffic priority-aware channel access. To improve the quality and efficiency of multi-channel communication in DSNs, this research presents a quality-aware Directional MAC, Q-DMAC a novel medium access control protocol. The novelty of our protocol lies in its ability to leverage multi-channel resources effectively, enabling reuse through nodes equipped with directional antennas while ensuring the Quality of Service (QoS). Proposed Q-DMAC assigns a higher number of slots inside a data window to nodes carrying more packets waiting to be sent and prioritizes medium access to nodes carrying a significant amount of data packets. Also by employing regret matching of game theory Q-DMAC selects channels resulting in reduced channel switching cost as well as computation cost. Simulation results demonstrate that Q-DMAC outperforms existing protocols such as CAR MAC, 2D-CMDMAC, and DCDS-MAC, achieving up to 20% higher throughput, 15% lower latency, 12% better packet delivery ratio, 25% longer network lifetime, and 15% reduced overhead.
- 17 March 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 13, 50996-51011
Open AccessQ-DMAC: Quality-Aware Distributed Channel Selection MAC Protocol for Enhancing Multi-Channel Communication in Directional Sensor Networks
The performance of wireless networks has continuously improved using Directional Wireless Sensor Networks (DSNs). DSN achieves this by enhancing data sensing accuracy, maximizing spectrum use, decreasing interference and collisions, and lengthening the network’s operating lifetime.Within the domain of Directional Sensor Networks (DSNs), the primary difficulties are coordinating transmission and reception possibilities for a variety of sender-receiver pairs and offering traffic priority-aware channel access. To improve the quality and efficiency of multi-channel communication in DSNs, this research presents a quality-aware Directional MAC, Q-DMAC a novel medium access control protocol. The novelty of our protocol lies in its ability to leverage multi-channel resources effectively, enabling reuse through nodes equipped with directional antennas while ensuring the Quality of Service (QoS). Proposed Q-DMAC assigns a higher number of slots inside a data window to nodes carrying more packets waiting to be sent and prioritizes medium access to nodes carrying a significant amount of data packets. Also by employing regret matching of game theory Q-DMAC selects channels resulting in reduced channel switching cost as well as computation cost. Simulation results demonstrate that Q-DMAC outperforms existing protocols such as CAR MAC, 2D-CMDMAC, and DCDS-MAC, achieving up to 20% higher throughput, 15% lower latency, 12% better packet delivery ratio, 25% longer network lifetime, and 15% reduced overhead.
- Open Access
- 19 December 2024
- Published by Association for Computing Machinery (ACM)
- p. 96-103
An Efficient Burst Traffic-Aware Clustering and Mobility Management model for mobile-sink based Heterogeneous Wireless Sensor Networks
- 19 December 2024
- Published by Association for Computing Machinery (ACM)
- p. 96-103
Open AccessAn Efficient Burst Traffic-Aware Clustering and Mobility Management model for mobile-sink based Heterogeneous Wireless Sensor Networks
- Top Cited PapersOpen Access
- 19 March 2024
- Published by Springer Nature in International Journal of Information Security
- Vol. 23 (3) , 2139-2158
MLSTL-WSN: machine learning-based intrusion detection using SMOTETomek in WSNs
In the domain of cyber-physical systems, wireless sensor networks (WSNs) play a pivotal role as infrastructures, encompassing both stationary and mobile sensors. These sensors self-organize and establish multi-hop connections for communication, collectively sensing, gathering, processing, and transmitting data about their surroundings. Despite their significance, WSNs face rapid and detrimental attacks that can disrupt functionality. Existing intrusion detection methods for WSNs encounter challenges such as low detection rates, computational overhead, and false alarms. These issues stem from sensor node resource constraints, data redundancy, and high correlation within the network. To address these challenges, we propose an innovative intrusion detection approach that integrates machine learning (ML) techniques with the Synthetic Minority Oversampling Technique Tomek Link (SMOTE-TomekLink) algorithm. This blend synthesizes minority instances and eliminates Tomek links, resulting in a balanced dataset that significantly enhances detection accuracy in WSNs. Additionally, we incorporate feature scaling through standardization to render input features consistent and scalable, facilitating more precise training and detection. To counteract imbalanced WSN datasets, we employ the SMOTE-Tomek resampling technique, mitigating overfitting and underfitting issues. Our comprehensive evaluation, using the wireless sensor network dataset (WSN-DS) containing 374,661 records, identifies the optimal model for intrusion detection in WSNs. The standout outcome of our research is the remarkable performance of our model. In binary classification scenarios, it achieves an accuracy rate of 99.78%, and in multiclass classification scenarios, it attains an exceptional accuracy rate of 99.92%. These findings underscore the efficiency and superiority of our proposal in the context of WSN intrusion detection, showcasing its effectiveness in detecting and mitigating intrusions in WSNs.
- 19 March 2024
- Published by Springer Nature in International Journal of Information Security
- Vol. 23 (3) , 2139-2158
Top Cited PapersOpen AccessMLSTL-WSN: machine learning-based intrusion detection using SMOTETomek in WSNs
In the domain of cyber-physical systems, wireless sensor networks (WSNs) play a pivotal role as infrastructures, encompassing both stationary and mobile sensors. These sensors self-organize and establish multi-hop connections for communication, collectively sensing, gathering, processing, and transmitting data about their surroundings. Despite their significance, WSNs face rapid and detrimental attacks that can disrupt functionality. Existing intrusion detection methods for WSNs encounter challenges such as low detection rates, computational overhead, and false alarms. These issues stem from sensor node resource constraints, data redundancy, and high correlation within the network. To address these challenges, we propose an innovative intrusion detection approach that integrates machine learning (ML) techniques with the Synthetic Minority Oversampling Technique Tomek Link (SMOTE-TomekLink) algorithm. This blend synthesizes minority instances and eliminates Tomek links, resulting in a balanced dataset that significantly enhances detection accuracy in WSNs. Additionally, we incorporate feature scaling through standardization to render input features consistent and scalable, facilitating more precise training and detection. To counteract imbalanced WSN datasets, we employ the SMOTE-Tomek resampling technique, mitigating overfitting and underfitting issues. Our comprehensive evaluation, using the wireless sensor network dataset (WSN-DS) containing 374,661 records, identifies the optimal model for intrusion detection in WSNs. The standout outcome of our research is the remarkable performance of our model. In binary classification scenarios, it achieves an accuracy rate of 99.78%, and in multiclass classification scenarios, it attains an exceptional accuracy rate of 99.92%. These findings underscore the efficiency and superiority of our proposal in the context of WSN intrusion detection, showcasing its effectiveness in detecting and mitigating intrusions in WSNs.
- Open Access
MLSTL-WSN: Machine Learning-based Intrusion Detection using SMOTETomek in WSNs
Wireless Sensor Networks (WSNs) play a pivotal role as infrastructures, encompassing both stationary and mobile sensors. These sensors self-organize and establish multi-hop connections for communication, collectively sensing, gathering, processing, and transmitting data about their surroundings. Despite their significance, WSNs face rapid and detrimental attacks that can disrupt functionality. Existing intrusion detection methods for WSNs encounter challenges such as low detection rates, computational overhead, and false alarms. These issues stem from sensor node resource constraints, data redundancy, and high correlation within the network. To address these challenges, we propose an innovative intrusion detection approach that integrates Machine Learning (ML) techniques with the Synthetic Minority Oversampling Technique Tomek Link (SMOTE-TomekLink) algorithm. This blend synthesizes minority instances and eliminates Tomek links, resulting in a balanced dataset that significantly enhances detection accuracy in WSNs. Additionally, we incorporate feature scaling through standardization to render input features consistent and scalable, facilitating more precise training and detection. To counteract imbalanced WSN datasets, we employ the SMOTE-Tomek resampling technique, mitigating overfitting and underfitting issues. Our comprehensive evaluation, using the WSN Dataset (WSN-DS) containing 374,661 records, identifies the optimal model for intrusion detection in WSNs. The standout outcome of our research is the remarkable performance of our model. In binary, it achieves an accuracy rate of 99.78% and in multiclass, it attains an exceptional accuracy rate of 99.92%. These findings underscore the efficiency and superiority of our proposal in the context of WSN intrusion detection, showcasing its effectiveness in detecting and mitigating intrusions in WSNs.
Open AccessMLSTL-WSN: Machine Learning-based Intrusion Detection using SMOTETomek in WSNs
Wireless Sensor Networks (WSNs) play a pivotal role as infrastructures, encompassing both stationary and mobile sensors. These sensors self-organize and establish multi-hop connections for communication, collectively sensing, gathering, processing, and transmitting data about their surroundings. Despite their significance, WSNs face rapid and detrimental attacks that can disrupt functionality. Existing intrusion detection methods for WSNs encounter challenges such as low detection rates, computational overhead, and false alarms. These issues stem from sensor node resource constraints, data redundancy, and high correlation within the network. To address these challenges, we propose an innovative intrusion detection approach that integrates Machine Learning (ML) techniques with the Synthetic Minority Oversampling Technique Tomek Link (SMOTE-TomekLink) algorithm. This blend synthesizes minority instances and eliminates Tomek links, resulting in a balanced dataset that significantly enhances detection accuracy in WSNs. Additionally, we incorporate feature scaling through standardization to render input features consistent and scalable, facilitating more precise training and detection. To counteract imbalanced WSN datasets, we employ the SMOTE-Tomek resampling technique, mitigating overfitting and underfitting issues. Our comprehensive evaluation, using the WSN Dataset (WSN-DS) containing 374,661 records, identifies the optimal model for intrusion detection in WSNs. The standout outcome of our research is the remarkable performance of our model. In binary, it achieves an accuracy rate of 99.78% and in multiclass, it attains an exceptional accuracy rate of 99.92%. These findings underscore the efficiency and superiority of our proposal in the context of WSN intrusion detection, showcasing its effectiveness in detecting and mitigating intrusions in WSNs.
- Open Access
Machine learning-based network intrusion detection for big and imbalanced data using oversampling, stacking feature embedding and feature extraction
Cybersecurity has emerged as a critical global concern. Intrusion Detection Systems (IDS) play a critical role in protecting interconnected networks by detecting malicious actors and activities. Machine Learning (ML)-based behavior analysis within the IDS has considerable potential for detecting dynamic cyber threats, identifying abnormalities, and identifying malicious conduct within the network. However, as the number of data grows, dimension reduction becomes an increasingly difficult task when training ML models. Addressing this, our paper introduces a novel ML-based network intrusion detection model that uses Random Oversampling (RO) to address data imbalance and Stacking Feature Embedding based on clustering results, as well as Principal Component Analysis (PCA) for dimension reduction and is specifically designed for large and imbalanced datasets. This model's performance is carefully evaluated using three cutting-edge benchmark datasets: UNSW-NB15, CIC-IDS-2017, and CIC-IDS-2018. On the UNSW-NB15 dataset, our trials show that the RF and ET models achieve accuracy rates of 99.59% and 99.95%, respectively. Furthermore, using the CIC-IDS2017 dataset, DT, RF, and ET models reach 99.99% accuracy, while DT and RF models obtain 99.94% accuracy on CIC-IDS2018. These performance results continuously outperform the state-of-art, indicating significant progress in the field of network intrusion detection. This achievement demonstrates the efficacy of the suggested methodology, which can be used practically to accurately monitor and identify network traffic intrusions, thereby blocking possible threats.
Open AccessMachine learning-based network intrusion detection for big and imbalanced data using oversampling, stacking feature embedding and feature extraction
Cybersecurity has emerged as a critical global concern. Intrusion Detection Systems (IDS) play a critical role in protecting interconnected networks by detecting malicious actors and activities. Machine Learning (ML)-based behavior analysis within the IDS has considerable potential for detecting dynamic cyber threats, identifying abnormalities, and identifying malicious conduct within the network. However, as the number of data grows, dimension reduction becomes an increasingly difficult task when training ML models. Addressing this, our paper introduces a novel ML-based network intrusion detection model that uses Random Oversampling (RO) to address data imbalance and Stacking Feature Embedding based on clustering results, as well as Principal Component Analysis (PCA) for dimension reduction and is specifically designed for large and imbalanced datasets. This model's performance is carefully evaluated using three cutting-edge benchmark datasets: UNSW-NB15, CIC-IDS-2017, and CIC-IDS-2018. On the UNSW-NB15 dataset, our trials show that the RF and ET models achieve accuracy rates of 99.59% and 99.95%, respectively. Furthermore, using the CIC-IDS2017 dataset, DT, RF, and ET models reach 99.99% accuracy, while DT and RF models obtain 99.94% accuracy on CIC-IDS2018. These performance results continuously outperform the state-of-art, indicating significant progress in the field of network intrusion detection. This achievement demonstrates the efficacy of the suggested methodology, which can be used practically to accurately monitor and identify network traffic intrusions, thereby blocking possible threats.
- Top Cited PapersOpen Access
- 14 August 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 11, 87694-87708
A Hybrid Dependable Deep Feature Extraction and Ensemble-Based Machine Learning Approach for Breast Cancer Detection
Breast cancer is a prevalent and life-threatening disease that requires effective detection and diagnosis methods to improve patient outcomes. Deep learning (DL) and machine learning (ML) techniques have emerged as powerful tools in breast cancer detection, offering benefits such as improved accuracy and efficiency. However, existing methods have scalability and performance limitations, emphasizing the need for further research. In this paper, we propose a hybrid dependable breast cancer detection approach that combines the power of DL using a pre-trained ResNet50V2 model and ensemble-based ML methods. The integration of DL enables the approach to learn and extract hidden patterns from complex breast cancer images, while ML algorithms contribute interpretability and generalization capabilities. We conducted extensive experiments using a breast histopathology image-based publicly available Invasive Ductal Carcinoma (IDC) dataset comprising samples of different sizes. The results obtained from our rigorous experiments provide compelling evidence for our hybrid model’s robustness and high performance. We achieved a higher accuracy rate of 95%, precision of 94.86%, recall of 94.32%, and F1 score of 94.57% compared to state-of-the-art models. We also identified Light Boosting Classifier (LGB) as the most suitable ML model in conjunction with the ResNet50V2 architecture. The results of this research offer significant contributions to breast cancer detection through an innovative approach, comprehensive performance analysis, and dependable assessment. Moreover, it has the potential to assist medical professionals in making informed decisions, improving patient care, and enhancing outcomes for breast cancer patients.
- 14 August 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 11, 87694-87708
Top Cited PapersOpen AccessA Hybrid Dependable Deep Feature Extraction and Ensemble-Based Machine Learning Approach for Breast Cancer Detection
Breast cancer is a prevalent and life-threatening disease that requires effective detection and diagnosis methods to improve patient outcomes. Deep learning (DL) and machine learning (ML) techniques have emerged as powerful tools in breast cancer detection, offering benefits such as improved accuracy and efficiency. However, existing methods have scalability and performance limitations, emphasizing the need for further research. In this paper, we propose a hybrid dependable breast cancer detection approach that combines the power of DL using a pre-trained ResNet50V2 model and ensemble-based ML methods. The integration of DL enables the approach to learn and extract hidden patterns from complex breast cancer images, while ML algorithms contribute interpretability and generalization capabilities. We conducted extensive experiments using a breast histopathology image-based publicly available Invasive Ductal Carcinoma (IDC) dataset comprising samples of different sizes. The results obtained from our rigorous experiments provide compelling evidence for our hybrid model’s robustness and high performance. We achieved a higher accuracy rate of 95%, precision of 94.86%, recall of 94.32%, and F1 score of 94.57% compared to state-of-the-art models. We also identified Light Boosting Classifier (LGB) as the most suitable ML model in conjunction with the ResNet50V2 architecture. The results of this research offer significant contributions to breast cancer detection through an innovative approach, comprehensive performance analysis, and dependable assessment. Moreover, it has the potential to assist medical professionals in making informed decisions, improving patient care, and enhancing outcomes for breast cancer patients.
- 10 October 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Mobile Computing
- Vol. 23 (1) , 29-41
A Hyper Heuristic Algorithm for Efficient Resource Allocation in 5G Mobile Edge Clouds
Emergence of intelligent devices and mobile edge clouds (MECs) in 5G networks has exponentially increased the number of applications that demand low latency services. However, their resource heterogeneity, limited computing power and storage including congestion in the ultra-dense 5G network, make the real-time services challenging. Existing works are limited either by addressing application delay requirements or computational load balancing. This article develops an efficient resource allocation framework for selecting optimal servers and routing paths in the 5G MEC network by jointly optimizing latency, computational, and network load variances. First, we formulate the above multi-objective problem as a mixed-integer non-linear programming problem. Further, we adopt a hyper-heuristic (AWSH) algorithm by leveraging the combined powers of A nt Colony, W hale, S ine-Cosine, and H enry Gas Solubility Optimization algorithms. The proposed AWSH algorithm works at the higher level, and it explores and exploits one of the three lower-level heuristics in each iteration to efficiently capture the dynamically varying environmental parameters and thereby address the resource allocation problem. Their collaborative effort helps to achieve a global optimum in allocating resources of 5G MEC network. Simulation results prove the superiority of the AWSH algorithm compared to state-of-the-art solutions in terms of service latency, successful offloading ratio, and load balancing.
- 10 October 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Mobile Computing
- Vol. 23 (1) , 29-41
A Hyper Heuristic Algorithm for Efficient Resource Allocation in 5G Mobile Edge Clouds
Emergence of intelligent devices and mobile edge clouds (MECs) in 5G networks has exponentially increased the number of applications that demand low latency services. However, their resource heterogeneity, limited computing power and storage including congestion in the ultra-dense 5G network, make the real-time services challenging. Existing works are limited either by addressing application delay requirements or computational load balancing. This article develops an efficient resource allocation framework for selecting optimal servers and routing paths in the 5G MEC network by jointly optimizing latency, computational, and network load variances. First, we formulate the above multi-objective problem as a mixed-integer non-linear programming problem. Further, we adopt a hyper-heuristic (AWSH) algorithm by leveraging the combined powers of A nt Colony, W hale, S ine-Cosine, and H enry Gas Solubility Optimization algorithms. The proposed AWSH algorithm works at the higher level, and it explores and exploits one of the three lower-level heuristics in each iteration to efficiently capture the dynamically varying environmental parameters and thereby address the resource allocation problem. Their collaborative effort helps to achieve a global optimum in allocating resources of 5G MEC network. Simulation results prove the superiority of the AWSH algorithm compared to state-of-the-art solutions in terms of service latency, successful offloading ratio, and load balancing.