B. Karthikeyan
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
B. Karthikeyan has a total of 34 co-authors. Here are the top co-authors by number of shared publications:
J. Anand
School of Electronics Engineering, Vellore Institute of Technology, Vellore, India
3 shared publications- V. Mahalakshmi
School of Electronics Engineering, Vellore Institute of Technology, Thiruvalam Road, Vellore 632014, Tamil Nadu, India
3 shared publications A Aishwariya
School of Electronics Engineering, VIT University, Vellore 632014, Tamil Nadu, India
1 shared publication- Abbas Z. Kouzani
School of Engineering, Deakin University, Geelong, VIC 3216, Australia
1 shared publication Akshay Chandak
School of Electronics Engineering, VIT University, Vellore, Tamil Nadu 632014, India
1 shared publicationAneeshsingh Bhatkhande
1 shared publication
Latest Publications
- Top Cited PapersOpen Access
- 7 February 2026
- Published by Springer Nature in Scientific Reports
- Vol. 16 (1)
Adaptive and intelligent customized deep Q-network for energy-efficient task offloading in mobile edge computing environments
The rapid expansion of edge-cloud infrastructures and latency-sensitive Internet of Things (IoT) applications has intensified the challenge of intelligent task offloading in dynamic and resource-constrained environments. This paper presents an Adaptive and Intelligent Customized Deep Q-Network (AICDQN), a novel reinforcement learning-based framework for real-time, priority-aware task scheduling in mobile edge computing systems. The proposed model formulates task offloading as a Markov Decision Process (MDP) and integrates a hybrid Gated Recurrent Unit-Long Short-Term Memory (GRU-LSTM) load prediction module to forecast workload fluctuations and task urgency trends. This foresight enables a Dynamic Dueling Double Deep Q-Network \((\text {D}^{4}\text {QN})\) agent to make informed offloading decisions across local, edge, and cloud tiers. The system models compute nodes using priority-aware M/M/1, M/M/c and M/M/\(\infty\) queuing systems, enabling delay-sensitive and queue-aware decision-making. A dynamic priority scoring function integrates task urgency, deadline proximity, and node-level queue saturation, ensuring real-time tasks are prioritized effectively. Furthermore, an energy-aware scheduling policy proactively transitions underutilized servers into low-power states without compromising performance. Extensive simulations demonstrate that AICDQN achieves up to 33.39% reduction in delay, 57.74% improvement in energy efficiency, and 81.25% reduction in task drop rate compared with existing offloading algorithms, including Deep Deterministic Policy Gradient (DDPG), Distributed Dynamic Task Offloading (DDTO-DRL), Potential Game based Offloading Algorithm (PGOA), and the User-Level Online Offloading Framework (ULOOF). These results validate AICDQN as a scalable and adaptive solution for next-generation edge-cloud systems requiring efficient, intelligent, and energy-constrained task offloading.- 7 February 2026
- Published by Springer Nature in Scientific Reports
- Vol. 16 (1)
Top Cited PapersOpen AccessAdaptive and intelligent customized deep Q-network for energy-efficient task offloading in mobile edge computing environments
The rapid expansion of edge-cloud infrastructures and latency-sensitive Internet of Things (IoT) applications has intensified the challenge of intelligent task offloading in dynamic and resource-constrained environments. This paper presents an Adaptive and Intelligent Customized Deep Q-Network (AICDQN), a novel reinforcement learning-based framework for real-time, priority-aware task scheduling in mobile edge computing systems. The proposed model formulates task offloading as a Markov Decision Process (MDP) and integrates a hybrid Gated Recurrent Unit-Long Short-Term Memory (GRU-LSTM) load prediction module to forecast workload fluctuations and task urgency trends. This foresight enables a Dynamic Dueling Double Deep Q-Network \((\text {D}^{4}\text {QN})\) agent to make informed offloading decisions across local, edge, and cloud tiers. The system models compute nodes using priority-aware M/M/1, M/M/c and M/M/\(\infty\) queuing systems, enabling delay-sensitive and queue-aware decision-making. A dynamic priority scoring function integrates task urgency, deadline proximity, and node-level queue saturation, ensuring real-time tasks are prioritized effectively. Furthermore, an energy-aware scheduling policy proactively transitions underutilized servers into low-power states without compromising performance. Extensive simulations demonstrate that AICDQN achieves up to 33.39% reduction in delay, 57.74% improvement in energy efficiency, and 81.25% reduction in task drop rate compared with existing offloading algorithms, including Deep Deterministic Policy Gradient (DDPG), Distributed Dynamic Task Offloading (DDTO-DRL), Potential Game based Offloading Algorithm (PGOA), and the User-Level Online Offloading Framework (ULOOF). These results validate AICDQN as a scalable and adaptive solution for next-generation edge-cloud systems requiring efficient, intelligent, and energy-constrained task offloading. - Open Access
- 1 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 14, 135233-135253
Adversarial Attacks in Autonomous Vehicles: A Comprehensive Review of Threats and Defence Strategies
- 1 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 14, 135233-135253
Open AccessAdversarial Attacks in Autonomous Vehicles: A Comprehensive Review of Threats and Defence Strategies
- Open Access
- 1 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 14, 101586-101615
NSCT-RSFound: A Dynamically Weighted Multi-Stream Vision Transformer for Robust Remote Sensing Scene Classification
- 1 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 14, 101586-101615
Open AccessNSCT-RSFound: A Dynamically Weighted Multi-Stream Vision Transformer for Robust Remote Sensing Scene Classification
- 1 December 2025
- Published by World Scientific Pub Co Pte Ltd in International Journal on Artificial Intelligence Tools
- Vol. 34 (07n08)
Context-Aware Acquisition Module using Vehicle to Roadside Unit Unicast Communication for Intelligent Transport System in Edge Computing
Intelligent transportation systems (ITSs) are evolving with the integration of edge computing and the Internet of vehicles (IoVs), enabling responsive, data-driven traffic management. This paper proposes a context-aware acquisition module (CAAM) to enhance real-time vehicular decision-making using unicast communication between vehicles and roadside units (RSUs). Unlike conventional broadcast approaches, CAAM selectively acquires and transmits critical contextual information such as vehicle speed, acceleration, and [ ] emissions, thereby reducing communication overhead and improving response accuracy. The system architecture incorporates RSUs, edge servers, and on-board units (OBUs), with RSUs periodically transmitting wave short messages (WSMs) using the IEEE 802.11p WAVE protocol. CAAM dynamically filters context data and enables targeted, low-latency control actions in complex urban environments. Simulations are conducted using OMNeT++ and VEINS, integrated with SUMO for realistic vehicular mobility and communication modeling. Performance is evaluated using key metrics, including vehicle speed, acceleration profiles, [ ] emissions, communication busy time, and packet delivery ratio (PDR). In a representative scenario, a vehicle responded to an RSU-issued deceleration command, reducing its speed from 13.89[ ]m/sec to 1.38[ ]m/sec. CAAM achieved the lowest end-to-end delay (0.48[ ]sec) and highest PDR (100%), outperforming adaptive V2R, shortest distance, and random selection strategies, supporting its alignment with emerging 5G-V2X standards.
- 1 December 2025
- Published by World Scientific Pub Co Pte Ltd in International Journal on Artificial Intelligence Tools
- Vol. 34 (07n08)
Context-Aware Acquisition Module using Vehicle to Roadside Unit Unicast Communication for Intelligent Transport System in Edge Computing
Intelligent transportation systems (ITSs) are evolving with the integration of edge computing and the Internet of vehicles (IoVs), enabling responsive, data-driven traffic management. This paper proposes a context-aware acquisition module (CAAM) to enhance real-time vehicular decision-making using unicast communication between vehicles and roadside units (RSUs). Unlike conventional broadcast approaches, CAAM selectively acquires and transmits critical contextual information such as vehicle speed, acceleration, and [ ] emissions, thereby reducing communication overhead and improving response accuracy. The system architecture incorporates RSUs, edge servers, and on-board units (OBUs), with RSUs periodically transmitting wave short messages (WSMs) using the IEEE 802.11p WAVE protocol. CAAM dynamically filters context data and enables targeted, low-latency control actions in complex urban environments. Simulations are conducted using OMNeT++ and VEINS, integrated with SUMO for realistic vehicular mobility and communication modeling. Performance is evaluated using key metrics, including vehicle speed, acceleration profiles, [ ] emissions, communication busy time, and packet delivery ratio (PDR). In a representative scenario, a vehicle responded to an RSU-issued deceleration command, reducing its speed from 13.89[ ]m/sec to 1.38[ ]m/sec. CAAM achieved the lowest end-to-end delay (0.48[ ]sec) and highest PDR (100%), outperforming adaptive V2R, shortest distance, and random selection strategies, supporting its alignment with emerging 5G-V2X standards.
- 7 October 2025
- Published by IOP Publishing in Engineering Research Express
- Vol. 7 (4) , 045310
Context reasoning-based vehicle traffic accident severity prediction using a machine learning algorithm in edge computing
The rapid development of the Internet of Things (IoT) has transformed intelligent transportation systems. However, traffic accidents remain a significant challenge, and it is imperative to develop effective strategies to reduce their frequency and severity. This research introduces a hybrid model that integrates Adaptive Bayesian Optimization (ABO) with Random Forest (RF) to enhance the prediction of traffic accident severity. By leveraging ABO to optimize RF parameters, the model achieves superior accuracy compared to conventional approaches while maintaining interpretability through relative importance and partial dependence plots. Furthermore, utilizing real-time vehicle data, the system implements context-based vehicle accident detection within vehicular ad hoc networks (VANETs). The integration of Enhanced Local Adaptive Machine Learning (ELAML), RF, and ABO-RF demonstrates robust performance in collision detection, outperforming models such as Support Vector Machines (SVM) and Artificial Neural Networks (ANN), with a detection accuracy of 90%. Compared to existing models, the proposed ABO-RF achieved the highest accuracy (90%), perfect recall (100%), and the best F1 score (94.74%), with lower error rate (10%), FAR (6.00%), MDR (25.00%), and detection latency (25.24%) than conventional methods such as MLP, ResNet-50, AdaBoost and XGBoost. These findings offer valuable insights into improving road safety and mitigating accident severity in the Internet of Vehicles (IoV)-enabled intelligent transportation systems.
- 7 October 2025
- Published by IOP Publishing in Engineering Research Express
- Vol. 7 (4) , 045310
Context reasoning-based vehicle traffic accident severity prediction using a machine learning algorithm in edge computing
The rapid development of the Internet of Things (IoT) has transformed intelligent transportation systems. However, traffic accidents remain a significant challenge, and it is imperative to develop effective strategies to reduce their frequency and severity. This research introduces a hybrid model that integrates Adaptive Bayesian Optimization (ABO) with Random Forest (RF) to enhance the prediction of traffic accident severity. By leveraging ABO to optimize RF parameters, the model achieves superior accuracy compared to conventional approaches while maintaining interpretability through relative importance and partial dependence plots. Furthermore, utilizing real-time vehicle data, the system implements context-based vehicle accident detection within vehicular ad hoc networks (VANETs). The integration of Enhanced Local Adaptive Machine Learning (ELAML), RF, and ABO-RF demonstrates robust performance in collision detection, outperforming models such as Support Vector Machines (SVM) and Artificial Neural Networks (ANN), with a detection accuracy of 90%. Compared to existing models, the proposed ABO-RF achieved the highest accuracy (90%), perfect recall (100%), and the best F1 score (94.74%), with lower error rate (10%), FAR (6.00%), MDR (25.00%), and detection latency (25.24%) than conventional methods such as MLP, ResNet-50, AdaBoost and XGBoost. These findings offer valuable insights into improving road safety and mitigating accident severity in the Internet of Vehicles (IoV)-enabled intelligent transportation systems.
- Open Access
- 27 June 2025
- Published by Elsevier in Results in Engineering
- Vol. 27
EADRL: Efficiency-aware adaptive deep reinforcement learning for dynamic task scheduling in edge-cloud environments
- 27 June 2025
- Published by Elsevier in Results in Engineering
- Vol. 27
Open AccessEADRL: Efficiency-aware adaptive deep reinforcement learning for dynamic task scheduling in edge-cloud environments
- Open Access
- 13 February 2025
- Published by Elsevier in Results in Engineering
- Vol. 25
Dynamic priority-based task scheduling and adaptive resource allocation algorithms for efficient edge computing in healthcare systems
- 13 February 2025
- Published by Elsevier in Results in Engineering
- Vol. 25
Open AccessDynamic priority-based task scheduling and adaptive resource allocation algorithms for efficient edge computing in healthcare systems
- Open Access
- 15 March 2024
- Published by MDPI AG in Engineering Proceedings
- Vol. 62 (1) , 17
Edge Computing in Context Awareness: A Comprehensive Study
Mobile edge computing (MEC), which is now gaining a lot of momentum, allows users to use its services with low latency, location awareness, and mobility assistance to offset the drawbacks of cloud computing. The quality of the experience, reduced latency, and boosted performance are the ultimate context-aware goals. There have been many context-aware efforts in the past. In this study, we reviewed many elements of the proposed context-aware approach for edge cloud computing, including their benefits and drawbacks. Additionally, we looked at such context-aware techniques to determine which were practical given the situation. We anticipate that the survey will be carefully considered in the creation of new context-aware methods. Future directions and problems have also been looked at.
- 15 March 2024
- Published by MDPI AG in Engineering Proceedings
- Vol. 62 (1) , 17
Open AccessEdge Computing in Context Awareness: A Comprehensive Study
Mobile edge computing (MEC), which is now gaining a lot of momentum, allows users to use its services with low latency, location awareness, and mobility assistance to offset the drawbacks of cloud computing. The quality of the experience, reduced latency, and boosted performance are the ultimate context-aware goals. There have been many context-aware efforts in the past. In this study, we reviewed many elements of the proposed context-aware approach for edge cloud computing, including their benefits and drawbacks. Additionally, we looked at such context-aware techniques to determine which were practical given the situation. We anticipate that the survey will be carefully considered in the creation of new context-aware methods. Future directions and problems have also been looked at.
- Open Access
- 29 February 2024
- Published by MDPI AG in Engineering Proceedings
- Vol. 62 (1) , 5
Self-Adaptive Waste Management System: Utilizing Convolutional Neural Networks for Real-Time Classification
This research presents a novel Self-Adaptive Waste Management System (SAWMS) that integrates advanced technology to address the pressing challenges of waste sorting and classification. SAWMS leverages Convolutional Neural Networks (CNNs) in conjunction with conveyor belt technology to achieve real-time object classification and self-training capabilities. The system utilizes sensors for object detection and a camera for image capture, enabling an accurate initial classification of waste objects into predefined categories such as food waste, metal, and plastic bottles. Notably, our proposed system sets itself apart by its unique ability to adapt and self-train based on classification errors, ensuring ongoing accuracy even in the face of changing waste compositions. Through dynamic adjustments of the conveyor belt’s destination, it efficiently directs waste objects to their appropriate bins for disposal or recycling. This research demonstrates the potential of SAWMS to revolutionize waste management practices, offering an agile and sustainable solution to the evolving challenges of waste sorting and disposal.
- 29 February 2024
- Published by MDPI AG in Engineering Proceedings
- Vol. 62 (1) , 5
Open AccessSelf-Adaptive Waste Management System: Utilizing Convolutional Neural Networks for Real-Time Classification
This research presents a novel Self-Adaptive Waste Management System (SAWMS) that integrates advanced technology to address the pressing challenges of waste sorting and classification. SAWMS leverages Convolutional Neural Networks (CNNs) in conjunction with conveyor belt technology to achieve real-time object classification and self-training capabilities. The system utilizes sensors for object detection and a camera for image capture, enabling an accurate initial classification of waste objects into predefined categories such as food waste, metal, and plastic bottles. Notably, our proposed system sets itself apart by its unique ability to adapt and self-train based on classification errors, ensuring ongoing accuracy even in the face of changing waste compositions. Through dynamic adjustments of the conveyor belt’s destination, it efficiently directs waste objects to their appropriate bins for disposal or recycling. This research demonstrates the potential of SAWMS to revolutionize waste management practices, offering an agile and sustainable solution to the evolving challenges of waste sorting and disposal.