Maira Khalid
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Maira Khalid has a total of 23 co-authors. Here are the top co-authors by number of shared publications:
- Jehad Ali
Ajou University
8 shared publications - Byeong-Hee Roh
Ajou University
7 shared publications - Ahmed Raza Mohsin
Department of AI Convergence Network, Ajou University, Suwon, South Korea
3 shared publications - Ahmed Raza Mohsin
Department of AI Convergence Network, Ajou University, Suwon, South Korea
2 shared publications - Abdul Razzaq
Institute of Computing Muhammad Nawaz Sharif University of Agriculture Multan Punjab Pakistan
1 shared publication Ahmed Raza Mohsin
Department of AI Convergence Network, Ajou University, Suwon, South Korea
1 shared publication
Latest Publications
- 1 November 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Consumer Electronics
- Vol. 72 (4) , 10811-10822
An Effective Continual Few-Shot Video Learning for Dynamic and Resource-Limited Consumer Edge Environments
- 1 November 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Consumer Electronics
- Vol. 72 (4) , 10811-10822
An Effective Continual Few-Shot Video Learning for Dynamic and Resource-Limited Consumer Edge Environments
- 20 October 2026
- Published by Springer Nature in Wireless Networks
- Vol. 32 (6) , 3735-3753
Privacy-preserving federated learning with homomorphic encryption and LLM based semantic adaptation for smart home activity recognition
- 20 October 2026
- Published by Springer Nature in Wireless Networks
- Vol. 32 (6) , 3735-3753
Privacy-preserving federated learning with homomorphic encryption and LLM based semantic adaptation for smart home activity recognition
- Open Access
- 8 September 2026
- Published by MDPI AG in Multimedia
- Vol. 2 (3) , 15
DMART-HAR: Dynamic Multimodal Transformer Learning for Cross-Domain Human Activity Recognition
Human activity recognition (HAR) in smart environments plays a critical role in applications such as healthcare monitoring, intelligent transportation systems, and ambient assisted living; however, existing approaches are limited by their inability to effectively handle heterogeneous multimodal sensor data, capture long-range temporal dependencies, and generalize across diverse real-world environments under domain shifts. In this work, we present DMART-HAR, a Dynamic Multimodal Activity Recognition Transformer framework that unifies structured multimodal representation learning, transformer-based temporal modeling, cross-modal interaction, and adversarial domain adaptation within a single architecture. Specifically, the developed method introduces a sensor tokenization mechanism to encode heterogeneous IoT data into a unified representation space, followed by a transformer encoder to capture global contextual dependencies, while a cross-modal attention module enables deep interaction among sensor modalities and an adversarial domain adaptation strategy enhances robustness to unseen environments. Extensive experiments on benchmark datasets, including CASAS, PAMAP2, and Opportunity, demonstrate that DMART-HAR consistently outperforms both conventional baselines and recent state-of-the-art methods, achieving accuracy/F1-scores of 94.3%/92.8%, 96.2%/94.7%, and 89.8%/88.1%, respectively, and consistently outperforms the strongest competing approaches under cross-domain evaluation settings. These findings demonstrate the effectiveness of modeling temporal dynamics, multimodal relationships, and domain invariance simultaneously, establishing DMART-HAR as a scalable and robust solution for real-world HAR applications.
- 8 September 2026
- Published by MDPI AG in Multimedia
- Vol. 2 (3) , 15
Open AccessDMART-HAR: Dynamic Multimodal Transformer Learning for Cross-Domain Human Activity Recognition
Human activity recognition (HAR) in smart environments plays a critical role in applications such as healthcare monitoring, intelligent transportation systems, and ambient assisted living; however, existing approaches are limited by their inability to effectively handle heterogeneous multimodal sensor data, capture long-range temporal dependencies, and generalize across diverse real-world environments under domain shifts. In this work, we present DMART-HAR, a Dynamic Multimodal Activity Recognition Transformer framework that unifies structured multimodal representation learning, transformer-based temporal modeling, cross-modal interaction, and adversarial domain adaptation within a single architecture. Specifically, the developed method introduces a sensor tokenization mechanism to encode heterogeneous IoT data into a unified representation space, followed by a transformer encoder to capture global contextual dependencies, while a cross-modal attention module enables deep interaction among sensor modalities and an adversarial domain adaptation strategy enhances robustness to unseen environments. Extensive experiments on benchmark datasets, including CASAS, PAMAP2, and Opportunity, demonstrate that DMART-HAR consistently outperforms both conventional baselines and recent state-of-the-art methods, achieving accuracy/F1-scores of 94.3%/92.8%, 96.2%/94.7%, and 89.8%/88.1%, respectively, and consistently outperforms the strongest competing approaches under cross-domain evaluation settings. These findings demonstrate the effectiveness of modeling temporal dynamics, multimodal relationships, and domain invariance simultaneously, establishing DMART-HAR as a scalable and robust solution for real-world HAR applications.
- 21 August 2026
- Published by Springer Nature in Telecommunication Systems
- Vol. 89 (3) , 124
An efficient software-defined WiFi-based multi-objective access point load balancing
- 21 August 2026
- Published by Springer Nature in Telecommunication Systems
- Vol. 89 (3) , 124
An efficient software-defined WiFi-based multi-objective access point load balancing
- 1 April 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Intelligent Transportation Systems
- Vol. PP (15249050) , 1-18
STEP-SAFE: Smart TinyML and Few-Shot Enabled Pedestrian and Cyclist Safety Framework for Intelligent Transportation System
Ensuring the safety of pedestrians and cyclists referred to as vulnerable road users (VRUs) is a central challenge in intelligent transportation systems (ITS), especially in urban environments where visual occlusion, behavioral unpredictability, and data scarcity pose significant barriers to real-time detection. This paper introduces STEP-SAFE, a novel few-shot learning framework optimized for TinyML platforms that enables accurate and efficient VRU recognition and alert generation with minimal supervision and extremely low computational overhead. The proposed model utilizes a prototype based meta learning architecture in combination with a lightweight convolutional encoder to support rapid generalization from as few as one to five labeled samples per class. To simulate embedded deployment, we integrate quantization aware training and sparsity based pruning, reducing the model size to under 300 KB and enabling inference latency below 64 milliseconds on micro-controller class hardware. Additionally, STEP-SAFE incorporates a dynamic risk aware alert mechanism that converts classification confidence into real-time safety decisions. The model is evaluated across multiple few-shot configurations (1-shot to 10-shot, 3-way to 5-way) and benchmarked against state-of-the-art methods. Moreover, results show that STEP-SAFE achieves up to 85.2% accuracy and 0.85mAP while requiring significantly fewer computational resources and support samples. Confusion matrix and episodic performance analyses confirm its robustness in distinguishing visually similar and occlusion prone VRU behaviors. By combining compact model design, few-shot adaptability, and interpretable alert logic, STEP-SAFE represents a deployable and intelligent solution for future ready ITS safety infrastructure.
- 1 April 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Intelligent Transportation Systems
- Vol. PP (15249050) , 1-18
STEP-SAFE: Smart TinyML and Few-Shot Enabled Pedestrian and Cyclist Safety Framework for Intelligent Transportation System
Ensuring the safety of pedestrians and cyclists referred to as vulnerable road users (VRUs) is a central challenge in intelligent transportation systems (ITS), especially in urban environments where visual occlusion, behavioral unpredictability, and data scarcity pose significant barriers to real-time detection. This paper introduces STEP-SAFE, a novel few-shot learning framework optimized for TinyML platforms that enables accurate and efficient VRU recognition and alert generation with minimal supervision and extremely low computational overhead. The proposed model utilizes a prototype based meta learning architecture in combination with a lightweight convolutional encoder to support rapid generalization from as few as one to five labeled samples per class. To simulate embedded deployment, we integrate quantization aware training and sparsity based pruning, reducing the model size to under 300 KB and enabling inference latency below 64 milliseconds on micro-controller class hardware. Additionally, STEP-SAFE incorporates a dynamic risk aware alert mechanism that converts classification confidence into real-time safety decisions. The model is evaluated across multiple few-shot configurations (1-shot to 10-shot, 3-way to 5-way) and benchmarked against state-of-the-art methods. Moreover, results show that STEP-SAFE achieves up to 85.2% accuracy and 0.85mAP while requiring significantly fewer computational resources and support samples. Confusion matrix and episodic performance analyses confirm its robustness in distinguishing visually similar and occlusion prone VRU behaviors. By combining compact model design, few-shot adaptability, and interpretable alert logic, STEP-SAFE represents a deployable and intelligent solution for future ready ITS safety infrastructure.
- 2 January 2026
- Published by Springer Nature
- p. 107-130
Quantum Intelligence: Accelerating Big Data Analysis with Quantum Computing
- 2 January 2026
- Published by Springer Nature
- p. 107-130
Quantum Intelligence: Accelerating Big Data Analysis with Quantum Computing
- 22 December 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Sensors Journal
- Vol. 26 (3) , 4664-4683
Enhancing Driver Safety With ResNet-RG—A Deep Learning-Based Distraction Detection Approach
Distracted driving, caused by behaviors such as texting, eating, or adjusting in-car systems, presents significant challenges in detection, owing to subtle distractions, varying environmental conditions, and multi-tasking behaviors. Existing driver distraction detection methods often exhibit limited feature extraction capabilities, high computational requirements, and insufficient generalizability to real-world scenarios. This study developed ResNet-RG (Regularization and Generalization), a deep-learning-based architecture optimized for real-time driver distraction detection. ResNet-RG integrates advanced residual learning and an optimized attention mechanism, allowing it to dynamically prioritize critical distraction cues, such as gaze shifts, head movements, and facial expressions, while maintaining computational efficiency. The model effectively emulates multi sensor fusion using vision based learning, capturing motion patterns, lighting variations, and gaze dynamics, similar to the operation of infrared tracking and Inertial Measurement Unit (IMU) based motion sensors. Evaluations on multiple benchmark datasets demonstrated ResNet-RG’s robustness, achieving 99.91% accuracy on the State Farm Distracted Driver Detection (SFDD) dataset, 99.12% on the MRL-Eye dataset, and 98.73% on the Drive&Act dataset. Additionally, on a custom dataset, MyDataset, designed to incorporate diverse environmental conditions, the model achieved 96.71% accuracy, surpassing traditional neural network models. ResNet-RG is proposed herein as a lightweight vision only architecture that unifies residual learning, attention mechanisms, and grayscale adaptation to deliver multimodal level robustness with real-time efficiency, establishing its novelty for Advanced Driver Distraction Detection (ADDD) systems.
- 22 December 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Sensors Journal
- Vol. 26 (3) , 4664-4683
Enhancing Driver Safety With ResNet-RG—A Deep Learning-Based Distraction Detection Approach
Distracted driving, caused by behaviors such as texting, eating, or adjusting in-car systems, presents significant challenges in detection, owing to subtle distractions, varying environmental conditions, and multi-tasking behaviors. Existing driver distraction detection methods often exhibit limited feature extraction capabilities, high computational requirements, and insufficient generalizability to real-world scenarios. This study developed ResNet-RG (Regularization and Generalization), a deep-learning-based architecture optimized for real-time driver distraction detection. ResNet-RG integrates advanced residual learning and an optimized attention mechanism, allowing it to dynamically prioritize critical distraction cues, such as gaze shifts, head movements, and facial expressions, while maintaining computational efficiency. The model effectively emulates multi sensor fusion using vision based learning, capturing motion patterns, lighting variations, and gaze dynamics, similar to the operation of infrared tracking and Inertial Measurement Unit (IMU) based motion sensors. Evaluations on multiple benchmark datasets demonstrated ResNet-RG’s robustness, achieving 99.91% accuracy on the State Farm Distracted Driver Detection (SFDD) dataset, 99.12% on the MRL-Eye dataset, and 98.73% on the Drive&Act dataset. Additionally, on a custom dataset, MyDataset, designed to incorporate diverse environmental conditions, the model achieved 96.71% accuracy, surpassing traditional neural network models. ResNet-RG is proposed herein as a lightweight vision only architecture that unifies residual learning, attention mechanisms, and grayscale adaptation to deliver multimodal level robustness with real-time efficiency, establishing its novelty for Advanced Driver Distraction Detection (ADDD) systems.
- 20 November 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Internet of Things Journal
- Vol. 13 (9) , 18143-18153
Intent-Based Networking With QoS-Aware Routing in SDN Using AHP-Based AI Prioritization and Optimization
The increasing heterogeneity of network services in next-generation environments requires intelligent and adaptive routing frameworks that can align with diverse and dynamic application-level service requirements. Software-defined networking (SDN) decouples the control plane from the data plane to shift the complexity from core network devices. Moreover, intent-based networking (IBN), when integrated with SDN, provides a centralized and programmable platform to translate high-level user intents into actionable network configurations. This article proposes a hybrid quality of service (QoS)-aware routing framework that combines the analytic hierarchy process (AHP) with adaptive artificial intelligence (AI) to support intent-driven decision-making in SDN. The AHP module interprets user-defined service intents into weighted QoS priorities, while the AI component dynamically refines these weights based on the network telemetry. This integration enables context-aware path selection that balances structure, adaptability, and explainability. Experimental results on real-world network topologies demonstrate that the proposed framework consistently outperforms benchmark and state-of-the-art approaches in terms of end-to-end (E2E) delay, packet loss, jitter, throughput, and intent satisfaction. Moreover, the system achieves fast re-routing convergence and maintains a low controller overhead, making it suitable for scalable, transparent, and high-performance deployment in IBN-compliant SDN infrastructures.
- 20 November 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Internet of Things Journal
- Vol. 13 (9) , 18143-18153
Intent-Based Networking With QoS-Aware Routing in SDN Using AHP-Based AI Prioritization and Optimization
The increasing heterogeneity of network services in next-generation environments requires intelligent and adaptive routing frameworks that can align with diverse and dynamic application-level service requirements. Software-defined networking (SDN) decouples the control plane from the data plane to shift the complexity from core network devices. Moreover, intent-based networking (IBN), when integrated with SDN, provides a centralized and programmable platform to translate high-level user intents into actionable network configurations. This article proposes a hybrid quality of service (QoS)-aware routing framework that combines the analytic hierarchy process (AHP) with adaptive artificial intelligence (AI) to support intent-driven decision-making in SDN. The AHP module interprets user-defined service intents into weighted QoS priorities, while the AI component dynamically refines these weights based on the network telemetry. This integration enables context-aware path selection that balances structure, adaptability, and explainability. Experimental results on real-world network topologies demonstrate that the proposed framework consistently outperforms benchmark and state-of-the-art approaches in terms of end-to-end (E2E) delay, packet loss, jitter, throughput, and intent satisfaction. Moreover, the system achieves fast re-routing convergence and maintains a low controller overhead, making it suitable for scalable, transparent, and high-performance deployment in IBN-compliant SDN infrastructures.
- 30 August 2025
- Published by Springer Nature in Cluster Computing
- Vol. 28 (9) , 1-22
Optimization of recurrent neural networks for high-performance intrusion detection in network traffic
- 30 August 2025
- Published by Springer Nature in Cluster Computing
- Vol. 28 (9) , 1-22
Optimization of recurrent neural networks for high-performance intrusion detection in network traffic
- 15 March 2025
- Published by Springer Nature in The Journal of Supercomputing
- Vol. 81 (4) , 1-39
Deep learning techniques for enhanced security and privacy in 6G terrestrial–nonterrestrial network architecture
- 15 March 2025
- Published by Springer Nature in The Journal of Supercomputing
- Vol. 81 (4) , 1-39
Deep learning techniques for enhanced security and privacy in 6G terrestrial–nonterrestrial network architecture