Ahmed Refaey Hussein
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
Ahmed Refaey Hussein has a total of 52 co-authors. Here are the top co-authors by number of shared publications:
- Yahuza Bello
University of Guelph, Guelph, Ontario, Canada
6 shared publications - Abdallah Shami
Western University
4 shared publications Jaspreet Singh
Manhattan College, New York, NY, USA
3 shared publications- Ahmed Sallam
Western University, London, Canada
2 shared publications - Aiman Erbad
Qatar University
2 shared publications - Amr Mohamed
Qatar University
2 shared publications
Latest Publications
- 1 September 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Vehicular Technology Magazine
- Vol. 21 (3) , 88-97
Causally Explainable and Responsible AI: Safety-Critical Intelligent Transportation Systems
- 1 September 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Vehicular Technology Magazine
- Vol. 21 (3) , 88-97
Causally Explainable and Responsible AI: Safety-Critical Intelligent Transportation Systems
- Open Access
- 29 December 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in TechRxiv
Foundation Models and Large Language Models in Next Generation Networks: Comparison, Opportunities and Challenges
Next-generation networks (NGNs) face growing demands for more bandwidth, lower delay, and better intelligence. Foundation Models (FMs) and Large Language Models (LLMs) are powerful generative artificial intelligence models that could help meet these demands. However, there is confusion about their differences and how best to use them in NGNs. This research survey presents a detailed evaluation of FMs against LLMs to illustrate the distinct characteristics and benefits of both models. The work also examines NGN applications, which include network traffic analysis enhancement and network management automation, as well as security improvement through the use of these models. The research paper discusses vital enabling techniques that are needed to train and customize the current models. Research results show that NGNs experience improved intelligent operations through the use of FMs together with LLMs. The process of implementing FMs and LLMs, along with their training, encounters obstacles that include demanding data as well as high computational expenses and security issues. The study emphasizes that future NGNs require FMs and LLMs to succeed; yet, further studies are necessary to overcome existing obstacles and reach their complete capabilities.
- 29 December 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in TechRxiv
Open AccessFoundation Models and Large Language Models in Next Generation Networks: Comparison, Opportunities and Challenges
Next-generation networks (NGNs) face growing demands for more bandwidth, lower delay, and better intelligence. Foundation Models (FMs) and Large Language Models (LLMs) are powerful generative artificial intelligence models that could help meet these demands. However, there is confusion about their differences and how best to use them in NGNs. This research survey presents a detailed evaluation of FMs against LLMs to illustrate the distinct characteristics and benefits of both models. The work also examines NGN applications, which include network traffic analysis enhancement and network management automation, as well as security improvement through the use of these models. The research paper discusses vital enabling techniques that are needed to train and customize the current models. Research results show that NGNs experience improved intelligent operations through the use of FMs together with LLMs. The process of implementing FMs and LLMs, along with their training, encounters obstacles that include demanding data as well as high computational expenses and security issues. The study emphasizes that future NGNs require FMs and LLMs to succeed; yet, further studies are necessary to overcome existing obstacles and reach their complete capabilities.
- 11 August 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Intelligent Transportation Systems
- Vol. 26 (10) , 18180-18193
Distributed Traffic Control in Complex Dynamic Roadblocks: A Multi-Agent Deep Reinforcement Learning Approach
Autonomous Vehicles (AVs) represent a transformative advancement in the transportation industry. These vehicles have sophisticated sensors, advanced algorithms, and powerful computing systems that allow them to navigate and operate without direct human intervention. However, AVs’ systems still get overwhelmed when they encounter a complex dynamic change in the environment resulting from an accident or a roadblock for maintenance. The advanced features of Sixth Generation (6G) technology are set to offer strong support to AVs, enabling real-time data exchange and management of complex driving maneuvers. This paper proposes a Multi-Agent Reinforcement Learning (MARL) framework to improve AVs’ decision-making in dynamic and complex Intelligent Transportation Systems (ITS) utilizing 6G-V2X communication. The primary objective is to enable AVs to avoid roadblocks efficiently by changing lanes while maintaining optimal traffic flow and maximizing the mean harmonic speed. To ensure realistic operations, key constraints such as minimum vehicle speed, roadblock count, and lane change frequency are integrated. We train and test the proposed MARL model with two traffic simulation scenarios using the SUMO and TraCI interface. Through extensive simulations, we demonstrate that the proposed model adapts to various traffic conditions and achieves efficient and robust traffic flow management. Specifically, the proposed approach results in a harmonic mean speed increase of up to 15% and a reduction in lane-change frequency by 10%. The trained model effectively navigates dynamic roadblocks, promoting improved traffic efficiency in AV operations with more than 70% efficiency over other benchmark solutions.
- 11 August 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Intelligent Transportation Systems
- Vol. 26 (10) , 18180-18193
Distributed Traffic Control in Complex Dynamic Roadblocks: A Multi-Agent Deep Reinforcement Learning Approach
Autonomous Vehicles (AVs) represent a transformative advancement in the transportation industry. These vehicles have sophisticated sensors, advanced algorithms, and powerful computing systems that allow them to navigate and operate without direct human intervention. However, AVs’ systems still get overwhelmed when they encounter a complex dynamic change in the environment resulting from an accident or a roadblock for maintenance. The advanced features of Sixth Generation (6G) technology are set to offer strong support to AVs, enabling real-time data exchange and management of complex driving maneuvers. This paper proposes a Multi-Agent Reinforcement Learning (MARL) framework to improve AVs’ decision-making in dynamic and complex Intelligent Transportation Systems (ITS) utilizing 6G-V2X communication. The primary objective is to enable AVs to avoid roadblocks efficiently by changing lanes while maintaining optimal traffic flow and maximizing the mean harmonic speed. To ensure realistic operations, key constraints such as minimum vehicle speed, roadblock count, and lane change frequency are integrated. We train and test the proposed MARL model with two traffic simulation scenarios using the SUMO and TraCI interface. Through extensive simulations, we demonstrate that the proposed model adapts to various traffic conditions and achieves efficient and robust traffic flow management. Specifically, the proposed approach results in a harmonic mean speed increase of up to 15% and a reduction in lane-change frequency by 10%. The trained model effectively navigates dynamic roadblocks, promoting improved traffic efficiency in AV operations with more than 70% efficiency over other benchmark solutions.
- 1 December 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Communications Magazine
- Vol. 62 (12) , 64-65
Series Editorial: Internet of Things
- 1 December 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Communications Magazine
- Vol. 62 (12) , 64-65
Series Editorial: Internet of Things
- 16 October 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Network and Service Management
- Vol. 22 (1) , 807-821
Dynamic Policy Decision/Enforcement Security Zoning Through Stochastic Games and Meta Learning
Securing Next Generation Networks (NGNs) remains a prominent topic of discussion in academia and industries alike, driven by the rapid evolution of cyber attacks. As these attacks become increasingly complex and dynamic, it is crucial to develop sophisticated security strategies with automated dynamic policy enforcement. In this paper, we propose a security strategy based on the zero-trust model, incorporating dynamic policy decisions through the utilization of stochastic games and Reinforcement Learning (RL). Our approach involves the development of an attack and defense strategy evolution model, specifically tailored to combat cyber attacks in NGNs. To achieve this, we employ RL techniques to update and adapt dynamic policies. To train the agents, we utilize the Generalized Proximal Policy Optimization with sample reuse (GePPO) algorithm, including its modified version, GePPO-ML, which incorporates meta-learning to initialize the agent’s policy and parameters. Additionally, we employ the Sample Dropout PPO with meta-learning (SDPPO-ML), a modified version of the SD-PPO algorithm, to train the agents. To evaluate the performance of these algorithms, we conduct a comparative analysis against the REINFORCE and PPO algorithms. The results illustrate the superior performance of both GePPO-ML and SDPPO-ML when compared to these baseline algorithms, with GePPO-ML exhibiting the best performance.
- 16 October 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Network and Service Management
- Vol. 22 (1) , 807-821
Dynamic Policy Decision/Enforcement Security Zoning Through Stochastic Games and Meta Learning
Securing Next Generation Networks (NGNs) remains a prominent topic of discussion in academia and industries alike, driven by the rapid evolution of cyber attacks. As these attacks become increasingly complex and dynamic, it is crucial to develop sophisticated security strategies with automated dynamic policy enforcement. In this paper, we propose a security strategy based on the zero-trust model, incorporating dynamic policy decisions through the utilization of stochastic games and Reinforcement Learning (RL). Our approach involves the development of an attack and defense strategy evolution model, specifically tailored to combat cyber attacks in NGNs. To achieve this, we employ RL techniques to update and adapt dynamic policies. To train the agents, we utilize the Generalized Proximal Policy Optimization with sample reuse (GePPO) algorithm, including its modified version, GePPO-ML, which incorporates meta-learning to initialize the agent’s policy and parameters. Additionally, we employ the Sample Dropout PPO with meta-learning (SDPPO-ML), a modified version of the SD-PPO algorithm, to train the agents. To evaluate the performance of these algorithms, we conduct a comparative analysis against the REINFORCE and PPO algorithms. The results illustrate the superior performance of both GePPO-ML and SDPPO-ML when compared to these baseline algorithms, with GePPO-ML exhibiting the best performance.
- 1 July 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Communications Magazine
- Vol. 62 (7) , 42-43
Series Editorial: Internet of Things
- 1 July 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Communications Magazine
- Vol. 62 (7) , 42-43
Series Editorial: Internet of Things
- 10 April 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Wireless Communications
- Vol. 31 (2) , 74-80
Toward Zero-Trust 6GC: A Software Defined Perimeter Approach with Dynamic Moving Target Defense Mechanism
The upcoming Sixth Generation (6G) network is projected to grapple with a range of security concerns, encompassing access control, authentication, secure connections among 6G Core (6GC) entities, and trustworthiness. Classical Virtual Private Networks (VPNs), extensively deployed in Evolved Packet Core (EPC) network infrastructure, are notoriously susceptible to a variety of attacks, including man-in-the-middle incursions, Domain Name System (DNS) hijacking, Denial of Service (DoS) attacks, port scanning, and persistent unauthorized access attempts. This paper introduces the concept of Software Defined Perimeter (SDP) as an innovative solution, providing an alternative to VPNs with the goal of fostering a secure zero-trust milieu within the 6G Core networks. We capitalize on the SDP controller-based authentication and authorization mechanisms to secure the EPC network's control and data plane functions, conceiving an architecture that is expansible to the 6G network. Further, we augment the SDP zero-trust capabilities via the incorporation of a dynamic component, the Moving Target Defense (MTD). This enhances the network's resilience against attacks targeting traditionally static network environments established via VPNs. Following rigorous testbed analysis, our proposed framework manifests superior resilience against DoS and port scanning attacks when jux-taposed with traditional VPN methodologies.
- 10 April 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Wireless Communications
- Vol. 31 (2) , 74-80
Toward Zero-Trust 6GC: A Software Defined Perimeter Approach with Dynamic Moving Target Defense Mechanism
The upcoming Sixth Generation (6G) network is projected to grapple with a range of security concerns, encompassing access control, authentication, secure connections among 6G Core (6GC) entities, and trustworthiness. Classical Virtual Private Networks (VPNs), extensively deployed in Evolved Packet Core (EPC) network infrastructure, are notoriously susceptible to a variety of attacks, including man-in-the-middle incursions, Domain Name System (DNS) hijacking, Denial of Service (DoS) attacks, port scanning, and persistent unauthorized access attempts. This paper introduces the concept of Software Defined Perimeter (SDP) as an innovative solution, providing an alternative to VPNs with the goal of fostering a secure zero-trust milieu within the 6G Core networks. We capitalize on the SDP controller-based authentication and authorization mechanisms to secure the EPC network's control and data plane functions, conceiving an architecture that is expansible to the 6G network. Further, we augment the SDP zero-trust capabilities via the incorporation of a dynamic component, the Moving Target Defense (MTD). This enhances the network's resilience against attacks targeting traditionally static network environments established via VPNs. Following rigorous testbed analysis, our proposed framework manifests superior resilience against DoS and port scanning attacks when jux-taposed with traditional VPN methodologies.
- 30 October 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Systems Journal
- Vol. PP (19328184) , 1-12
Learning-Based Neighboring Station Coverage Identification and Dynamic Resource Utilization for 3-D Cells in NR-U Networks
With ongoing proliferation of wireless networks and dramatic growth of data traffic, offloading cellular network traffic to unlicensed band (UB) enabled new radio (NR) base stations (BSs) becomes one of the critical alternatives to address the radio resource constraint. Due to network densification and reduced small cell coverage, this solution can be further improved through efficient and dynamic utilization of 3-D spatially distributed common UB radio resources. The overall objective of this study is to better estimate the 3-D location-specific interference from neighboring UB BSs and to support efficient and dynamic radio resource utilization in the overlapped coverage space. In the first part of this study, the problem of determining the location-specific interference of a neighbor NR-Unlicensed (NR-U) BS is addressed. This problem is solved with a deep regression neural network assisted algorithm while predicting 3-D location-based interference power of a neighbor NR-U BS and bypassing the complex path loss parameter estimation processes. In the second part of this study, the problem of efficient and dynamic UB radio resource utilization is considered for NR-U transmissions. To solve this problem, a collaborative double $Q$ -learning (DQL) algorithm is developed. Under ideal, estimated and theoretical categories of predictions, approximately 11%, 23%, and 20% better performance and approximately 59%, 212%, and 67% faster algorithm convergence are shown by the DQL assisted approach over conventional $Q$ -learning-based solutions, respectively.
- 30 October 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Systems Journal
- Vol. PP (19328184) , 1-12
Learning-Based Neighboring Station Coverage Identification and Dynamic Resource Utilization for 3-D Cells in NR-U Networks
With ongoing proliferation of wireless networks and dramatic growth of data traffic, offloading cellular network traffic to unlicensed band (UB) enabled new radio (NR) base stations (BSs) becomes one of the critical alternatives to address the radio resource constraint. Due to network densification and reduced small cell coverage, this solution can be further improved through efficient and dynamic utilization of 3-D spatially distributed common UB radio resources. The overall objective of this study is to better estimate the 3-D location-specific interference from neighboring UB BSs and to support efficient and dynamic radio resource utilization in the overlapped coverage space. In the first part of this study, the problem of determining the location-specific interference of a neighbor NR-Unlicensed (NR-U) BS is addressed. This problem is solved with a deep regression neural network assisted algorithm while predicting 3-D location-based interference power of a neighbor NR-U BS and bypassing the complex path loss parameter estimation processes. In the second part of this study, the problem of efficient and dynamic UB radio resource utilization is considered for NR-U transmissions. To solve this problem, a collaborative double $Q$ -learning (DQL) algorithm is developed. Under ideal, estimated and theoretical categories of predictions, approximately 11%, 23%, and 20% better performance and approximately 59%, 212%, and 67% faster algorithm convergence are shown by the DQL assisted approach over conventional $Q$ -learning-based solutions, respectively.