Maria Diamanti
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Maria Diamanti has a total of 49 co-authors. Here are the top co-authors by number of shared publications:
- Symeon Papavassiliou
National Technical University of Athens
22 shared publications - Eirini Eleni Tsiropoulou
Arizona State University
18 shared publications Symeon Papavassiliou
15 shared publications- Panagiotis Charatsaris
National Technical University of Athens
14 shared publications Eirini Eleni Tsiropoulou
8 shared publications- Sofia Barkatsa
National Technical University of Athens
4 shared publications
Latest Publications
- 1 November 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Consumer Electronics
- Vol. 72 (4) , 11655-11668
Proactive Cyber Deception in Attack Graphs: A Risk-Aware Deployment Framework
- 1 November 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Consumer Electronics
- Vol. 72 (4) , 11655-11668
Proactive Cyber Deception in Attack Graphs: A Risk-Aware Deployment Framework
- 1 May 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Consumer Electronics
- Vol. 72 (2) , 5117-5132
Balancing Accuracy-Latency Tradeoff in Hybrid Approximate-Delayed Computing Systems
- 1 May 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Consumer Electronics
- Vol. 72 (2) , 5117-5132
Balancing Accuracy-Latency Tradeoff in Hybrid Approximate-Delayed Computing Systems
- 23 February 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Vehicular Technology
- Vol. PP (00189545) , 1-16
Latency and Energy Minimization in RIS-enabled AirComp Federated Learning Networks
Over-the-Air Computation (AirComp) in Federated Learning (FL) has captivated research interest for fast and spectrally efficient model aggregation across distributed clients. To mitigate the adverse effects of channel fading on the aggregated global model, Reconfigurable Intelligent Surfaces (RIS) have been further integrated into the AirComp FL network architecture. In this context, the overwhelming majority of the literature focuses on active and passive beamforming design toward minimizing the superimposed signal's Mean Squared Error (MSE). In this paper, we take a step further by investigating the relationship between the MSE, overall latency, and energy consumption in RIS-enabled AirComp FL networks while explicitly incorporating the computing aspects of FL, which are often overlooked. To this end, we formulate and solve the weighted overall latency and energy consumption minimization problem due to both clients' transmissions and local computations. The optimization targets the design of the RIS phase shifts, the receive beamformer, and allocating local computing frequencies at the client devices. The original non-convex problem is decomposed into subproblems, each of which can be reformulated into a convex form and solved either directly or via an iterative alternating optimization approach. The paper concludes with the overall optimization algorithm and extensive numerical results, comparing the proposed analytical solution with general-purpose optimization toolboxes. The performance gains of RIS and the benefits of joint communication-computing resource allocation are also demonstrated.
- 23 February 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Vehicular Technology
- Vol. PP (00189545) , 1-16
Latency and Energy Minimization in RIS-enabled AirComp Federated Learning Networks
Over-the-Air Computation (AirComp) in Federated Learning (FL) has captivated research interest for fast and spectrally efficient model aggregation across distributed clients. To mitigate the adverse effects of channel fading on the aggregated global model, Reconfigurable Intelligent Surfaces (RIS) have been further integrated into the AirComp FL network architecture. In this context, the overwhelming majority of the literature focuses on active and passive beamforming design toward minimizing the superimposed signal's Mean Squared Error (MSE). In this paper, we take a step further by investigating the relationship between the MSE, overall latency, and energy consumption in RIS-enabled AirComp FL networks while explicitly incorporating the computing aspects of FL, which are often overlooked. To this end, we formulate and solve the weighted overall latency and energy consumption minimization problem due to both clients' transmissions and local computations. The optimization targets the design of the RIS phase shifts, the receive beamformer, and allocating local computing frequencies at the client devices. The original non-convex problem is decomposed into subproblems, each of which can be reformulated into a convex form and solved either directly or via an iterative alternating optimization approach. The paper concludes with the overall optimization algorithm and extensive numerical results, comparing the proposed analytical solution with general-purpose optimization toolboxes. The performance gains of RIS and the benefits of joint communication-computing resource allocation are also demonstrated.
- 1 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Networking
- Vol. 34, 7949-7964
Distributed Resource Control in Over-the-Air Computation Based on Multi-Aggregative Games
- 1 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Networking
- Vol. 34, 7949-7964
Distributed Resource Control in Over-the-Air Computation Based on Multi-Aggregative Games
- 1 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Wireless Communications
- Vol. 25, 22032-22046
Radio and Compute Resource Allocation for SWIPT and RIS-Assisted AirComp Federated Learning
- 1 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Wireless Communications
- Vol. 25, 22032-22046
Radio and Compute Resource Allocation for SWIPT and RIS-Assisted AirComp Federated Learning
- Open Access
- 1 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Open Journal of the Communications Society
- Vol. 7, 10067-10082
On-Orbit Computing and Caching: A Distributed Regret Learning Approach
- 1 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Open Journal of the Communications Society
- Vol. 7, 10067-10082
Open AccessOn-Orbit Computing and Caching: A Distributed Regret Learning Approach
- 22 December 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Network Science and Engineering
- Vol. 13, 5056-5073
Risk-Conscious Computing in Space-Air-Ground IoT Networks: A Prospect-Theoretic Game Perspective
Space-Air-Ground Integrated Networks (SAGIN) have been recognized as key enablers of 6G systems for ubiquitous service provisioning, unlocking Internet of Things (IoT) applications in geographically dispersed areas. In this paper, we study the problem of computation task offloading for remotely deployed IoT devices to either a limited-capability Uncrewed Aerial Vehicle (UAV)-mounted Multi-access Edge Computing (MEC) server or a cloud server via satellite relaying. The problem is formulated as a non-cooperative game, where each IoT device autonomously determines the percentage of task offloaded to each server to minimize the aggregate time and energy overhead due to transmissions and remote processing. Diverging from the prevailing literature, in this paper, we model the IoT devices' risk-seeking and loss-averse behavior in sharing the common pools of computing resources, i.e., cloud and MEC server. By incorporating risk-consciousness in their computation offloading decision-making, IoT devices strive to balance the total incurred overhead with the likelihood of task rejection due to overexploitation of limited shared edge resources. To this end, the IoT devices' utility function is modeled using Prospect Theory and Tragedy of the Commons. Two solutions based on normal and satisfaction-form games are derived, targeting to maximize or achieve a minimum value for the prospect-theoretic utility, providing insights from both device and system perspectives, respectively. Numerical results show the effectiveness of the overall risk-conscious computing framework in the achieved time and energy overhead, as well as task probability of failure.
- 22 December 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Network Science and Engineering
- Vol. 13, 5056-5073
Risk-Conscious Computing in Space-Air-Ground IoT Networks: A Prospect-Theoretic Game Perspective
Space-Air-Ground Integrated Networks (SAGIN) have been recognized as key enablers of 6G systems for ubiquitous service provisioning, unlocking Internet of Things (IoT) applications in geographically dispersed areas. In this paper, we study the problem of computation task offloading for remotely deployed IoT devices to either a limited-capability Uncrewed Aerial Vehicle (UAV)-mounted Multi-access Edge Computing (MEC) server or a cloud server via satellite relaying. The problem is formulated as a non-cooperative game, where each IoT device autonomously determines the percentage of task offloaded to each server to minimize the aggregate time and energy overhead due to transmissions and remote processing. Diverging from the prevailing literature, in this paper, we model the IoT devices' risk-seeking and loss-averse behavior in sharing the common pools of computing resources, i.e., cloud and MEC server. By incorporating risk-consciousness in their computation offloading decision-making, IoT devices strive to balance the total incurred overhead with the likelihood of task rejection due to overexploitation of limited shared edge resources. To this end, the IoT devices' utility function is modeled using Prospect Theory and Tragedy of the Commons. Two solutions based on normal and satisfaction-form games are derived, targeting to maximize or achieve a minimum value for the prospect-theoretic utility, providing insights from both device and system perspectives, respectively. Numerical results show the effectiveness of the overall risk-conscious computing framework in the achieved time and energy overhead, as well as task probability of failure.
- 17 December 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Network and Service Management
- Vol. 23, 1312-1328
Resilient RAN Selection and SFC Deployment in Dependable Wireless Edge Cloud Networks
The evolution toward sixth-generation (6G) networks necessitates integrated resource management solutions to address the interdependencies between network segments, such as Radio Access Network (RAN) and Edge Cloud (EC) infrastructures. Unified management of network and compute fabrics is crucial for achieving seamless service delivery, end-to-end power efficiency, and delay guarantees, while resiliency becomes a key enabler for adapting to various application demands and diverse network segment conditions. In this context, this paper proposes a unified framework for dependable wireless EC networks that jointly addresses the problems of RAN selection and Service Function Chain (SFC) embedding to minimize the total power consumption across network segments under end-to-end delay SFC deployment constraints. The framework iteratively solves these problems, considering the interdependencies between RAN ingress points and the EC network resource constraints. To deal with the high dimensionality of the considered parameters and achieve timely and scalable decision-making, a coalition formation game optimizes RAN selection, while a delay-aware heuristic approach undertakes the power-efficient embedding of multiple SFCs within the EC network. Simulation results demonstrate the framework’s efficiency in reducing power consumption compared to segment-specific approaches, highlighting the importance of cross-segment dependencies. Also, the adaptability of the proposed unified modeling and the framework’s scalability are demonstrated, ensuring resilient performance under varying network parameter settings.
- 17 December 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Network and Service Management
- Vol. 23, 1312-1328
Resilient RAN Selection and SFC Deployment in Dependable Wireless Edge Cloud Networks
The evolution toward sixth-generation (6G) networks necessitates integrated resource management solutions to address the interdependencies between network segments, such as Radio Access Network (RAN) and Edge Cloud (EC) infrastructures. Unified management of network and compute fabrics is crucial for achieving seamless service delivery, end-to-end power efficiency, and delay guarantees, while resiliency becomes a key enabler for adapting to various application demands and diverse network segment conditions. In this context, this paper proposes a unified framework for dependable wireless EC networks that jointly addresses the problems of RAN selection and Service Function Chain (SFC) embedding to minimize the total power consumption across network segments under end-to-end delay SFC deployment constraints. The framework iteratively solves these problems, considering the interdependencies between RAN ingress points and the EC network resource constraints. To deal with the high dimensionality of the considered parameters and achieve timely and scalable decision-making, a coalition formation game optimizes RAN selection, while a delay-aware heuristic approach undertakes the power-efficient embedding of multiple SFCs within the EC network. Simulation results demonstrate the framework’s efficiency in reducing power consumption compared to segment-specific approaches, highlighting the importance of cross-segment dependencies. Also, the adaptability of the proposed unified modeling and the framework’s scalability are demonstrated, ensuring resilient performance under varying network parameter settings.
- 13 October 2025
- Published by Springer Nature
- p. 187-206
Secure and Power-Efficient Multi-user Scheduling in Semi-Grant-Free NOMA Networks: A Coalition Game Approach
- 13 October 2025
- Published by Springer Nature
- p. 187-206
Secure and Power-Efficient Multi-user Scheduling in Semi-Grant-Free NOMA Networks: A Coalition Game Approach
- 3 October 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Consumer Electronics
- Vol. 71 (4) , 9768-9779
Energy-Efficient and Data-Optimized Federated Learning for Distributed On-Device Intelligence
The advent of Artificial Intelligence (AI), requiring large data from diverse devices, necessitates novel architectures that incorporate edge devices in the pipeline of AI operations. Federated Learning (FL) is becoming the state-of-practice for distributed, collaborative training of AI models at the network edge, respecting data privacy and anonymity concerns. However, implementing FL on battery-powered edge devices with limited computing capabilities necessitates innovative approaches to balance training accuracy and energy efficiency. In this work, we propose a framework for energy-efficient and data-optimized FL to facilitate on-device intelligence. The proposed framework jointly optimizes the (a) selection of the most important data samples from the devices’ local training datasets to maintain FL model accuracy, and their (b) computing frequency and (c) uplink transmission power to minimize the energy consumption during local computation and communication phases. A multi-processor model is considered for each device, extending data selection to allocate data samples to each device’s processors. The initially formulated non-convex and combinatorial optimization problem is decomposed into sub-problems, each equivalently transformed into a convex form. An Alternating Optimization (AO) approach is then applied to the sub-problems, yielding a close-to-optimal solution to the original problem. The proposed algorithm is distinguished in terms of scalability, achieving at least 30% energy reduction on the MNIST and CIFAR10 datasets compared to the second-best benchmarking schemes, while maintaining FL model accuracy at nearly the same level. Overall, the proposed framework proves scalable and suitable for deployment in realistic resource-constrained edge environments.
- 3 October 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Consumer Electronics
- Vol. 71 (4) , 9768-9779
Energy-Efficient and Data-Optimized Federated Learning for Distributed On-Device Intelligence
The advent of Artificial Intelligence (AI), requiring large data from diverse devices, necessitates novel architectures that incorporate edge devices in the pipeline of AI operations. Federated Learning (FL) is becoming the state-of-practice for distributed, collaborative training of AI models at the network edge, respecting data privacy and anonymity concerns. However, implementing FL on battery-powered edge devices with limited computing capabilities necessitates innovative approaches to balance training accuracy and energy efficiency. In this work, we propose a framework for energy-efficient and data-optimized FL to facilitate on-device intelligence. The proposed framework jointly optimizes the (a) selection of the most important data samples from the devices’ local training datasets to maintain FL model accuracy, and their (b) computing frequency and (c) uplink transmission power to minimize the energy consumption during local computation and communication phases. A multi-processor model is considered for each device, extending data selection to allocate data samples to each device’s processors. The initially formulated non-convex and combinatorial optimization problem is decomposed into sub-problems, each equivalently transformed into a convex form. An Alternating Optimization (AO) approach is then applied to the sub-problems, yielding a close-to-optimal solution to the original problem. The proposed algorithm is distinguished in terms of scalability, achieving at least 30% energy reduction on the MNIST and CIFAR10 datasets compared to the second-best benchmarking schemes, while maintaining FL model accuracy at nearly the same level. Overall, the proposed framework proves scalable and suitable for deployment in realistic resource-constrained edge environments.