Emmanuel Thepie Fapi
Emmanuel Thepie Fapi has a total of 23 co-authors. Here are the top co-authors by number of shared publications:
- Zhongwen Zhu
Ericsson GAIA
5 shared publications - Jun Yan
Concordia University
3 shared publications - Aditya Mahajan
Department of Electrical and Computer Engineering, McGill University, Montreal, Canada
2 shared publications - Ben Liang
University of Toronto
2 shared publications - Georges Kaddoum
Université du Québec
2 shared publications - Hardik Dalal
Ericsson Global AI Accelerator, Montral, QC, Canada
2 shared publications
Latest Publications
- 1 July 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Mobile Computing
- Vol. 25 (7) , 10876-10892
O-RAN Xapps Conflict Prediction Using Graph Convolutional Networks
- 1 July 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Mobile Computing
- Vol. 25 (7) , 10876-10892
O-RAN Xapps Conflict Prediction Using Graph Convolutional Networks
- 9 January 2025
- Published by Springer Nature
- p. 182-195
Development Pitfalls: A Case Study in Developing a Smart Grid Co-simulation Platform Based on HELICS
- 9 January 2025
- Published by Springer Nature
- p. 182-195
Development Pitfalls: A Case Study in Developing a Smart Grid Co-simulation Platform Based on HELICS
- 11 December 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE/ACM Transactions on Networking
- Vol. 33 (2) , 761-776
Communication-Efficient Network Topology in Decentralized Learning: A Joint Design of Consensus Matrix and Resource Allocation
In decentralized machine learning over a network of workers, each worker updates its local model as a weighted average of its local model and all models received from its neighbors. Efficient consensus weight matrix design and communication resource allocation can increase the training convergence rate and reduce the wall-clock training time. In this paper, we jointly consider these two factors and propose a novel algorithm termed Communication-Efficient Network Topology (CENT), which reduces the latency in each training iteration by removing unnecessary communication links. CENT enforces communication graph sparsity by iteratively updating, with a fixed step size, a trade-off factor between the convergence factor and a weighted graph sparsity. We further extend CENT to one with an adaptive step size (CENT-A), which adjusts the trade-off factor based on the feedback of the objective function value, without introducing additional computation complexity. We show that both CENT and CENT-A preserve the training convergence rate while avoiding the selection of poor communication links. Numerical studies with real-world machine learning data in both homogeneous and heterogeneous scenarios demonstrate the efficacy of CENT and CENT-A and their performance advantage over state-of-the-art algorithms.
- 11 December 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE/ACM Transactions on Networking
- Vol. 33 (2) , 761-776
Communication-Efficient Network Topology in Decentralized Learning: A Joint Design of Consensus Matrix and Resource Allocation
In decentralized machine learning over a network of workers, each worker updates its local model as a weighted average of its local model and all models received from its neighbors. Efficient consensus weight matrix design and communication resource allocation can increase the training convergence rate and reduce the wall-clock training time. In this paper, we jointly consider these two factors and propose a novel algorithm termed Communication-Efficient Network Topology (CENT), which reduces the latency in each training iteration by removing unnecessary communication links. CENT enforces communication graph sparsity by iteratively updating, with a fixed step size, a trade-off factor between the convergence factor and a weighted graph sparsity. We further extend CENT to one with an adaptive step size (CENT-A), which adjusts the trade-off factor based on the feedback of the objective function value, without introducing additional computation complexity. We show that both CENT and CENT-A preserve the training convergence rate while avoiding the selection of poor communication links. Numerical studies with real-world machine learning data in both homogeneous and heterogeneous scenarios demonstrate the efficacy of CENT and CENT-A and their performance advantage over state-of-the-art algorithms.
- 1 April 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Industrial Informatics
- Vol. 20 (6) , 8873-8883
Adversarial Artificial Intelligence in Blind False Data Injection in Smart Grid AC State Estimation
Artificial intelligence (AI) plays an imperative role in next-generation critical infrastructures like the smart grid, whose power can be harnessed by not only operators, but also cyber adversaries. This article investigates a potential threat from adversarial AI in blind false data injection attacks (FDIA) targeting the ac state estimators in the smart grid. Assuming no access to the grid topology required in most FDIA, we propose an adversarial model based on artificial neural networks (ANNs) to infer grid topology from historical measurements. Following the topology inference, a substitute bad data detector (BDD) model is further proposed in the attack model to filter the false data before injection, reducing the risk of detection given potential bad data in normal operations. We also refine the common evaluation of FDI stealthiness by including the presence of bad data among normal and false data when assessing the detection performance. Simulations on the IEEE 30-bus system reveal that significant deviations can be inflicted stealthily by the proposed blind FDI attack. Detailed analyses of the stealthiness, impacts, and parameters are also presented to shed more light on the threats for further studies and effective countermeasures.
- 1 April 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Industrial Informatics
- Vol. 20 (6) , 8873-8883
Adversarial Artificial Intelligence in Blind False Data Injection in Smart Grid AC State Estimation
Artificial intelligence (AI) plays an imperative role in next-generation critical infrastructures like the smart grid, whose power can be harnessed by not only operators, but also cyber adversaries. This article investigates a potential threat from adversarial AI in blind false data injection attacks (FDIA) targeting the ac state estimators in the smart grid. Assuming no access to the grid topology required in most FDIA, we propose an adversarial model based on artificial neural networks (ANNs) to infer grid topology from historical measurements. Following the topology inference, a substitute bad data detector (BDD) model is further proposed in the attack model to filter the false data before injection, reducing the risk of detection given potential bad data in normal operations. We also refine the common evaluation of FDI stealthiness by including the presence of bad data among normal and false data when assessing the detection performance. Simulations on the IEEE 30-bus system reveal that significant deviations can be inflicted stealthily by the proposed blind FDI attack. Detailed analyses of the stealthiness, impacts, and parameters are also presented to shed more light on the threats for further studies and effective countermeasures.
- 24 July 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Vehicular Technology
- Vol. 72 (12) , 16249-16265
Multi-Cell Over-the-Air Computation Systems With Spectrum Sharing: A Perspective From $\alpha$-Fairness
Wireless data aggregation (WDA) is a pivotal enabling technique in the era of Internet of Things (IoT). Recently, an emerging WDA method, over-the-air computation (AirComp), has been proposed to perform fast data aggregation, with improved spectrum utilization and shortened transmission delay. In this paper, we consider a multi-cell AirComp system with single-input multiple-output (SIMO) communications. All users in the system share the same wireless spectrum. The mean squared error (MSE) is used as the metric to quantify the computation accuracy of each cell. To achieve the fairness of data aggregation among different cells, we formulate a unified optimization problem to minimize the MSE-based objective function, from the perspective of $\alpha$ -fairness. Due to the non-convexity issue, the optimization problem is first divided into four cases according to the choice of $\alpha$ . Thereafter, an efficient solution based on the block coordinate descent (BCD) method is proposed for each case. Specifically, in the first case, the optimal solutions are obtained for the beamforming at the AP and the power allocation at the user. In the second and third cases, a tractable solution to determine the power allocation is devised by adopting the successive convex approximation (SCA) method. In the last case, the majorization minimization (MM) method is applied to tackle the power allocation problem. Furthermore, by analyzing the MSE performance, we arrive at an in-depth insight. Namely, the impact of the inter-cell interference disappears when the number of antennas at the AP is sufficiently large. The effectiveness of our proposed schemes is validated by the numerical results.
- 24 July 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Vehicular Technology
- Vol. 72 (12) , 16249-16265
Multi-Cell Over-the-Air Computation Systems With Spectrum Sharing: A Perspective From $\alpha$-Fairness
Wireless data aggregation (WDA) is a pivotal enabling technique in the era of Internet of Things (IoT). Recently, an emerging WDA method, over-the-air computation (AirComp), has been proposed to perform fast data aggregation, with improved spectrum utilization and shortened transmission delay. In this paper, we consider a multi-cell AirComp system with single-input multiple-output (SIMO) communications. All users in the system share the same wireless spectrum. The mean squared error (MSE) is used as the metric to quantify the computation accuracy of each cell. To achieve the fairness of data aggregation among different cells, we formulate a unified optimization problem to minimize the MSE-based objective function, from the perspective of $\alpha$ -fairness. Due to the non-convexity issue, the optimization problem is first divided into four cases according to the choice of $\alpha$ . Thereafter, an efficient solution based on the block coordinate descent (BCD) method is proposed for each case. Specifically, in the first case, the optimal solutions are obtained for the beamforming at the AP and the power allocation at the user. In the second and third cases, a tractable solution to determine the power allocation is devised by adopting the successive convex approximation (SCA) method. In the last case, the majorization minimization (MM) method is applied to tackle the power allocation problem. Furthermore, by analyzing the MSE performance, we arrive at an in-depth insight. Namely, the impact of the inter-cell interference disappears when the number of antennas at the AP is sufficiently large. The effectiveness of our proposed schemes is validated by the numerical results.
- 21 March 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Network and Service Management
- Vol. 20 (4) , 4951-4963
Unsupervised GAN-Based Intrusion Detection System Using Temporal Convolutional Networks and Self-Attention
Fifth-generation (5G) networks provide connectivity to a massive number of devices and boost a plethora of applications in several different domains. However, the large adoption of connected devices increases attack surfaces and introduces several security threats that can severely damage physical objects and risk people’s lives. Despite existing intrusion detection systems (IDSs), there are still several challenges to be addressed in the detection of cyber-attacks. For instance, while unsupervised IDSs are required to detect zero-day attacks, they usually present high false positive rates. Moreover, most existing IDSs rely on long short-term memory (LSTM) networks to consider time-dependencies among data. However, LSTM networks have recently been shown to present several drawbacks and limitations, which put into question their performance on sequence modeling tasks. Thus, in this paper, we investigate generative adversarial networks (GANs), a promising unsupervised approach to detecting attacks by implicitly modeling systems, and alternatives to LSTM networks to consider temporal dependencies among data. We propose a novel unsupervised GAN-based IDS that uses temporal convolutional networks (TCNs) and self-attention to detect cyber-attacks. The proposed IDS leverages edge computing and is proposed for edge servers, which bring computation resources closer to end nodes. Experiment results show that our proposed IDS can be configured to satisfy different detection rate and detection time requirements. Moreover, they show that our IDS is more accurate and at least 3.8 times faster than two state-of-the-art GAN-based IDSs that are used as baselines.
- 21 March 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Network and Service Management
- Vol. 20 (4) , 4951-4963
Unsupervised GAN-Based Intrusion Detection System Using Temporal Convolutional Networks and Self-Attention
Fifth-generation (5G) networks provide connectivity to a massive number of devices and boost a plethora of applications in several different domains. However, the large adoption of connected devices increases attack surfaces and introduces several security threats that can severely damage physical objects and risk people’s lives. Despite existing intrusion detection systems (IDSs), there are still several challenges to be addressed in the detection of cyber-attacks. For instance, while unsupervised IDSs are required to detect zero-day attacks, they usually present high false positive rates. Moreover, most existing IDSs rely on long short-term memory (LSTM) networks to consider time-dependencies among data. However, LSTM networks have recently been shown to present several drawbacks and limitations, which put into question their performance on sequence modeling tasks. Thus, in this paper, we investigate generative adversarial networks (GANs), a promising unsupervised approach to detecting attacks by implicitly modeling systems, and alternatives to LSTM networks to consider temporal dependencies among data. We propose a novel unsupervised GAN-based IDS that uses temporal convolutional networks (TCNs) and self-attention to detect cyber-attacks. The proposed IDS leverages edge computing and is proposed for edge servers, which bring computation resources closer to end nodes. Experiment results show that our proposed IDS can be configured to satisfy different detection rate and detection time requirements. Moreover, they show that our IDS is more accurate and at least 3.8 times faster than two state-of-the-art GAN-based IDSs that are used as baselines.
- 2 August 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Cognitive Communications and Networking
- Vol. 8 (4) , 1881-1897
Structure-Aware Reinforcement Learning for Node-Overload Protection in Mobile Edge Computing
Mobile Edge Computing (MEC) involves placing computational capability and applications at the edge of the network, providing benefits such as reduced latency, reduced network congestion, and improved performance of applications. The performance and reliability of MEC degrades significantly when the edge server(s) in the cluster are overloaded. In this work, an adaptive admission control policy to prevent edge node from getting overloaded is presented. This approach is based on a recently-proposed low complexity RL (Reinforcement Learning) algorithm called SALMUT (Structure-Aware Learning for Multiple Thresholds), which exploits the structure of the optimal admission control policy in multi-class queues for an average-cost setting. We extend the framework to work for node overload-protection problem in a discounted-cost setting. The proposed solution is validated using several scenarios mimicking real-world deployments in two different settings — computer simulations and a docker testbed. Our empirical evaluations show that the total discounted cost incurred by SALMUT is similar to state-of-the-art deep RL algorithms such as PPO (Proximal Policy Optimization) and A2C (Advantage Actor Critic) but requires an order of magnitude less time to train, outputs easily interpretable policy, and can be deployed in an online manner.
- 2 August 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Cognitive Communications and Networking
- Vol. 8 (4) , 1881-1897
Structure-Aware Reinforcement Learning for Node-Overload Protection in Mobile Edge Computing
Mobile Edge Computing (MEC) involves placing computational capability and applications at the edge of the network, providing benefits such as reduced latency, reduced network congestion, and improved performance of applications. The performance and reliability of MEC degrades significantly when the edge server(s) in the cluster are overloaded. In this work, an adaptive admission control policy to prevent edge node from getting overloaded is presented. This approach is based on a recently-proposed low complexity RL (Reinforcement Learning) algorithm called SALMUT (Structure-Aware Learning for Multiple Thresholds), which exploits the structure of the optimal admission control policy in multi-class queues for an average-cost setting. We extend the framework to work for node overload-protection problem in a discounted-cost setting. The proposed solution is validated using several scenarios mimicking real-world deployments in two different settings — computer simulations and a docker testbed. Our empirical evaluations show that the total discounted cost incurred by SALMUT is similar to state-of-the-art deep RL algorithms such as PPO (Proximal Policy Optimization) and A2C (Advantage Actor Critic) but requires an order of magnitude less time to train, outputs easily interpretable policy, and can be deployed in an online manner.
- 13 June 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-9
Joint Consensus Matrix Design and Resource Allocation for Decentralized Learning
In decentralized machine learning over a network of workers, each worker updates its local model as a weighted average of its local model and all models received from its neighbors. Efficient consensus weight matrix design and communication resource allocation can increase the training convergence rate and reduce the wall-clock training time. In this paper, we jointly consider these two factors and propose a novel algorithm termed Communication-Efficient Network Topology (CENT), which reduces the latency in each training iteration by removing unnecessary communication links. CENT preserves the training convergence rate while enforcing communication graph sparsity and avoiding selecting poor communication links. Numerical study with real-world machine learning data demonstrates the efficacy of the proposed solution and its performance advantage over state-of-the-art algorithms.
- 13 June 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-9
Joint Consensus Matrix Design and Resource Allocation for Decentralized Learning
In decentralized machine learning over a network of workers, each worker updates its local model as a weighted average of its local model and all models received from its neighbors. Efficient consensus weight matrix design and communication resource allocation can increase the training convergence rate and reduce the wall-clock training time. In this paper, we jointly consider these two factors and propose a novel algorithm termed Communication-Efficient Network Topology (CENT), which reduces the latency in each training iteration by removing unnecessary communication links. CENT preserves the training convergence rate while enforcing communication graph sparsity and avoiding selecting poor communication links. Numerical study with real-world machine learning data demonstrates the efficacy of the proposed solution and its performance advantage over state-of-the-art algorithms.
- 13 April 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Communications Letters
- Vol. 26 (7) , 1583-1587
Multi-Objective GAN-Based Adversarial Attack Technique for Modulation Classifiers
Deep learning is increasingly being used for many tasks in wireless communications, such as modulation classification. However, it has been shown to be vulnerable to adversarial attacks, which introduce specially crafted imperceptible perturbations, inducing models to make mistakes. This letter proposes an input-agnostic adversarial attack technique that is based on generative adversarial networks (GANs) and multi-task loss. Our results show that our technique reduces the accuracy of a modulation classifier more than a jamming attack and other adversarial attack techniques. Furthermore, it generates adversarial samples at least 335 times faster than the other techniques evaluated, which raises serious concerns about using deep learning-based modulation classifiers.
- 13 April 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Communications Letters
- Vol. 26 (7) , 1583-1587
Multi-Objective GAN-Based Adversarial Attack Technique for Modulation Classifiers
Deep learning is increasingly being used for many tasks in wireless communications, such as modulation classification. However, it has been shown to be vulnerable to adversarial attacks, which introduce specially crafted imperceptible perturbations, inducing models to make mistakes. This letter proposes an input-agnostic adversarial attack technique that is based on generative adversarial networks (GANs) and multi-task loss. Our results show that our technique reduces the accuracy of a modulation classifier more than a jamming attack and other adversarial attack techniques. Furthermore, it generates adversarial samples at least 335 times faster than the other techniques evaluated, which raises serious concerns about using deep learning-based modulation classifiers.
- Open Access
Structure-aware reinforcement learning for node-overload protection in mobile edge computing
Mobile Edge Computing (MEC) refers to the concept of placing computational capability and applications at the edge of the network, providing benefits such as reduced latency in handling client requests, reduced network congestion, and improved performance of applications. The performance and reliability of MEC are degraded significantly when one or several edge servers in the cluster are overloaded. Especially when a server crashes due to the overload, it causes service failures in MEC. In this work, an adaptive admission control policy to prevent edge node from getting overloaded is presented. This approach is based on a recently-proposed low complexity RL (Reinforcement Learning) algorithm called SALMUT (Structure-Aware Learning for Multiple Thresholds), which exploits the structure of the optimal admission control policy in multi-class queues for an average-cost setting. We extend the framework to work for node overload-protection problem in a discounted-cost setting. The proposed solution is validated using several scenarios mimicking real-world deployments in two different settings - computer simulations and a docker testbed. Our empirical evaluations show that the total discounted cost incurred by SALMUT is similar to state-of-the-art deep RL algorithms such as PPO (Proximal Policy Optimization) and A2C (Advantage Actor Critic) but requires an order of magnitude less time to train, outputs easily interpretable policy, and can be deployed in an online manner.
Open AccessStructure-aware reinforcement learning for node-overload protection in mobile edge computing
Mobile Edge Computing (MEC) refers to the concept of placing computational capability and applications at the edge of the network, providing benefits such as reduced latency in handling client requests, reduced network congestion, and improved performance of applications. The performance and reliability of MEC are degraded significantly when one or several edge servers in the cluster are overloaded. Especially when a server crashes due to the overload, it causes service failures in MEC. In this work, an adaptive admission control policy to prevent edge node from getting overloaded is presented. This approach is based on a recently-proposed low complexity RL (Reinforcement Learning) algorithm called SALMUT (Structure-Aware Learning for Multiple Thresholds), which exploits the structure of the optimal admission control policy in multi-class queues for an average-cost setting. We extend the framework to work for node overload-protection problem in a discounted-cost setting. The proposed solution is validated using several scenarios mimicking real-world deployments in two different settings - computer simulations and a docker testbed. Our empirical evaluations show that the total discounted cost incurred by SALMUT is similar to state-of-the-art deep RL algorithms such as PPO (Proximal Policy Optimization) and A2C (Advantage Actor Critic) but requires an order of magnitude less time to train, outputs easily interpretable policy, and can be deployed in an online manner.