Selected Publications
Please, not that this is a list of selected publications and it does not represent an exhaustive list of publications. For a complete and up-to-date publications, please, check the Google Scholar profile.
Research Interests
Artificial intelligence and machine learning
Neural data compression, knowledge distillation, generative models (i.e., variational autoencoders, GAN, Gaussian mixture model), time series analysis, reinforcement learning, and transfer learning.
Wireless communications
Link adaptation, CSI compression in MIMO systems, channel selection, traffic load forecasting, MAC selection, power saving and energy saving for 5GB communication systems, and frequency .offset estimation.
Journals
Enabling Efficient Data Integration of Industry 5.0 Nodes Through Highly Accurate Neural CSI Feedback
Mostafa Hussien, et al.,
IEEE Transactions on Consumer Electronics (IF=4.4) | Q1
In this work, we propose a new method for CSI compression by learning an approximation for a sufficient statistics function. Our method establishes a new category of compression techniques based on the theory of sufficient statistics. Moreover, we present a detailed analysis of the upper bound of the prediction error in our specific scenario. We develop a Bayesian optimization framework to optimally select the adopted neural network architecture.
PDF | Link
A Learning Framework for Bandwidth-Efficient Distributed Inference in Wireless IoT
Mostafa Hussien, Kim Khoa Nguyen, and Mohamed Cheriet
IEEE Sensors Journal (IF=4.3) | Q1
In this work, we argue that data compression mechanisms and entropy quantizers should be co-designed with the sensing goal, specifically for machine-consumed data. To this end, we propose a novel deep learning-based framework for compressing and quantizing the observations of correlated sensors.
PDF | Link
Self Supervised Learning for CSI Compression in FDD Massive MIMO Systems
Mostafa Hussien, Kim Khoa Nguyen, and Mohamed Cheriet
IEEE Communication Letters (IF=3.55) | Q1
we propose a novel learning-based technique for CSI compression. Our method exploits the bias/variance tradeoff for CSI compression based on a shallow neural network. Unlike prior work that adopts projection-based techniques for compression, our method interprets the model weights as the compressed representation.
Carrier Frequency Offset Estimation in 5G NR: Introducing Gradient Boosting Machines
Mostafa Hussien, et al.
IEEE Access (IF=3.36) | Q1
In this work, we propose an ML-based approach for CFO estimation in OFDM systems. Specifically, we propose a Gradient-Boosting Machine (GBM)-based solution to predict the CFO given the received Primary Synchronization Signal (PSS) and Secondary Synchronization Signal (SSS).
Conferences
Efficient Neural Data Compression For Machine Type Communications via Knowledge Distillation
Mostafa Hussien, Yi Tian Xu, Di Wu, Xue Liu, Gregory Dudek
IEEE Global Communications Conference (GLOBECOM 2022)
In this paper, we propose a novel encoder for data compression in mMTC communications, which is termed Distillation Encoder (DE). Unlike prior work, the design of the proposed DE aims to achieve high compression ratios while preserving the accuracy of the inferred decisions. DE inherits the knowledge of a large teacher model (trained on the raw data) through knowledge distillation.
PDF | Link
Modeling and Optimizing Resource-Constrained Instance-Based Transfer Learning
Mohammad Askarizadeh, Mostafa Hussien, Alireza Morsali, and Kim Khoa Nguyen
IEEE Global Communications Conference (GLOBECOM 2022)
In this paper, we propose a new notion namely, regret of learner (RoL) as a quantitative measure for the learner’s performance, computation costs, and communication resources of TL. Then, we use a convex combination of the empirical source and target errors with respect to the feasibility and resource constraints to design an optimization model called OPTL that deploys a TL model in a resource-constrained environment to avoid NT.
PDF | Link
PRVNet: A Novel Partially-Regularized Variational Autoencoders for Massive MIMO CSI Feedback
Mostafa Hussien, Kim Khoa Nguyen, and Mohamed Cheriet
IEEE Wireless Communications and Networking Conference (WCNC 2022)
In this paper, we introduce PRVNet, a neural network architecture inspired by variational autoencoders (VAE) to compress the CSI matrix before feeding it back to the base station under noisy channel conditions. Moreover, we propose a customized loss function that best suits the special characteristics of the problem being addressed. We also introduce an additional regularization hyperparameter for the learning objective, which proved to be crucial for achieving competitive performance. In addition, we provide an efficient way to tune this hyperparameter using KL-annealing.
PDF | Link
Towards More Reliable Deep Learning-Based Link Adaptation for WiFi 6
Mostafa Hussien, Mohammed F. A . Ahmed, Ghassan Dahman, Kim Khoa Nguyen, Gwenael Poitau, and Mohamed Cheriet
IEEE International Conference on Communications (ICC 2021)
In this paper, we proposed modelling the adaptive modulation and coding (AMC) problem as a multi-label, multi-class classification. We proposed a new powerful loss function to improve the reliability of the proposed CNN-based framework.
PDF | Link | Certificate
Optimized Transfer Learning For Wireless Channel Selection
Mohammad Askarizadeh, Mostafa Hussien, Kim Khoa Nguyen, and Masoumeh Zare
IEEE Global Communications Conference (GLOBECOM 2021)
In this paper, we propose a rigorous model to evaluate the feasibility and optimality of transfer learning for CMAB-based channel selection in communication systems. To this end, we introduce a utility model for evaluating these economical aspects. Leveraging the Best Approximation Theory, we introduce a new similarity concept and a transfer rule applied in the context of channel selection.
PDF | Link
Fault-Tolerant 1-bit Representation for DistributedInference Tasks in Wireless IoT
Mostafa Hussien, Kim Khoa Nguyen, and Mohamed Cheriet
IEEE Conference on Network and Service Management (CNSM 2021)
In this work, we consider compressing the data of correlated sensors in a way that maximizes the inferred-decision accuracy at the FC. We propose an end-to-end learning framework that considers high-latency communication channels. The framework is augmented with a novel loss function and a three-stage training algorithm for learning discriminative low-bit (binary) features at each sensor.
Optimized Transfer Learning : Application for Wireless Channel Selection
Mohammad Askarizadeh, Mostafa Hussien, Kim Khoa Nguyen, and Masoumeh Zare
IEEE Conference on Network and Service Management (CNSM 2021)
We introduce a new concept of similarity based on the Best Approximation Theory and a general transfer rule. Then, we propose a model to evaluate the feasibility and optimality of TL. We verify our proposed model in the context of the wireless channel selection problem using contextual multi-armed bandits. Experimental results show optimal TL decisions can be made, and Extra Action is an efficient technique for TL in channel selection.
PDF | Link