The sixth generation (6 G) of wireless communication demands advanced personalization, low latency, and scalability to support emerging applications such as extended reality (XR) and intelligent digital twins. Conventional federated learning (FL) methods struggle to address these requirements due to their limited adaptability to diverse user contexts. To tackle these issues, we suggest FCAFL, a new federated learning method that uses Formal Concept Analysis (FCA) to group users based on their needs and combine models in a way that considers their specific situations. FCA-FL achieves state-of-the-art performance with 82% accuracy, $63 \%$ communication overhead, and 92 Mbps throughput, significantly outperforming traditional approaches. Additionally, it maintains robust signal quality with an SINR range of $26 \mathrm{~dB}-22 \mathrm{~dB}$, ensuring scalability in high-density networks. Future work will explore dynamic clustering, quantum-enhanced FL, and real-world deployments to further enhance FCA-FL’s potential as a transformative solution for personalized 6G services.

This research investigates the application of Vision Transformers (ViT) for facial emotion recognition through extensive experimentation on three benchmark datasets: CK+, FER2013, and AffectNet. Our study systematically examines how dataset characteristics influence model performance, revealing substantial variations in recognition accuracy based on data quality and environmental conditions. The ViT architecture demonstrates outstanding performance on the controlled CK+ dataset, achieving 95.2% accuracy and 93.8% F1-score, confirming its effectiveness under laboratory conditions. However, performance decreases considerably on real-world datasets, with $\mathbf{6 0. 4 \%}$ accuracy on FER2013 and $58.1 \%$ on AffectNet, emphasizing the challenges of practical deployment in uncontrolled environments. Comparative evaluation with existing methods shows that our ViT approach surpasses traditional CNN architectures by $\mathbf{1} \boldsymbol{-} \mathbf{3} \boldsymbol{\%}$ across all datasets while maintaining computational efficiency. The research identifies consistent overfitting patterns during training, with happiness demonstrating the highest recognition performance and fear-sadness pairs presenting the most significant classification challenges across all conditions.

Radio resource allocation involves managing, distributing, and optimizing available wireless network resources (frequency spectrum, transmission power, network access time, antennas or beams). The advent of 6G networks, with billions of heterogeneous connected devices and increasingly complex infrastructure, necessitates intelligent and equitable resource allocation to ensure optimal Quality of Service (QoS) while preventing congestion. In this context, we propose a Deep Reinforcement Learning (DRL)-based allocation method for 6G V2X (Vehicle-toEverything) networks. Connected vehicles continuously generate sensitive data (location, trajectories, speed), whose centralized collection for DRL training poses significant risks including data breaches, GDPR (General Data Protection Regulation) noncompliance (due to challenging anonymization requirements), and user distrust of personal data collection systems. To balance performance with privacy preservation, we introduce a hybrid DRL-FL (Federated Learning) architecture that addresses 6G vehicular network challenges while ensuring data confidentiality. Our approach reduces base station energy consumption while ensuring rigorous user privacy protection.

Third Generation Partnership Project (3GPP) introduced Vehicle-to-Everything (V2X) in Long Term Evolution (LTE) Release 14 and Release 15. The present work aimed to investigate the resource allocation problem for V2X communications. To this end, we proposed applying the Variable Neighborhood Search (VNS) algorithm to solve the resource allocation problem. The proposed method aims to maximize the total throughput in the system while maintaining a minimum latency and reliability for both Cellular User Equipment (CUEs) and Vehicle User Equipment (VUEs). To verify the effectiveness of the proposed method, the VNS-based resource allocation scheme was compared to two other schemes; the first is based on the Particle Swarm Optimization (PSO) algorithm, and the second on the Ant Colony Optimization (ACO) Algorithm. The simulation was carried out using MATLAB software. The obtained results prove the performance of the proposed scheme in terms of resource allocation for V2X communications. The performance gains of the proposed approach demonstrate its feasibility and utility for V2X communications.

Research on cooperative intelligent traffic issues has enhanced ground transportation's efficiency, safety, and comfort. The present work is interested in the resource allocation problem in V2X communications. We propose a new hybrid metaheuristic for a resource allocation scheme to maximize the system's total sum rate. This method is based on the particle swarm optimization (PSO) algorithm and a new suggested local search. We try to leverage the strengths and mitigate the weaknesses of both algorithms. The standard PSO algorithm has issues converging to optimal solutions because it lacks exploitation abilities. The local search aims to expand the search space vertically, allowing for a more balanced approach and addressing global exploration and local exploitation. We compared the proposed approach to the PSO and the Ant Colony Optimization (ACO) algorithms. The simulation was conducted using the MATLAB software platform. The results demonstrate that The algorithms proposed in this article significantly improve the system throughput and access rate of vehicular user equipment (VUEs) while ensuring the data rate of cellular user equipment (CUEs).'The results demonstrate the superiority of the proposed scheme.

Over the past decade, polar codes have gained momentum in academia and industry. They are selected for the control channel in the 3rd Generation Partnership Project (3GPP) for the 5th generation (5G) New Radio (NR) standard for mobile communications. To this end, the present work proposes a new polar code for ultra-reliable and low latency (URLLC) use case of 5G NR. Since polar codes are known by their performance in terms of error correction, two scenarios are considered. The first is interested in the evolution of Block Error Rate (BLER) as a function of downlink Signal Noise Ratio (SNR) while using Quadrature phase shift keying (QPSK) modulation on an Additive White Gaussian Noise (AWGN) channel. The second scenario is interested in the evolution of BLER as a function of uplink SNR while using QPSK modulation on an AWGN channel. Simulation results prove the performance of the new polar code type for 5G NR networks, especially in terms of throughput and BLER.

It becomes very common to use cell phones in public transportation and the cars. Vehicular networking has a major problem which is the degradation of signal quality due to interference and the large number of mobile devices. Artificial intelligence (AI) is a promising technique for next-generation wireless networks. Deep learning is a type of AI derived from machine learning; here the machine can learn by itself, unlike programming where it is content to execute rules to the letter predetermined. In addition, AI can be explored in order to solve various problems. In this paper, we tackle the problem of resource allocation in a vehicular small cell network (VSCN). Indeed, we propose a new mechanism based on deep reinforcement learning denoted Resource Allocation based Deep Reinforcement Learning (RA-DRL). The main goal of our proposed method is to maximize the total system sum rate (throughput) while guaranteeing minimum interferences, Quality of Service (QoS) and the demand for all users. Simulation results demonstrate that our proposed RA-DRL algorithm exhibits better performance comparing to the other methods, by maximizing the total system sum rate while maintaining inter-VSCs interferences and a minimum latency

Device-to-Device (D2D) enables direct communication among Vehicle-to-Everything (V2X) applications. Due to the high mobility in a vehicular environment, it is necessary to take into account the fast channel variations when defining resource allocation schemes in V2X networks. Furthermore, an efficient power control for V2X communications is needed to meet the large and growing number of vehicular devices and the growing demand for data traffic. In addition and for several types of networks, the use of transmission beamformer selection to reduce the power is a challenge. In this paper, we consider the power control problem in V2X networks. In addition, a joint beamforming and power control (BFPC) method for downlink V2X communication is proposed. It aims to reduce the total transmit power in the system while guaranteeing Quality of Service (QoS) for both, Vehicle User Equipments (VUEs) and Cellular User Equipments (CUEs). The problem is formulated as a non-convex optimization that we propose to solve it by using SemiDefinite Relaxation (SDR). The simulation results show the effectiveness of our BFPC proposed method compared to fixed power control method.

Femtocells, also known as Home enhanced Node B (HeNB), are low cost and low power wireless Access Point (AP). This device is habitually deployed in an indoor environment to provide in-building coverage enhancement. Although femtocells are low power base stations, a vast deployment of femtocells causes many technical problems such as: interferences, user Quality of Service (QoS) and power control and consumption of femtocells. To solve such problems, most exciting works focus on power control and QoS in femtocell networks. In this paper, we provide a new intelligent QoS and power control method, denoted Users and Applications QoS Priorities (UAQP) power control method. This method aims to optimize the energy consumption of Femtocell Base Station (FBS) through adjusting their transmitting power according to two types of priorities, users and applications priorities. The simulation demonstrates the effectiveness of our proposed power control method and their performance compared to fixed power control method.