Rafik Lasri

Accurate and timely maturity classification of Rosa damascena is important for optimizing harvest scheduling and preserving the quality of rose-derived products. However, deploying deep learning models on resource-constrained edge devices remains challenging due to computational and memory limitations. This study proposes GWOEdgeNet, a lightweight convolutional neural network architecture optimized through Grey Wolf Optimization (GWO) for real-time maturity classification on embedded systems. The GWO search identified an ultra-compact architecture with only 10,946 parameters. Knowledge distillation from a ResNet18 teacher network improved classification accuracy from 92.94% to 95.35% without increasing model complexity. Comprehensive benchmarking on Raspberry Pi 4B demonstrated that GWOEdgeNet-KD achieves 38.1 ms inference latency (26.2 FPS), outperforming eight state-of-the-art lightweight architectures including MobileNetV3, ShuffleNetV2, and EfficientNet-B0. The proposed model requires 31× fewer parameters than ShuffleNetV2-x0.5 and achieves 6.8× faster inference than EfficientNet-B0 while maintaining competitive accuracy. Experimental results confirm balanced per class performance (94.9% for immature, 94.5% for mature) and consistent inference timing under field conditions. The proposed approach demonstrates that metaheuristic neural architecture search combined with knowledge distillation enables accurate and efficient image classification on low cost agricultural hardware.

Urban congestion causes further increases in travel times, fuel consumption and greenhouse-gas emissions. In this regard, we conduct a systematic study of a Genetic Algorithm (GA) for real-time routing in an urban scenario in Bethlehem City, based on a SUMO microsimulation that has been calibrated using real data from the field. Our work makes four main contributions: (i) the implementation of a reproducible GA framework for dynamic routing with explicit constraints and adaptive termination criterion; (ii) design of a weight sensitivity study for studying a multi term fitness function with travel time and waiting time, and optionally fuel usage; (iii) an edge-assisted distributed architecture on roadside units (RSUs) supported by cloud services; and (iv) specifying and refining the data set description and experimental protocol with a planned statistical analysis. Empirical evidence from the Bethlehem case study shows a consistent decline in total travel time under high congestion cases. Variations in the waiting time between different scenarios are exhibited, reflecting the trade-offs in the fitness weighting scheme. We recognize that we have some limitations, including the manual resolution of data and the inherent problem of differences between simulations and real world, and we are proposing a road-map towards a pilot deployment that handles these issues. Rather than proposing a new GA variant, we present a deployment-oriented framework-an edge- assisted GA with explicit protocols and a latency envelope, and a reproducible multi-objective tuning procedure validated on a city-scale network under severe congestion.

The increasing integration of renewable energy sources (RES) is accelerating the shift from traditional synchronous machine-based power systems to inverter-dominated grids. This transition poses significant frequency stability challenges, as power electronic interfaces lack the inherent kinetic energy storage of conventional generators, resulting in low system inertia. To address these challenges, this study proposes an adaptive virtual inertia control system based on fuzzy logic, which offers notable advancements in frequency dynamics. The proposed controller dynamically adjusts the virtual inertia constant in real-time by leveraging inputs such as frequency and the rate of change of frequency (RoCoF). This adaptive approach overcomes the limitations of fixed inertia systems, ensuring improved frequency stability and superior transient performance during load disturbances. Simulation results validate the system's effectiveness, showing reduced frequency overshoots, minimized deviations, and faster recovery to nominal frequency compared to conventional fixed inertia methods. By rapidly damping oscillations and enhancing transient stability, the proposed system significantly outperforms traditional techniques. Moreover, the study reviews current virtual inertia strategies, control topologies, and explores future research directions for integrating advanced virtual inertia into modern grids. These findings demonstrate the robustness of fuzzy logic based adaptive inertia for stabilizing low-inertia microgrids with high RES penetration.

This paper presents the design, simulation, and test measurements of a microstrip bandstop filter operating at 1.5 GHz, incorporating six split ring resonator (SRR) unit cells. The substrate employed is an FR-4 with a thickness of 1.6 mm and tangent losses of 0.025. In the initial phase, the design is conceptualized, simulated using computer simulation technology (CST) studio and advanced design system (ADS) Agilent simulators, and validated through test measurements. Building upon this foundation, the filter is transformed into a reconfigurable variant by integrating four SMV2019 varactor diodes. These varactors are modeled to ensure the reconfigurability of the bandwidth. The integration of varactors introduces dynamic tuning capabilities to the considered bandstop filter.