Tuğba Selcen Navruz

Missile warning systems play a critical role in ensuring the survivability of airborne platforms by enabling early detection of incoming threats. Due to low signal-to-noise ratios, dynamic backgrounds, and the small size of targets in the early stages, missile plume detection in passive electro-optical approaches is still difficult. Solar-blind ultraviolet (SBUV) sensing offers improved target-to-background contrast under daylight conditions, making it a suitable modality for missile plume detection. This study assesses different You Look Only Once (YOLO) based object detection architectures (YOLOv5, YOLOv8, YOLOv10, and YOLOv11) for missile plume detection using a synthetic SBUV dataset in order to assess the models' detection performance and computational efficiency. They are trained and compared under the same conditions. To provide a thorough analysis, standard evaluation metrics, which are precision, recall, f1 score and mean average precision (mAP) are combined with measurements of inference speed. The results indicate that all evaluated models achieve high detection performance, while notable differences emerge in inference speed and robustness across varying confidence thresholds. Although newer architectures such as YOLOv10 and YOLOv11 introduce more advanced design components, they do not consistently provide lower end-to-end latency under the evaluated conditions. In contrast, YOLOv8 achieves the lowest inference time, with YOLOv5 demonstrating comparable performance and more stable behavior across confidence thresholds. Furthermore, confusion matrix analysis reveals that newer YOLO models reduce background-related false detections compared to YOLOv5, indicating improved discrimination capability in complex scenes. Overall, the findings confirm that YOLO-based approaches are well-suited for SBUV-based missile plume detection, and that recent architectural developments primarily enhance robustness and detection quality rather than consistently reducing inference latency in practical deployment scenarios.

In this research, we delve deep into the realm of Targeted Social Engineering Email Detection, presenting a novel approach that harnesses the power of Lambda Architecture (LA). Our innovative methodology strategically segments the BERT model into two distinct components: the embedding generator and the classification segment. This segmentation not only optimizes resource consumption but also improves system efficiency, making it a pioneering step in the field. Our empirical findings, derived from a rigorous comparison between the fastText and BERT models, underscore the superior performance of the latter. Specifically, The BERT model has high precision rates for identifying malicious and benign emails, with impressive recall values and F1 scores. Its overall accuracy rate was 0.9988, with a Matthews Correlation Coefficient value of 0.9978. In comparison, the fastText model showed lower precision rates. Leveraging principles reminiscent of the Lambda architecture, our study delves into the performance dynamics of data processing models. The Separated-BERT (Sep-BERT) model emerges as a robust contender, adept at managing both real-time (stream) and large-scale (batch) data processing. Compared to the traditional BERT, Sep-BERT showcased superior efficiency, with reduced memory and CPU consumption across diverse email sizes and ingestion rates. This efficiency, combined with rapid inference times, positions Sep-BERT as a scalable and cost-effective solution, aligning well with the demands of Lambda- inspired architectures. This study marks a significant step forward in the fields of big data and cybersecurity. By introducing a novel methodology and demonstrating its efficacy in detecting targeted social engineering emails, we not only advance the state of knowledge in these domains but also lay a robust foundation for future research endeavors, emphasizing the transformative potential of integrating advanced big data frameworks with machine learning models.

This study introduces a novel methodology designed to assess the accuracy of data processing in the Lambda Architecture (LA), an advanced big-data framework qualified for processing streaming (data in motion) and batch (data at rest) data. Distinct from prior studies that have focused on hardware performance and scalability evaluations, our research uniquely targets the intricate aspects of data-processing accuracy within the various layers of LA. The salient contribution of this study lies in its empirical approach. For the first time, we provide empirical evidence that validates previously theoretical assertions about LA, which have remained largely unexamined due to LA’s intricate design. Our methodology encompasses the evaluation of prospective technologies across all levels of LA, the examination of layer-specific design limitations, and the implementation of a uniform software development framework across multiple layers. Specifically, our methodology employs a unique set of metrics, including data latency and processing accuracy under various conditions, which serve as critical indicators of LA’s accurate data-processing performance. Our findings compellingly illustrate LA’s “eventual consistency”. Despite potential transient inconsistencies during real-time processing in the Speed Layer (SL), the system ultimately converges to deliver precise and reliable results, as informed by the comprehensive computations of the Batch Layer (BL). This empirical validation not only confirms but also quantifies the claims posited by previous theoretical discourse, with our results indicating a 100% accuracy rate under various severe data-ingestion scenarios. We applied this methodology in a practical case study involving air/ground surveillance, a domain where data accuracy is paramount. This application demonstrates the effectiveness of the methodology using real-world data-intake scenarios, therefore distinguishing this study from hardware-centric evaluations. This study not only contributes to the existing body of knowledge on LA but also addresses a significant literature gap. By offering a novel, empirically supported methodology for testing LA, a methodology with potential applicability to other big-data architectures, this study sets a precedent for future research in this area, advancing beyond previous work that lacked empirical validation.

A lattice matched triple junction solar cell (TJSC) structure with a GaAs_0.58 P_0.42 top cell and bandgap tunable GaN_xAs_1-x-zP_z middle and bottom cells on virtual SiGe substrate is proposed in this study. SiGe/Si substrate is preferred as it is a low-cost substrate and because it provides a lattice constant at which bandgap tunable dilute nitride materials that are appropriate for highly efficient multijunction solar cells can be obtained. By changing the nitrogen content in GaN_xAs_1-x-zP_z, the bandgap of the middle and bottom subcells is adjusted to the optimum values. The bandgap of the top cell is constant at 1.95 eV. Three models with different values of surface recombination velocities and Shockley–Read–Hall recombination lifetimes are applied to the presented TJSC structure. Peak efficiencies of 48.9%, 40.6% and 33.7% are achieved at E_G2 = 1.45 eV and E_G3 = 1.04 eV for Model 1, E_G2 = 1.45 eV and E_G3 = 1.15 eV for Model 2, and E_G2 = 1.5 eV and E_G3 = 1.17 eV for Model 3, respectively. A fourth bandgap adjustable GaN_xAs_1-x-zP_z junction is inserted into the system and a significant improvement is obtained under high sun concentration for Models 1 and 2. The presented original results are very promising because the variable bandgaps provide very efficient absorption of incoming spectrum.

The integration of compact concepts and advances in permanent-magnet technology improve the safety, usability, endurance, and simplicity of unmanned aerial vehicles (UAVs) while also providing long-term operation without maintenance and larger air gap use. These developments have revealed the demand for the use of magnetic couplers to magnetically isolate aircraft engines and starter-generator shafts, allowing contactless torque transmission. This paper explores the design aspects of an active cylindrical-type magnetic coupler based on finite element analyses to achieve an optimum model for hybrid UAVs using a piston engine. The novel model is parameterised in Ansys Maxwell for optimetric solutions, including magnetostatics and transients. The criteria of material selection, coupler types, and topologies are discussed. The Torque-Speed bench is set up for dynamic and static tests. The highest torque density is obtained in the 10-pole configuration with an embrace of 0.98. In addition, the loss of synchronisation caused by the piston engine shaft locking and misalignment in the case of bearing problems is also examined. The magnetic coupler efficiency is above 94% at the maximum speed. The error margin of the numerical simulations is 8% for the Maxwell 2D and 4.5% for 3D. Correction coefficients of 1.2 for the Maxwell 2D and 1.1 for 3D are proposed.

In this study, an adaptive FastSLAM (AFastSLAM) algorithm, which is obtained by estimating the time‐varying noise statistics and improving FastSLAM algorithm, is proposed. This improvement was accomplished by using maximum likelihood estimation and expectation maximization criterion and a one‐step smoothing algorithm in importance sampling. In addition, innovation covariance estimation (ICE) method was used to prevent loss of positive definiteness of the process and measurement noise covariance matrices. The proposed method was compared with FastSLAM by calculating the root mean square error (RMSE) using different particle numbers at varying initial process and measurement noise values. Simulation studies have shown that AFastSLAM provides much more accurate, consistent, and successful estimates than FastSLAM for both robot and landmark positions.