Queueing theory is a fundamental framework for analyzing the behavior of telecommunication systems, particularly in packet-switched networks where stochastic arrivals and limited service capacity lead to queue formation. Although these concepts are typically introduced through mathematical models, students often encounter difficulties in relating theoretical results to practical network behavior. Hence, this paper presents a simulation tool designed to support the teaching of queueing theory in communication networks. Our Python-based solution enables both the generation of synthetic traffic based on statistical distributions and the use of real traffic data imported from packet traces or external datasets. The tool supports multiple queueing disciplines, including first-come-first-served (FCFS), priority-based scheduling, round-robin, and shortest processing next (SPN), allowing users (students) to analyze their impact under identical traffic conditions, i.e., input flows. A set of representative simulation scenarios is used to demonstrate the functionality of the tool and to illustrate differences between queueing disciplines.
Accurate wind power forecasting for offshore wind farms in Arctic climatic conditions is challenging due to the nonlinear, intermittent, and non-stationary nature of wind generation, further intensified by climate-change-induced variability. This study presents a short-term forecasting framework using a 40-year hourly dataset that integrates synthetic wind farm power generation data with satellite-based meteorological reanalysis variables, thereby capturing both turbine-specific power characteristics and atmospheric conditions relevant to offshore wind forecasting. After feature selection and physics-aware outlier removal based on turbine operational constraints, the multivariate time series is modelled using a Gated Recurrent Unit (GRU) neural network for one-hour-ahead wind power forecasting. Particle Swarm Optimization (PSO) is employed to tune key hyperparameters, while early stopping is used to reduce overfitting. The optimized GRU model is evaluated using ReLU and Leaky ReLU activation functions to assess their effect on forecasting performance. The proposed integration of physics-based data cleaning, PSO-based optimization, and GRU-based temporal modelling provides an effective approach for large-scale offshore wind power forecasting and supports smart-grid operation, generation scheduling, and renewable energy management.
This paper presents a smartphone-like software approach to enable user applications in vehicles, including the capability of accessing critical vehicle functions. Base is the so-called AGEDA framework driven by Aumovio which acts as middleware between user-downloadable software and the closed and secured vehicle system. Unlike existing frameworks mainly designed for entertainment or navigation in manually-driven cars, it will allow after-sales installation and updates of functions with possible influence on the vehicle’s behavior. This includes assistance functions as well as extended functionality for higher automation levels, and the further connection to application-specific cloud services. After an overview of this middleware is given, the paper presents two example applications where functional reconfiguration is necessary and beneficial: collective and cloud-enhanced traffic awareness, and decentral logistics via physical internet. Both examples are demonstrated in real-life with hardware in vehicles and prototypic back-end services. To show the future potential, we present further ideas on how our framework can provide benefits to end-users, fleet providers, and manufacturers.
This work is the result of research that addresses the issue of energy dynamics of unmanned aerial vehicles (UAVs) depending on environmental stressors such as temperature, wind and turbulence. It evaluates their impact on performance and battery life. The goal is to improve the understanding of the effectiveness of unmanned aerial vehicles in different climatic conditions when deployed in agriculture and infrastructure monitoring. According to the sources available to date, no similar research has been published to this one, in which the thermal, aerodynamic and electrochemical variables in question would be combined into one common indicator. The research results show that environmental loading affects both flight duration and energy requirements. Turbulence increases energy expenditure due to aerodynamic drag and constant stabilization. We synthesized these factors into a coherent energy consumption model. This approach overcomes the limitations of previously tested models by incorporating empirical correction factors that refine the accuracy of the calculation under unstable operating conditions. This study fills a gap in energy-aware mission planning and provides insights into trajectory optimization and parameter tuning. Our findings expand the understanding of UAV performance and demonstrate that performance must be evaluated beyond nominal battery specifications. Our research supports endurance prediction, route design, and energy assessment methodologies in challenging terrain.
The reliability of electronic assessment depends not only on the correctness of test responses, but also on the response processes through which those answers are produced. This study proposes an exploratory multimodal triangulation framework for analyzing response behaviour in electronic assessment by integrating three human–computer interaction indicators: spatial mouse activity, gaze distribution, and response-time patterns. The initial sample consisted of twenty-five students who completed a thirty-item multiple-choice test in Health Management while mouse movement, gaze activity, and response time were recorded simultaneously using a custom logging application and an Eye Tribe eye-tracking device. Following gaze-data quality screening, seven participants were retained for the complete multimodal analysis. Mouse-based and gaze-based heatmaps were generated, normalized, and compared using Intersection over Union, Pearson correlation, and Spearman rank correlation, together with participant-level response-time statistics. Participant-specific translational correction was applied to compensate for gaze drift, with final offsets confirmed using recognizable and spatially stable interface elements. The results showed limited strict spatial overlap between mouse and gaze heatmaps, with a mean IoU of 0.114 at the 0.5 threshold, but moderate distributional correspondence, particularly in rank-based similarity. Response time was not directly associated with item-level accuracy, while participant-level temporal indicators became more informative when interpreted together with spatial measures. The strongest exploratory association with average test performance was observed for Spearman mouse–gaze correspondence. The retained and excluded participants showed no statistically detectable differences in test performance or average response time, although the possibility of selection bias cannot be conclusively excluded because of the small sample. These findings suggest that mouse movement, gaze behaviour, and response time should be interpreted as complementary indicators of the response process rather than as isolated predictors. The proposed framework provides exploratory evidence that multimodal HCI data may support a richer analysis of response behaviour and assessment integrity in electronic testing, but validation on larger samples and with more robust eye-tracking procedures is required.
