Radoslav Furnadzhiev

This paper presents a comparative study of two unsupervised machine learning methods, Isolation Forest (IF) and Gaussian Mixture Model (GMM), for anomaly detection in Automatic Dependent Surveillance–Broadcast (ADS-B) data. While ADS-B provides critical flight telemetry, its high volume and inherent noise pose significant challenges for automated surveillance. To address this, we developed a distributed Extract Transform Load (ETL) pipeline using Apache Spark to analyze a dataset of 4.82 billion records. The results show a low Pearson correlation between the two models, suggesting that they capture complementary types of anomalies: IF identifies structural outliers through recursive partitioning, while GMM detects statistical deviations from learned probability densities. This study demonstrates that an ensemble of lightweight methods can effectively scale to billion-record datasets, with IF offering superior computational efficiency (∼1.7 h) compared to GMM (∼4.8 h). These results provide a robust baseline for developing scalable, real-time aviation monitoring systems to enhance global air traffic safety.

Kubernetes clusters are deployed across data centers for geo-redundancy and low-latency access, resulting in new challenges in scheduling workloads optimally. This paper presents a practical evaluation of network-aware scheduling in a distributed Kubernetes cluster that spans multiple network zones. A custom scheduling plugin is implemented within the scheduling framework to incorporate real-time network telemetry (inter-node ping latency) into pod placement decisions. The assessment methodology combines a custom scheduler plugin, realistic network latency measurements, and representative distributed benchmarks to assess the impact of scheduling on traffic patterns. The results provide strong empirical confirmation of the findings previously established through simulation, offering a validated path forward to integrate not only network metrics, but also other performance-critical metrics such as energy efficiency, hardware utilization, and fault tolerance.

Distributed Kubernetes clusters provide robust solutions for geo-redundancy and fault tolerance in modern cloud architectures. However, default scheduling mechanisms primarily optimize for resource availability, often neglecting network topology, inter-node latency, and global resource efficiency, leading to suboptimal task placement in multi-region deployments. This paper proposes network-aware scheduling plugins that integrate heuristic, metaheuristic, and linear programming methods to optimize resource utilization and inter-zone communication latency for containerized workloads, particularly Apache Spark batch-processing tasks. Unlike the default scheduler, the presented approach incorporates inter-node latency constraints and prioritizes locality-aware scheduling, ensuring efficient pod distribution while minimizing network overhead. The proposed plugins are evaluated using the kube-scheduler-simulator, a tool that replicates Kubernetes scheduling behavior without deploying real workloads. Experiments cover multiple cluster configurations, varying in node count, region count, and inter-region latencies, with performance metrics recorded for scheduler efficiency, inter-zone communication impact, and execution time across different optimization algorithms. The obtained results indicate that network-aware scheduling approaches significantly improve latency-aware placement decisions, achieving lower inter-region communication delays while maintaining resource efficiency.

Cloud computing has revolutionized the way organizations manage their IT infrastructure by providing a flexible, scalable, and cost-effective solution for deliv Cloud computing has revolutionized the way organizations manage their IT infrastructure by providing a flexible, scalable, and cost-effective solution for deliv

Driver behavior models are an important part of road traffic simulation modeling. They encompass characteristics such as mood, fatigue, and response to distracting conditions. The relationships between external factors and the way drivers perform tasks can also be represented in models. This article proposes a methodology for establishing parameters of driver behavior models. The methodology is based on road traffic data and determines the car-following model and routing algorithm and their parameters that best describe driving habits. Sequential and parallel implementation of the methodology through the urban mobility simulator SUMO and Python are proposed. Four car-following models and three routing algorithms and their parameters are investigated. The results of the performed simulations prove the applicability of the methodology. Based on more than 7000 simulations performed, it is concluded that in future experiments of the traffic in Plovdiv it is appropriate to use a Contraction Hierarchies routing algorithm with the default routing step and the Krauss car-following model with the default configuration parameters.

Containerization is still a relatively new and fast evolving technology. It has been widely adopted in cloud environments, because of its lower overhead compared to virtualization. Besides the benefits of using containers for running applications in virtual environments a lot of challenges in terms of security arise. The paper presents an overview of the risks and thread plaguing modern OS-virtualized systems and the security mechanisms on every level of containerized deployments. It outlines some of the existing open issues and suggests the use of a multi-layered security approach that should not only address all current security issues, but provide resilience against future threats.