Abstract
The growing volume of telemetry and network data in Internet of Things (IoT) environments imposes new requirements on information processing infrastructures in the field of cybersecurity. The high rate of event generation, heterogeneity of data sources, and the necessity for real-time response justify the use of Big Data technologies as a foundational platform for security analytics. A multi-layer architectural model of a Big Data infrastructure for IoT security is proposed, encompassing data ingestion and initial stream processing, distributed data processing, large-scale storage, analytical processing, and integration with security monitoring systems. The model incorporates stream-processing mechanisms with feature extraction in sliding time windows, distributed storage systems such as NoSQL and Data Lake solutions, and scalable tools for handling high-throughput data streams. The core component of the architecture is the integral risk indicator R(t), derived from a multidimensional feature vector and enabling a formalized quantitative assessment of anomalous activity. This approach ensures the integration of stream analytics, machine learning, and event correlation mechanisms within a scalable distributed Big Data infrastructure. The research methodology is based on simulation modeling of a DDoS attack scenario under increasing event intensity, followed by analysis of latency, detection accuracy, and risk dynamics metrics. The results confirm the scalability of the proposed architecture, the absence of exponential growth in processing latency, and the robustness of the detection mechanism under peak loads. The practical significance of the study lies in the applicability of the proposed Big Data infrastructure as a foundation for adaptive cybersecurity systems in IoT, SCADA, and industrial information environments.

This publication has 16 references indexed in Scilit: