Abstract
The widespread adoption of artificial intelligence (AI) and machine learning (ML) technologies for managing complex network infrastructures is constrained by the lack of standardized, relevant, and labeled datasets needed for training and validation. This paper presents a methodology for developing and validating a deterministic simulation model of a distributed corporate network in Cisco Packet Tracer. The proposed model implements a hierarchical corporate network architecture that supports VLANs, dynamic BGP-based routing, fault-tolerance mechanisms, and comprehensive cybersecurity policies. The model serves not only as an educational and methodological platform for developing practical networking skills, but also as a controlled testbed for generating synthetic telemetry data. A key feature of the proposed approach is the integration of an automated assessment system that ensures network configuration compliance with predefined reference parameters, thereby guaranteeing the correctness and reproducibility of the generated data. Configuration verification ensures reproducible network states and enables accurate automated data labeling. Experimental validation confirmed the model's deterministic nature and suitability for generating training and test datasets for intelligent network management algorithms, including anomaly detection, QoS optimization, and automated incident response. The proposed approach can be effectively applied in both educational and research contexts.