This paper addresses the problem of designing a software architecture for an ontology-driven system for managing complex objects, aimed at ensuring efficient, adaptive, and explainable operation under conditions of dynamic workloads and heterogeneous resources. The relevance of the study is driven by the increasing complexity of modern information and communication infrastructures, particularly university corporate networks, which are characterized by high-intensity user requests, variability of workload profiles, and strict quality-of-service requirements.
An integrated approach is proposed that combines ontology-based modeling, mathematical methods for analysis and optimization, and software integration mechanisms within a unified multi-layer architecture. A formal model of an ontology-driven system is developed, providing a consistent representation of domain knowledge, dynamic data, and decision-making procedures. A dynamic load distribution model for terminal cluster centers is introduced, taking into account temporal characteristics of request flows, structural properties of the network, and multi-criteria constraints.
The software architecture of the system is substantiated, including data acquisition and preprocessing, ontology-based knowledge representation, semantic processing, analytical and optimization modules, integration mechanisms, and presentation components. A key feature of the architecture is the integration of declarative (ontology-based) and procedural (algorithmic) layers, enabling semantic interpretation of system states, application of logical reasoning mechanisms, and generation of well-founded control decisions.
The results demonstrate that combining knowledge-oriented approaches with mathematical modeling significantly improves load balancing efficiency, reduces response latency, and ensures more balanced utilization of computational and communication resources. The proposed approach is designed for integration into existing network management systems and can serve as a foundation for intelligent management of ICT infrastructures.
Future research directions include extending the models to handle uncertainty (e.g., interval and fuzzy representations), integrating machine learning techniques for workload prediction, and developing digital twins of network infrastructures for scenario-based analysis and strategic planning.
