COMPARATIVE ANALYSIS OF OPTIMIZATION MODELS FOR RESOURCE ALLOCATION IN UNCERTAIN IT ENVIRONMENTS

Аңдатпа

This study addresses the challenge of resource allocation in dynamic and uncertain environments by evaluating five mathematical models: Markov Decision Process (MDP), Integer Programming, Multi-Agent Systems, Scheduling Problems, and Assignment Models. The relevance of this problem lies in the need for effective decision-making tools under time and resource constraints, particularly in IT and service-oriented industries where operations are often unpredictable and require adaptive optimization strategies. The main aim of this research is to perform a comparative analysis of the selected models using performance metrics such as execution time, cost, and memory consumption. The research methodology involves conducting a series of simulations on synthetically generated datasets that reflect realistic operating conditions in terms of workload variability, task dependencies, and limited resources. Results indicate that Integer Programming delivers the lowest solution cost but requires significantly more execution time, making it suitable for scenarios where accuracy outweighs speed. On the other hand, MDP and Multi-Agent models offer faster computation and flexibility, but with relatively higher solution costs. Scheduling and Assignment Models demonstrate a balanced trade-off but are less scalable under complex constraints. These findings highlight the inherent trade-off between time efficiency, computational complexity, and solution optimality. The proposed comparison contributes to identifying suitable approaches based on task-specific priorities and operational goals. The theoretical significance of the work lies in the integration of multiple optimization techniques into a structured comparative framework, enhancing the understanding of model behaviour under uncertainty. The practical contribution focuses on guiding system architects and decision-makers in selecting the most appropriate modelling tools for real-time systems in IT and service environments. Future work will explore enhancements to scalability, hybrid modelling strategies, and adaptive algorithms to support industrial-scale implementations with dynamic data inputs.