Smart environments require adaptive resource management to handle dynamic workloads and system variability. Traditional solutions, which often rely on static configurations or heuristic adjustments, may not maintain performance as conditions change. This work presents Oraculum, an adaptive model that integrates real-time monitoring, predictive analytics, and automated decision-making. Unlike previous architectures that apply reactive or rule-based adaptations, Oraculum incorporates predictive reinforcement learning (TD3) to anticipate environmental changes and optimize reconfiguration decisions proactively. The model applies data-driven methods to adjust system configurations dynamically, improving both resource allocation and service quality. Experimental results demonstrate that Oraculum significantly reduces Mean Adaptation Time (MAT) compared to existing self-adaptive models while achieving an adaptation accuracy of 97%, overhead of 2%, and maintaining system stability at 98%. These findings highlight the advantages of predictive control in addressing the challenges of dynamic workloads and resource constraints in smart environments, offering a practical approach for maintaining consistent performance.

This work was partially funded by national funds through FCT - Foundation for Science and Technology, I.P., under projects UIDB/04466/2025, UIDP/04466/2025, and project 16,881 (LISBOA2030-FEDER-00816400).

Predictive analytics,Reinforcement learning,Real-time monitoring,Smart environments,Self-adaptive systems

• Computer and Information Sciences - Natural Sciences
• Other Engineering and Technology Sciences - Engineering and Technology
• Electrical Engineering, Electronic Engineering, Information Engineering - Engineering and Technology