A Hybrid AI-Based Resource Optimization Framework for Energy-Efficient and Low-Latency IoT Systems in Smart Infrastructure Applications
Abstract
Due to the explosive growth of Internet of Things (IoT) devices in the smart infrastructure setting, the scope of distributed sensing, data creation, and real-time service prompting has experienced unprecedented growth. Nevertheless, mass IoT implementations have severe issues to deal with in terms of computational resource limitations, network overload, energy waste, and time-constrained application needs. Traditional approaches to resource management commonly make use of fixed allocation policy or single model based intelligence systems which are not dynamically adjusted in response to the heterogeneous and time varying workloads. To overcome said shortcomings, the present paper will suggest a Hybrid AI-Based Resource Optimization (HAI-RO) model that would optimize the computational efficiency and decrease latency in an IoT-enabled smart infrastructure system. The suggested structure combines workload prediction based on machine learning and adaptive resource allocation based on reinforcement learning to reach intelligent and context-aware optimization on an edge and cloud layer. A dynamic energy aware scheduling system allows further minimization of power usage without affecting Quality of Service (QoS). Experimental testing shows there was a great improvement in the time of task completion, energy consumption, throughput and reduced latency in comparison to the traditional allocation models. The proposed HAI-RO model is a scalable, flexible, and sustainable solution to the next-generation IoT smart infrastructure applications.