Applying Artificial Intelligence Techniques For Resource Management in the Internet of Things (IoT)

Abstract

Internet of Things (IoT) applications in smart cities (SCs) rely on free-flow services streamlined by artificial intelligence (AI) paradigms. However, the nature of resource constraint prevails due to external infrastructure costs and energy-based allocations. Existing approaches to smart city resource distribution rely on static thresholds or reactive responses, which are not always sufficient. These approaches may limit system performance and scalability in dynamic IoT environments owing to increased energy consumption, postponed resource allocation, and frequent device failures. This article introduces a Concerted Resource Management (CRM) using the Leveled Reinforcement Training (LRT) method. The proposed method accurately identifies cost-complex and high energy-consuming sharing intervals based on service response time and device failure. The reinforcement learning and training concerts both energy and device incorporations for SC applications based on its demand. This process requires leveled training in resource management, from energy depletion to device activeness. The interrupted sessions are identified using resource allocation failures, and the active resources with optimal energy expenses are selected to pursue resource management. The training method thus identifies the demands based on independent or concerted resource allocations to mitigate the management constraints in an SC environment. This proposed method reduces the resource constraint-based waiting for allocations and allocation failures in any SC application services. Under the varying devices, the following is observed: Improvements: 9.1% (Allocation Rate), 10% (Device Detection), 11.88% (Constraint Mitigation— Energy), 9.06% (Constraint Mitigation—Resource Allocation); Reduced: 8.01% (Allocation Failure), 9.64% (Waiting Time).