Intelligent Edge Computing: AI-Powered Optimization for Smart IoT System

In the era of rapidly evolving Internet of Things (IoT) ecosystems, the convergence of Artificial Intelligence (AI) with Edge Computing has emerged as a transformative paradigm to meet the stringent requirements of low latency, reduced bandwidth usage, and enhanced data privacy. This study presents an Intelligent Edge Computing (IEC) framework powered by AI-based optimization techniques designed specifically for smart IoT systems operating in real-time environments. The proposed system utilizes lightweight Deep Reinforcement Learning (DRL) for dynamic task offloading and scheduling, and a Swarm Intelligence-based Resource Allocation (SIRA) algorithm to optimize energy consumption and computational load across edge nodes. Additionally, the system leverages Federated Learning (FL) for decentralized model training while maintaining data security and minimizing transmission overhead. Experimental evaluations conducted using the iFogSim simulator across smart home, industrial automation, and healthcare monitoring scenarios demonstrate the effectiveness of the IEC framework. Key results include a 32.8% reduction in average task latency, 27.4% improvement in energy efficiency, and 22.5% increase in task success rate compared to traditional cloud-based architectures. The IEC framework also achieved 94.6% model accuracy using FL with minimal privacy leakage. These results affirm that AI-powered edge optimization can significantly enhance the performance and scalability of smart IoT systems while ensuring sustainable and secure operations..