Emergent Architectures in Edge Computing Revolutionizing AI-Driven Approaches for Low Latency Applications

The rapid growth of real-time applications, such as intelligent surveillance and Internet of Things (IoT)-based automation, demands low-latency, high-efficiency computing architectures. Traditional cloud-based models struggle with high response times, bandwidth limitations and inefficient workload distribution, necessitating emergent architectures in edge computing for enhanced performance. To address these challenges, this study proposes the Emergent Edge Architecture (EEA), a three-tier hierarchical edge computing framework integrating artificial intelligence (AI)-driven dynamic load balancing, Function-as-a-Service (FaaS) and reinforcement learning (RL)-based task allocation. The architecture dynamically distributes workloads across Near Edge (user devices), Far Edge (edge servers) and Cloud Backup, ensuring optimal latency reduction, energy efficiency and real-time processing. A real-time video analytics use case leveraging You Only Look Once version 5 (YOLOv5) object detection in a 5G-enabled edge environment validates EEA's effectiveness. Experimental results demonstrate a 70% reduction in latency, a 142% increase in throughput and a 45.6% decrease in power consumption compared to cloud-based processing. Additionally, the system maintains a 67.7% improvement in task completion time under high concurrency. These findings confirm that EEA enhances performance, scalability and energy efficiency, making it well-suited for low-latency applications in smart surveillance, industrial IoT and intelligent traffic monitoring.