JoWUA

Abstract

The adoption rate of Internet of Things (IoT) devices has posed new challenges to computing systems, particularly to energy efficiency, data processing, and real-time decision making. Traditional cloud-centric frameworks face severe bottlenecks of high communication latency, bandwidth limitations, and energy costly data transmission. While offering unparalleled computational and storage capabilities, these frameworks suffer from high cloud-centric latency bottlenecks. Conversely, edge-centric frameworks process data nearer to the source, providing lower latency but suffering from limited processing capacity, memory, and scalability. These opposing concerns highlight the need for a collaborative framework. This paper proposes a dynamically adaptive task allocation algorithm, Hierarchical Edge-Cloud Collaborative AI Algorithm which seeks to balance workloads between edge nodes and cloud servers based on task intensity, latency, and energy constraints. The framework utilizes machine learning for workload forecasting to predict computational demand and reinforce task allocation in real-time to dynamically improve workload distribution efficiency. The proposed framework helps maintain the desired responsiveness and quality of service with optimal energy consumption by adaptively balancing workloads. Results from tests performed confirm the new algorithm outperformed the conventional IoT Management systems significantly. As indicated in the results, intelligent offloading decisions can provide up to 45% energy savings and also reduce latency by 30%. Enhancements in both operational efficiencies and user experience can be achieved with such offloading decisions. Furthermore, the system demonstrates strong scalability and retains performance with an increasing number of connected devices. These results demonstrate the promise of sustainability and practicality in managing IoT infrastructures using ubiquitous computing with collaborative hierarchical AI.