Quantifying the Effectiveness of Cloud and Edge Offloading: An Optimization Study on Energy Efficiency of Mobile Real-Time Systems

The rapid growth of IoT and AI technologies has driven the increased use of mobile embedded systems with real-time constraints. These battery-powered systems should minimize energy consumption while meeting task deadlines. Optimizing the trade-off between energy consumption and performance is crucial, particularly as workload demands fluctuate. Previous research has focused on optimizing system resources, such as processors and memory, by configuring low-power modes based on workload intensity. Task offloading to cloud or edge servers has also been studied, leveraging the ample resources of cloud servers for resource-constrained mobile systems and the ability of edge servers to efficiently meet deadline constraints with stable network connections. In this paper, we quantify the effectiveness of task offloading to cloud and edge servers in terms of power savings for mobile systems. Unlike previous approaches, we analyze the impact of offloading when co-optimized with other energy-saving techniques, such as dynamic voltage scaling and low-power memory configurations. Through extensive experiments, we explore the trade-offs between cloud and edge environments, accounting for workload intensity, network conditions, and server computing capabilities. Our findings offer valuable insights for designing optimized task offloading strategies tailored to specific mobile system characteristics, effectively balancing the benefits of cloud and edge environments.