Dynamic and Stochastic Models for Application Management in Distributed Computing Systems

Fog and edge computing have proven indispensable in tackling issues related to time-critical applications, high network congestion, user confidentiality, and data protection. While these emerging paradigms offer significant potential, substantial effort is required to study and design systems and applications tailored to their unique characteristics. This study conducts a comprehensive analysis of distributed application scheduling and offloading across cloud, fog, and edge environments. We developed multiple prototypes to investigate the organization of distributed applications under various system scales and workloads. To evaluate the system’s effectiveness and reliability, we computed steady-state probabilities using enhanced Markov models specifically designed for cloud, fog, and edge settings. These probabilities were employed to establish key metrics for assessing the efficiency of distributed application scheduling and offloading, including network utilization, response delay, energy consumption, and associated costs.