Scheduling Bag-of-Task Jobs with Security Requirements and Partial Computations in a Fog – Cloud System

The increasing data output from IoT devices creates significant strain on the resources of conventional cloud data centers, hampering their capacity to effectively cater to the demands of IoT applications, especially those with time-sensitive requirements. Fog computing, a modern computing paradigm, extends the capabilities of cloud resources to the periphery of the network. Within these collaborative computing environments, diverse workloads consist of tasks distinguished by varying levels of security requirements. Therefore, implementing security-aware scheduling schemes is essential to ensure the proper execution of applications and attain the desired Quality of Service (QoS). Accordingly, this study delves into security-conscious scheduling methodologies tailored for Bag-of-Task (BoT) applications within a fog-cloud computing framework. Fog resources are identified as more suitable for processing BoT tasks with security prerequisites, designated as "fog jobs". Conversely, BoTs without security demands, termed "cloud jobs", may be processed on either cloud or fog resources. To accelerate fog job execution, a method leveraging partial computations for cloud job tasks is employed. This study, through simulation-based assessments, provides significant insights into the effect of these approximations on system performance, contingent upon the delay levels of cloud job tasks, albeit with a marginal decrease in result precision.