FUSION: A Fuzzy-Based Multi-Objective Task Management for Fog Networks

While the Guarantee Ratio (GR) is critically important in delay-sensitive fog applications, the existing deadline-aware task assignment strategies prioritize the balance of utilization over this criterion. Therefore, this paper introduces FUSION: a fuzzy-based task management policy, which provides a high GR with the least possible makespan. FUSION considers the effect of propagation, uplink/downlink delays, and also the bandwidth between the layers on the tasks’ completion time during offloading. It benefits from a fuzzy offloader, along with a VM-ranking strategy based on a fuzzy quantified proposition. Hence, it uses two simple and efficient fuzzy ranking approaches, i.e., Decomposition and OWA. By employing fuzzy-based models, FUSION can handle uncertainty in rapidly changing fog environments with time-varying task sets with minimal computation complexity against existing meta-heuristic algorithms. FUSION considers tasks’ size with respect to VM’s processing capacity (MIPS), arrival rate, length, deadline, processing time, and execution time. In addition to VMs’ load, and busy time, FUSION considers laxity as one of its VM-ranking objectives. FUSION also conducts load-balancing, but only when it can improve the rankings to not affect the GR. Based on the iFogSim simulations, FUSION provides a higher GR in 63% of the scenarios compared to state-of-the-art. Furthermore, evaluations of the offloader algorithm indicate that FUSION provides higher GR in more than 55% of the scenarios.