Workload Allocation in IoT-Fog-Cloud Architecture Using a Multi-Objective Genetic Algorithm

With the rapid growth of Internet-of-Things (IoT) applications, data volumes have been considerably increased. The processing resources of IoT nodes cannot cope with such huge workloads. Processing parts of the workload in clouds could solve this problem, but the quality of services for end-users will be decreased. Given the latency reduction for end-users, the concept of processing in the fog devices, which are at the edge of the network has been evolved. Optimizing the energy consumption of fog devices in comparison with cloud devices is a significant challenge. On the other hand, providing the expected-quality of service in processing the requested workloads is highly dependent on the propagation delay between fog devices and clouds, which due to the nature of the distribution of clouds with the different workloads, is highly variable. To date, none of the proposed solutions has solved the problem of workload allocation given the criteria of minimizing the energy and delay of fog devices and clouds, simultaneously. This paper presents a processing model for the problem in which a trade-off between energy consumption and delay in processing workloads in fog is formulated. This multi-objective model of the problem is solved using NSGAII algorithm. The numerical results show that by using the proposed algorithm for workload allocation in a fog-cloud scenario, both of energy-consumption and delay can be improved. Also, by allocating 25% of the IoT workloads to fog devices, the energy consumption and delay are both minimized.