Computation offloading model for smart factory

Abstract

Internet of Things (IoT) is coming up in a rapid pace in various application domains. In smart factories, IoT can be deployed using sensors and actuators for taking smart manufacturing decisions. To maintain the smartness, huge computational power is required to handle the generated data by the IoT sensors. Local servers, in smart factories, usually organize the sensors/actuators and takes decisions at the local level. However, they are not equipped with enough computational power to handle all types of computational tasks and therefore some tasks need to be offloaded to the upper layer such as Cloud. Hybrid cloud i.e. public cloud along with some local servers can better handle this requirement. Recently, an additional layer called fog computing is introduced in the cloud architecture to complement it with added power. Offloading of tasks, generated by the industrial applications of IoT devices, should be done only when existing computational power of local server is not able to meet the quality requirements of the tasks. An ultimate objective of the smart factory owner is to earn revenue and for that, IoT devices need to meet their quality of service expectation. For offloading, tasks can be categorized as delay sensitive and delay tolerant and making decision on offloading by the local server is non-trivial. This work proposes an offloading decision model using game theory in a non-cooperative environment considering the categorization of tasks and it is shown that dominant strategy exists for the local server. For the performance study of the proposed model, simulation is done using iFogSim simulator. A comparative study with state-of-art exhibits that the proposed offloading scheme outperforms.