An adaptive offloading framework for improving performance of applications in IoT devices using fuzzy multi criteria decision making

The recent advances of Internet of Things (IoT) technologies have changed the requirements of IoT device to not only provide basic sensing and communication services but also of executing more complex applications with different goals. These challenges have highlighted the need to provide high computation capability in IoT devices. However, common limited resources in IoT devices bring challenges to support application requirements as well as to deal with limited computation resources. To address with this problem, computation offloading can be applied. In this approach heavy computational tasks can be transferred and executed in the cloud computing service to get the result. However, sending heavy computational jobs along with the data to the cloud server are not always efficient, especially where the mobile environments where network performances may changes unpredictably. This paper proposes a prototype of smart offloading framework designed to work in IoT devices using the Fuzzy Multi Criteria Decision Making as the decision tool. The decision whether the job execution will be done in the IoT device itself or being uploaded to the cloud computing server is done by considering internal and external factors such as current network conditions. The smart offloading framework prototype has been developed and tested in a real IoT device. The experiment results showed that the smart offloading approach can improve the performance of applications running in an IoT device by deciding location of job executions in dynamic situations with good results.