Abstract
In the last few years, we have witnessed the huge popularity of one of the most promising technologies of the modern era: the Internet of Things. In IoT, various smart objects (smart sensors, embedded devices, PDAs, and smartphones) share their data with one another irrespective of their geographical locations using the Internet. The amount of data generated by these connected smart objects will be on the order of zettabytes in the coming years. This huge amount of data creates challenges with respect to storage and analytics given the resource constraints of these smart devices. Additionally, to process the large volume of information generated, the traditional cloud-based infrastructure may lead to long response time and higher bandwidth consumption. To cope up with these challenges, a new powerful technology, edge computing, promises to support data processing and service availability to end users at the edge of the network. However, the integration of IoT and edge computing is still in its infancy. Task scheduling will play a pivotal role in this integrated architecture. To handle all the above mentioned issues, we present a novel architecture for task selection and scheduling at the edge of the network using container-as-a-service (CoaaS). We solve the problem of task selection and scheduling by using cooperative game theory. For this purpose, we developed a multi-objective function in order to reduce the energy consumption and makespan by considering different constraints such as memory, CPU, and the user's budget. We also present a real-time internal and external container migration technique for minimizing the energy consumption. For task selection and scheduling, we have used lightweight containers instead of the conventional virtual machines to reduce the overhead and response time as well as the overall energy consumption of fog devices, that is, nano data centers (nDCs). Our empirical results demonstrate that the proposed scheme reduces the energy consumption and the average number of SLA violations by 21.75 and 11.82 percent, respectively.