With the widespread use of Internet of Things (IoT) applications, the fast response and efficient data storage have been the main concerns of the service users and providers. Thus, data offloading has become a hotspot in both industry and academia, especially for real-time applications. To achieve efficient data offloading, a great number of in-depth studies have been conducted. Nevertheless, when addressing the issue of data offloading, few studies have taken into account the unstable channel conditions, which is however more practical and really needs more attention. In this paper, we consider the unstable channel state in the communication model. Based on this, we propose the task reliability model, the energy consumption model, and the device reliability model. From the perspective of optimizing energy consumption, we propose an optimal task scheduling model. Moreover, an innovative Dynamic Energy-Efficient Data offloading scheduling algorithm-DEED is proposed. The purpose of DEED is to as much as possibly reduce the energy consumption while ensuring the task reliability. To verify the effectiveness of the proposed DEED, extensive experiments are conducted to compare it with three comparison algorithms: DRSD, DEPD, and DRPD. The experimental results under different channel conditions demonstrate the superiority of the DEED in terms of the energy saving, reliability, and robustness
The development of Internet of Things (IoT) has been hailed as an unprecedented success. In the near future, tens of billions of IoT devices will be applied in homes, schools, companies, hospitals, etc. However, the processing capacity of IoT devices cannot totally guarantee the completion of tasks on time. Consequently, offloading tasks to the network edges and processing them in the edges has become a mainstream paradigm . Therefore, Edge Computing (EC) , Mobile Edge Computing (MEC) , Mobile Cloud Computing (MCC) , Fog Computing (FC)  and other similar concepts have been proposed in recent years. To harvest the computing and storage resource of devices at the edge environment, both academia and industry have focused on the collaboration between edge network and IoT devices [27, 28]. Thus, data offloading becomes a critical technology for IoT applications, especially for applications running on mobile devices. Nowadays, data are regarded as one of the most promising resources, and lots of artificial intelligence systems need as many data as possible to improve its performance. Due to the lack of reliability and security of storing data on mobile devices, offloading data to the edge or data center becomes an important way to permanently store data. In IoT applications, collaborative data offloading still faces many challenges. With the increase of application scenarios, IoT devices are going to be expected to perform more and more sophisticated tasks such as surveillance, crowdsensing, and health monitoring. However, the battery capacity of IoT devices are limited, and recharging or replacing its battery frequently is impractical in most instances. Besides, for mobile IoT devices, they are often used in the network where communication quality dynamically fluctuates, so data loss or data offloading failure is inevitable. As a service pattern, the success rate and response speed of data offloading directly affect the Quality of Service (QoS).