Adaptive Task Offloading Auction for Industrial CPS in Mobile Edge Computing

The emerging intelligent applications in Industrial Cyber-Physical Systems (ICPS), such as product inspection by deep-learning-based image recognition technology, are highly computation-consuming. However, the smart devices without sufficient computing resources fail to handle this kind of applications. Moreover, the Internet has very high latency compared with the local network which fails to meet the requirements of time-sensitive tasks, therefore we can not offload these tasks over the cloud. Mobile Edge Computing (MEC) brings the opportunities to offload the tasks of ICPS to the MEC servers to satisfy strict latency requirements, as well as to meet the demand for security requirements. Considering MEC servers owned by the third parties, resource allocation in MEC should be solved jointly with network economics to maximize the utility of system. In this paper, we investigate the task offloading problem under the access capability, latency and security constraints. Specifically, we present a novel Adaptive Task Offloading (ATO) auction mechanism to determine which MEC server to offload with access capability and security constraints, and how to schedule tasks with various deadline constraints, which incentives the third party of MEC providers to share their computing resources with the maximum profit. According to our theoretical analysis, the proposed auction mechanism has the properties of individual rationality, computational efficiency and truthfulness. Extensive simulations have been conducted to evaluate the performance of ATO auction and the experimental results show our method provides better solutions with the classic greedy algorithms in terms of maximizing the utility of the MEC server.