Edge computing assisted privacy-preserving data computation for IoT devices

Abstract Along with the ubiquitous deployment of IoT devices, requirements on sensing data computation and analysis increase rapidly. However, the traditional cloud-based architecture is no longer sustained the computation load from these tremendous IoT devices, which bring the paradox of delay tolerance and bandwidth insufficiency. Fortunately, the edge computing is emerged and incorporated with the IoT network. Meanwhile, new questions arises. When and how to select among edge computing servers, and also achieve a well balance between consumed energy, transmission delay and data privacy. In this paper, we consider the problem that how IoT devices allocate their computation loads among edge computing servers and their on-chip computation units, to balance energy efficiency and data privacy in physical layer. Firstly, the optimization function of IoT devices is derived which reflects the energy consumption, transmission delay and also privacy requirement; Secondly, the direct transmission scenario is analyzed, and optimal transmit power are derived with or without privacy factors; Thirdly, we extend the model to relay transmission scenario when edge computing servers are far away, and propose the relay selection algorithm for IoT devices; Finally, by extensive simulations, two main conclusions are verified: the energy consumption remains the same with data privacy protection, energy saved 54.9% on average using relay IoT devices compared to direct transmission case.