Latency-constrained dynamic computation offloading with energy harvesting IoT devices
In this paper, we address the problem of dynamic computation offloading with Multi-Access Edge Computing (MEC), considering an Internet of Things (IoT) environment where computation requests are continuously generated locally at each device, and are handled through dynamic queue systems. In such context, we consider simple devices (e.g., sensors) with limited battery and energy harvesting capabilities. Hinging on stochastic optimization tools, we devise a dynamic algorithm that jointly optimize radio (e.g., power, energy) and computation (e.g., CPU cycles) resources, while guaranteeing a certain out of service probability (defined as the probability that the sum of local and remote queues exceeds a predefined value) and stability of the device batteries around prescribed operating levels. The method requires the solution of a convex optimization problem per time slot, and does not require apriori knowledge of channel, task and energy arrival distributions. Numerical results illustrate the advantages of the proposed method