Joint Resource Allocation for Latency-Constrained Dynamic Computation Offloading with MEC

In this paper, we address the problem of dynamic computation offloading with Multi-Access Edge Computing (MEC), where new requests for computations are continuously generated at each user equipment (UE), and are handled through dynamic queue systems. Building on stochastic optimization tools, we provide a dynamic algorithm that jointly optimize radio (i.e., power, bandwidth) and computation (i.e., CPU cycles) resources, while guaranteeing a target performance in terms of average latency and out of service probability, i.e., the probability that the (sum of) computation queues exceeds a predefined value. The method requires the solution of a convex optimization problem at each time slot, and does not need any apriori knowledge of channel and task arrival distributions. Finally, numerical results corroborate the potential benefits of our strategy.