Enabling Computation Offloading for Autonomous and Assisted Driving in 5G Networks

Abstract

Connected and automated vehicles currently leverage on-board resources to implement autonomous and assisted driving operations. Such functionalities, which are characterized by tight latency demands, require significant processing resources and can generate a considerable amount of data. Cloud computing is considered the one-stop solution for executing computationally intensive workloads. However, accommodating autonomous and assisted driving requirements using a centralized cloud computing platform is not always feasible due to the latency and reliability constraints they impose. In this paper, we introduce a multi-access edge computing platform suitable for offloading certain autonomous and assisted driving tasks to the edges of the network. We also illustrate how both paradigms (centralized and edge cloud computing) can coexist complementing each other in the challenging task of supporting autonomous and assisted driving, thus opening up new horizons for connected vehicles, for which service instantiation and migration needs to be seamless due to its impact on road safety.