A Survey on Deep Reinforcement Learning-driven Task Offloading in Aerial Access Networks

Abstract

Internet of Things computation offloading is a challenging problem, particularly in distant places where mobile edge computing (MEC) or cloud infrastructure is absent. Fortunately, aerial access networks (AANs), which include unmanned aerial vehicles and satellite communications, are employed as effective aerial platforms to deliver ubiquitous and reliable access. Furthermore, deep reinforcement learning (DRL) is a viable method to boost the efficiency of edge network resource management in achieving energy-efficient, low-delay MEC services. This paper investigates recent advances in DRL-based task offloading strategies in the MEC-based AANs. Research challenges and directions are also discussed.

Topics & Concepts