Abstract
Computation task offloading is one of the enabling technologies for
computation-intensive applications and edge intelligence, which experiences the
explosive growth of massive data generated. Different techniques, wireless
technologies and mechanisms have been proposed in the literature for task
offloading in order to improve the services provided to the users. Although
there is a rich literature of computation task offloading, the role of data in
the scope of it has not received much attention yet. This motivates us to
propose a survey which classified the state-of-the-art (SoTA) of computation
task offloading from the view point of data. First, a thorough literature
review is conducted to reveal the SoTA from various aspects with the
consideration of task generation, i.e., architecture, objective, offloading
strategy, task types, etc. It is found that types of task offloading is related
to the data and will affect the offloading procedure, which contains resource
allocation, task allocation etc. Then computation offloading is classified into
two categories based on task types, namely static task based offloading and
dynamic task based offloading. Finally, our views on future computation
offloading are provided with the corresponding challenges and opportunities.