4.7 Article

Vehicle-Based Cloudlet Relaying for Mobile Computation Offloading

Journal

IEEE TRANSACTIONS ON VEHICULAR TECHNOLOGY
Volume 67, Issue 11, Pages 11181-11191

Publisher

IEEE-INST ELECTRICAL ELECTRONICS ENGINEERS INC
DOI: 10.1109/TVT.2018.2870392

Keywords

Mobile computation offloading; vehicular networks; cloudlet relaying

Funding

  1. National Natural Science Foundation of China [61603031, 61501023]
  2. State Key Lab of Rail Traffic Control and Safety [RCS2018ZT015]

Ask authors/readers for more resources

This paper studies computation offloading for mobile users in legacy vehicles that are not equipped with powerful computing devices. While running a time-sensitive and computation-intensive application on a mobile device (MD) of these users, the MD may offload part or all the applications to a nearby smart vehicle that can serve as a cloudlet. This provides a possibility to complete the application in time while saving the battery energy of the MD. However, high mobility of the vehicles causes short-lived communication links between the vehicles. As a result, the MD may have to offload tasks to a new cloudlet each time it lases the connection to the current cloudlet. In this paper, we propose a vehicle-based cloudlet relaying (VCR) scheme for mobile computation offloading. The objective is to effectively utilize the computation resources available in surrounding smart vehicles of the MD in the highly dynamic network environment, where each vehicle-based cloudlet may only be available to help the MD execute a small portion of the application. Instead of offloading tasks directly from the MD to individual cloudlets, which may consume high power and a long time, especially when the link just becomes available or is about to disconnect, we propose computation relaying that allows the current executing cloudlet to offload tasks to the next executing cloudlet on behalf of the MD. A set of conditions are established to enable the inter-cloudlet offloading so that the computation relaying can be achieved between the vehicle cloudlets, while the results can be returned to the MD. In addition, the computation relaying is always backed up by direct offloading between the MD and individual cloudlets. An algorithm is proposed to find the time to switch to a new cloudlet and select the best cloudlet as the next execution site or return the execution back to the MD, in order to minimize the energy consumption of the MD while satisfying the delay constraint of the application. Extensive simulation results are shown to demonstrate the effectiveness of the proposed VCR scheme.

Authors

I am an author on this paper
Click your name to claim this paper and add it to your profile.

Reviews

Primary Rating

4.7
Not enough ratings

Secondary Ratings

Novelty
-
Significance
-
Scientific rigor
-
Rate this paper

Recommended

No Data Available
No Data Available