Title: Dependent task offloading based on proactive replication

Authors: Benhong Zhang; Cong He; Hao Xu; Xiang Bi

Addresses: School of Computer Science and Information Engineering, Hefei University of Technology, Hefei, 230009, China; Engineering Research Centre of Safety Critical Industrial Measurement and Control Technology, Ministry of Education, Hefei, 230009, China ' School of Computer Science and Information Engineering, Hefei University of Technology, Hefei, 230009, China; Engineering Research Centre of Safety Critical Industrial Measurement and Control Technology, Ministry of Education, Hefei, 230009, China ' School of Computer Science and Information Engineering, Hefei University of Technology, Hefei, 230009, China; Engineering Research Centre of Safety Critical Industrial Measurement and Control Technology, Ministry of Education, Hefei, 230009, China ' School of Computer Science and Information Engineering, Hefei University of Technology, Hefei, 230009, China; Engineering Research Centre of Safety Critical Industrial Measurement and Control Technology, Ministry of Education, Hefei, 230009, China

Abstract: Vehicle-to-vehicle (V2V) offloading is considered a promising solution to the problem of limited computing resources in vehicles. In practical applications, many computational tasks can be divided into interdependent subtasks. Existing studies rarely consider the dependency between subtasks and the prospect of subtask offloading failures. In this paper, we propose a DDPG-based proactive replication offloading algorithm of dependent tasks. To improve the success rate of offloading, interdependent subtasks are first categorised into critical subtasks and non-critical subtasks according to the degree of urgency, and critical subtasks will be offloaded to two service vehicles simultaneously. Then, the computational task offloading problem is considered as a linear integer programming problem and the optimal policy for task offloading is obtained using the DDPG algorithm. Simulation results show that this scheme performs well in improving the success rate of offloading and reducing delays.

Keywords: critical subtask; proactive replication offloading; vehicle-to-vehicle; V2V; dependent task; offloading reliability.

DOI: 10.1504/IJSNET.2024.138919

International Journal of Sensor Networks, 2024 Vol.45 No.2, pp.87 - 100

Received: 29 Nov 2023
Accepted: 03 Apr 2024
Published online: 03 Jun 2024 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article