A Multi-Objective Approach Based on Differential Evolution and Deep Learning Algorithms for VANETs

Mohammad Bany Taha, Chamseddine Talhi, Hakima Ould-Slimane, Saed Alrabaee, Kim Kwang Raymond Choo

Research output: Contribution to journalArticlepeer-review

Abstract

Intelligent transportation systems (ITS) are becoming more prominent in our society (for example, in smart cities), although a number of challenges remain to be (fully) addressed (e.g., high vehicle mobility). In this paper, we propose a scheme that combines both a cluster algorithm and a Multi-Objective Task Distribution algorithm based on Differential Evolution (MOTD-DE), designed to ensure stability and reliability in vehicular ad-hoc network (VANET) deployments. Specifically, we use Kubernetes container-base as the cluster algorithm to select various vehicles that fulfill the algorithm's conditions. Hence, this allows us to perform complex tasks on behalf of data owner vehicles. In our approach, the vehicles' information will be available on the master vehicle (data owner vehicle) when the vehicle joins the cluster, and a deep learning model is used to define the fit complexity of sub-tasks. The proposed MOTD-DE distributes sub-tasks between vehicle clusters to reduce the execution time and the resources (vehicles) used to perform a task. We also assume the sub-tasks to be independent. To evaluate our work, we propose scenarios with varying number of tasks, vehicles, CPU and memories values, and distances between cluster vehicles and data owner vehicle. A comparative summary of the evaluation findings between MOTD-DE and four other widely used approaches (i.e., Particle Swarm Optimization (PSO), Genetic Algorithm (GA), Ant-Colony algorithm (ACO), and Artificial-Bee-Colony (ABC) algorithm) shows that MOTD-DE outperforms these competing approaches.

Original languageEnglish
Pages (from-to)1-16
Number of pages16
JournalIEEE Transactions on Vehicular Technology
DOIs
Publication statusAccepted/In press - 2022

Keywords

  • Bee colony
  • Cloud computing
  • Costs
  • Delays
  • Differential Evolution
  • Heuristic algorithms
  • Kubernetes
  • Optimization
  • Particle swarm optimization
  • Task analysis
  • Task distribution
  • VANETs
  • Vehicular ad hoc networks

ASJC Scopus subject areas

  • Automotive Engineering
  • Aerospace Engineering
  • Electrical and Electronic Engineering
  • Applied Mathematics

Fingerprint

Dive into the research topics of 'A Multi-Objective Approach Based on Differential Evolution and Deep Learning Algorithms for VANETs'. Together they form a unique fingerprint.

Cite this