Loading...

A Machine Learning Simulator for Task Offloading and Resource Allocation in IoV

Alaei Tabatabaei, Ali | 2024

3 Viewed
  1. Type of Document: M.Sc. Thesis
  2. Language: Farsi
  3. Document No: 56885 (19)
  4. University: Sharif University of Technology
  5. Department: Computer Engineering
  6. Advisor(s): Bayat Sarmadi, Siavash
  7. Abstract:
  8. Recent years have witnessed significant advancements in the smart transportation industry. Timely execution of processes and tasks prior to deadlines is crucial and sensitive. With the advancement of the automotive industry and smart transportation vehicles, the number of smart vehicles is increasing day by day. However, not all devices can perform their respective tasks on time, hence the need for edge network servers to assist in task execution .Decision-making regarding task execution methods and resource allocation in mobile networks poses a significant challenge. In recent research, given the complexity of the issue and the advancements in machine learning algorithms, the efficacy of reinforcement learning methods in addressing this problem has been demonstrated. However, thus far, no systematic simulation tool has been provided for modeling this problem, presenting a challenge in effectively solving it in real-world dimensions. In this research, a modular and flexible simulation system is proposed, capable of accurately modeling the problem environment. This simulation system is developed using the Python programming language and is compatible with the GYM API. The proposed simulator can accurately calculate the delay and energy consumption of task transmissions by modeling communication channels according to 3GPP standards. In this simulator, the reward parameter, as a single objective of the optimization problem, is calculated based on energy consumption and task execution time. The proposed simulator can be used for training reinforcement learning models or even comparing different algorithms to solve this problem. In the final step, the performance of various reinforcement learning algorithms is examined in the implemented environment to further investigate the simulator's efficacy
  9. Keywords:
  10. Task Offloading ; Low Latency ; Resources Allocation ; Reinforcement Learning ; Mobile Edge Computing ; Smart Vehicles ; Vehicles Internet ; Simulator System

 Digital Object List

 Bookmark

...see more