Research indexJournal article
Reinforcement Learning Based Optimization Algorithm for Resource Offloading and Load Balancing in Connected Vehicles
Zhao, Duan, Guan
19 Jan 2024
NobleIDark:/48914/w1/20260525/D2F28B1DDOI10.1109/nnice61279.2024.10498913
Reinforcement Learning Based Optimization Algorithm for Resource Offloading and Load Balancing in Connected Vehicles
Zhao, Duan, Guan · 2024 · Subscription
NobleIDark:/48914/w1/20260525/D2F28B1DDOI10.1109/nnice61279.2024.10498913
The full text is at the publisher
You can read the abstract here, or open it at the publisher.
Abstract
Edge computing in Internet of vehicles accomplishes the objectives of low latency and energy consumption by offloading tasks to edge computing servers. However, how to reduce the task offloading latency and energy consumption of vehicle terminals while ensuring the priority of server computing and server load balancing is still a challenge. To solve this problem, this paper first constructs the channel model, communication model, delay model, load balancing model, and target optimization model of the Internet of Vehicles edge computing, and uses non-orthogonal multiple access (NOMA) communication technology. Then the paper proposes an improved Q-learning algorithm based on genetic factors (GAQ-learning algorithm) to calculate the vehicle task offloading scheme. Finally, comparative experiments were conducted between the proposed solution and various schemes. Simulation results demonstrate that the proposed solution can effectively reduce the average offloading latency of users, optimize the workload between servers, and effectively reduce energy consumption, with improved performance compared to three base offloading schemes.