@article{MF0E1BBA5,
title = "Dynamic Load Balancing Algorithm for Energy-Delay Tradeoff in a Cloud-RSU-Vehicle Architecture",
journal = "The Journal of Korean Institute of Communications and Information Sciences",
year = "2024",
issn = "1226-4717",
doi = "10.7840/kics.2024.49.3.377",
author = "Pyeongjun Choi, Pildo Yoon, Jeongho Kwak",
keywords = "Load balancing, Lyapunov optimization, Edge computing",
abstract = "Advanced autonomous driving services at Level 4 and above eliminate the need for constant driver supervision, enabling vehicles to respond autonomously to various driving scenarios. The integration of components under edge environment such as the On Board Unit (OBU), Road Side Units (RSUs), and cloud infrastructure, facilitated by V2X communication, enhances perception, decision making, and control capabilities by efficiently utilizing networking and computing resources scattered across each component. However, existing standards like LTE-V2X or 5G NR V2X focus on networking resources and lack guidance on efficient computing resource management in autonomous driving systems. To tackle this, we propose a dynamic computing load balancing algorithm for the cloud-RSU-vehicle architecture. The proposed algorithm minimizes the average energy consumption of the vehicle while keeping the average processing delay under a certain level through resource allocation and offloading decisions considering network conditions, computational capabilities, and processing queues using Lyapunov optimization techniques. Via simulations under autonomous vehicle scenario, we show that the proposed algorithm operates adaptively to dynamically changing network speeds and service requests; thereby it significantly saves the average energy consumption with similar average processing delay compared to a policy of only vehicle computing."
}