Abstract:
Unmanned aerial vehicles (UAVs), due to their flexible deployment, are used as a 3-D space assistant tool for vehicular edge computing (VEC) to cover moving vehicles. How...Show MoreMetadata
Abstract:
Unmanned aerial vehicles (UAVs), due to their flexible deployment, are used as a 3-D space assistant tool for vehicular edge computing (VEC) to cover moving vehicles. However, existing work generally assumes relatively uniform vehicle distribution while the actual road conditions vary over time. The time-varying location of vehicles and road congestion in peak hours pose challenges to VEC. First, traffic congestion can lead to imbalanced UAVs load, that is UAVs covering congested areas are overloaded while others remain idle. Second, after the high-speed moving vehicle leaves the service range of the current UAV, it is unable to receive computing results of the original request task, which means task processing failure. In this work, we propose a framework of UAV clusters adaptive deployment (UCAD) to address these issues. By clustering vehicles, UCAD gives a density-based adaptive region determination (DBARD) algorithm to determine congested areas and dynamically update them to adapt to dynamic network environments. After that, UCAD presents a particle swarm optimization-based Cluster deployment algorithm (PSOC), deploying UAV clusters within determined areas to provide continuous services for vehicles. Simulation results demonstrate that UCAD can adaptively deploy UAV clusters to assist vehicles based on traffic congestion conditions. Simulation results show that compared to existing works, CONEC, TU2V, and IELTS, the proposed UCAD can increase the task success rate by 8.7%, 18.5%, and 23.5%, respectively. UCAD can achieve better performance on UAVs workload balance.
Published in: IEEE Internet of Things Journal ( Volume: 11, Issue: 13, 01 July 2024)