Low-AoI data collection for multi-UAVs-UGVs assisted large-scale IoT systems based on workload balancing

Abstract

Unmanned Ground Vehicles (UGVs), due to their mobility and high power capacity, can serve as mobile base stations to assist Internet of Things (IoT) systems in remote areas lacking infrastructure for data collection. However, the slow speed of UGVs leads to significant transmission latency in large-scale IoT systems. Unmanned Aerial Vehicles (UAVs) offer advantages in terms of rapid mobility and flexibility. By deploying UAVs carried by UGVs to collaboratively perform data collection tasks, we can effectively enhance the performance of data collection. We refer to this system as an integrated UGV-UAV-assisted IoT system. In this system, multiple UGVs and UAVs are deployed to collect data from sensor nodes (SNs) over large areas. It is essential to consider the task regions assigned to each UGV and the UAVs they carry. UGVs equipped with more UAVs should handle data collection tasks for a greater number of SNs. To address this issue, we propose a low-latency data collection scheme for multi-UGVs-UAVs based on workload balancing (LMUWB). This scheme allocates appropriate task regions to each UGV based on the deployment locations of ground SNs and assigns an adequate number of UAVs according to the workload of each region. Additionally, deep reinforcement learning (DRL) is introduced to optimize the trajectories of UGVs and UAVs, enabling to reduce the system Age of Information (AoI), so as to ensure data freshness. Simulation experiments demonstrate that the LMUWB scheme can provide an effective solution for timely data collection in large-scale IoT systems.