Edge computing powers aerial swarms in sensing, communication, and planning

Abstract

The last decade has witnessed significant progress and extensive applications of unmanned aerial vehicles (UAVs), thanks to their exceptional mobility, cost-effectiveness, and versatile deployment capabilities.1Floreano D. Wood R.J. Science, technology and the future of small autonomous drones.Nature. 2015; 521: 460-466Crossref PubMed Scopus (893) Google Scholar With continuous advances in sensing, communication, and processing technologies, using UAV swarms to provide cost-effective services in future smart cities is expected to become ubiquitous. These services encompass a wide spectrum of applications, ranging from large-scale infrastructure inspection, geospatial land surveying, and traffic/crowd monitoring, to three-dimensional terrain mapping.2Mohamed N. Al-Jaroodi J. Jawhar I. et al.Unmanned aerial vehicles applications in future smart cities.Technol. Forecast. Soc. Change. 2020; 153: 119293Crossref Scopus (191) Google Scholar In these scenarios, UAVs are typically equipped with onboard sensors such as cameras and LiDARs, to gather environmental information by analyzing the sensor data such as captured videos and point clouds. The key technologies to achieve autonomous information-gathering flight for UAV swarms include information extraction from sensor data, transmission and fusion of the extracted information among UAVs, and planning and coordination of their trajectories to ensure safe and efficient mission accomplishment. However, the implementation of these algorithms in real-world UAV swarms poses critical challenges. First, information extraction from sensor data involves computationally intensive tasks, which are difficult to perform onboard UAVs due to their limited processing capabilities. Second, information fusion and trajectory coordination within the swarm heavily rely on inter-UAV communications, which are not consistently available in urban environments due to building occlusion. These challenges impose several limitations on current autonomous UAV swarms, such as reduced flight speed due to low onboard data processing rate. In addition, the need for reliable inter-UAV communications restricts the operational range of the swarm, as UAVs must fly closely to maintain effective communication links, which is challenging in large-scale urban areas. Recently, with rapid developments of cellular networks such as LTE and 5G, utilizing cellular-connected UAVs has become a promising solution in which the UAVs connect with cellular base stations (BSs) as aerial users.3Zeng Y. Wu Q. Zhang R. Accessing from the sky: A tutorial on UAV communications for 5G and beyond.Proc. IEEE. 2019; 107: 2327-2375Crossref Scopus (739) Google Scholar Cellular networks can provide ultra-reliable, low-latency communication services to UAVs over an extensive range.4Gao M. Wan L. Shen R. et al.SparkLink: A short-range wireless communication protocol with ultra-low latency and ultra-high reliability.Innovation. 2023; 4: 100386Google Scholar Moreover, the BSs equipped with edge servers (ESs) can provide powerful computing capabilities at the network edge, offering significant potential to overcome the limitations of current UAV swarms. First, UAVs can offload computationally intensive tasks to the BSs for fast edge processing of the sensing data and retrieve the processed results via communication links. Second, on the other hand, the UAVs can use the obtained results to enhance their awareness of the environment, in particular the wireless link state. This information enables them to adapt their trajectories to keep effective communication with the BSs, thereby accelerating their flight. Furthermore, by deploying multiple BSs that are connected via optical fibers, the information acquired from different UAVs can be efficiently fused on those BSs. Thus, the issues related to non-synchronous information fusion and long-range inter-UAV communication can be solved. With these potentials provided by edge computing, the primary question is how