The rapid increase in the number of commercial drones will trigger a transformation in next-generation wireless networks due to widespread mobility in three dimensions (3D). This 3D mobility can be used to improve the agility and the mobility of the network. Aerial network elements (ANEs) including not only drones and unnamed aerial vehicles/systems (UAVs/UASs), but also helikites, electric vertical take-off and landing (eVTOL) aircraft, and high altitude platform station (HAPS) systems are expected to serve as the main enabler of the 2D to 3D wireless network paradigm change. However, their presence brings new challenges as ANEs need to harmoniously coexist with the existing terrestrial flexible-use spectrum networks. For instance, ANEs may be served directly by the current terrestrial LTE networks in the U.S., which introduces spectrum coexistence problems between ANEs and terrestrial ground users. To tackle these concerns, this project studies the integration of ANEs into the existing terrestrial networks by developing new models, algorithms, and experiments. This groundbreaking research will contribute towards transforming wireless systems from 2D to 3D, thereby improving the agility and the mobility of the network and enhancing 6G systems. The US-Canada collaboration will foster an international transfer of expertise across the aforementioned areas, thus ensuring broad societal and technological impacts.
This joint US-Canada research project presents a cutting-edge solution of harmonious wireless coexistence between ANEs and the existing terrestrial networks by utilizing machine learning, edge computing, and optimization technologies to truly unleash the high potential of ANEs via: (i) designing a novel framework that uses the 3D ANE mobility aspects to improve spectrum sensing and detection performance; (ii) proposing an AI-based approach that can detect and classify signals in the band of interest with limited side information, recognize threats, and then deploy localization and tracking approaches that can adapt to specific scenarios and contexts; (iii) developing a reinforcement learning (RL) based framework to jointly optimize ANE trajectory design and resource management to maximize the performance of 3D networks and distributed AI algorithms. The developed models and methods will be tested in the NSF AERPAW Platform at North Carolina State University (NCSU), providing an opportunity for their validation and testing in a real-world environment.