Data-Driven Beam Management with Angular Domain Information for mmWave UAV Networks
Abstract : Unmanned aerial vehicles (UAVs) have extensive civilian and military applications, but establishing a UAV network providing high data rate communications with low delay is a challenge. Millimeter wave (mmWave), with its high bandwidth nature, can be adopted in the UAV network to achieve high speed data transfer. However, it is difficult to establish and maintain the mmWave communication links due to the mobility of UAVs. In this paper, a beam management scheme utilizing angular domain information (ADI) is proposed to rapidly establish and reliably maintain the communication links for the mmWave UAV network. Firstly, Gaussian process machine learning (GPML)-enabled position prediction is proposed to facilitate coarse-ADI acquisition through the proposed UAV clustering algorithm. Then, with the proposed confined-ADI acquisition which removes the redundancy in the coarse-ADI acquisition, fast beam tracking with respectively the single-beam pattern and the multi-beam pattern is achieved. Finally, a data-driven beam pattern selection scheme is proposed for improving the spectrum efficiency. Simulation results verify the outstanding performance of the proposed beam management for mmWave UAV networks.
? Most existing works on the analysis of mmWave-UAV communication performance adopt the channel models in (8) – (10) or make necessary simplification for the application scenario and analytical requirements.
? When UAVs operate at appropriate altitudes, the probability that an LoS path exists is very high.
? It was shown that the system performance is highly dependent on the flight altitude of the UAV because it impacts both the probability that an LoS path exists and the transmission distance.
? When an LoS path exists between the UAV and BS, the position-aided beam tracking is much more efficient.
? In this paper, we systematically evaluate the key issues and challenges of BM for IRS-assisted mmWave networks to bring insights into the future network design.
? In essence, the environmental awareness is a multiclassification problem with reference to the coordinate of online devices, which can be handled by DNN.
? Those observations indicate that IRS can provide an energy-efficient and low-cost solution to address the blockage issue of mmWave networks.
? The BM for IRS assisted mmWave networks is a very challenging issue, which may require an even more complicated searching procedure to find a near-optimal solution in real time.
• The authors of analyzed the requirements of UAV-aided mmWave communication in 5G ultra-dense networks (UDN), and proposed a novel link-adaptive constellation-division multiple access (CoDMA) mechanism.
• A compact lens antenna allowing mechanical beam steering in the 60 GHz band was proposed in, which can be readily adjusted for HAP applications.
• A frequency-selective channel estimation algorithm was proposed for pure LoS channels and could be extended to the environment with MPCs.
• A low-complexity 3D beam coverage strategy based on sub-array techniques was proposed in , where a UPA is equipped at the UAV to cover the target region.
? Most importantly, there are limited researches on BM issue for the new network topology, while efficient BM is the prerequisite to enable the high-performance mmWave network.
? However, the coexistence of IRSs and mmWave BSs complicates the network architecture, and thus poses great challenges for efficient beam management (BM) that is one critical prerequisite for high performance mmWave networks.
? Although IRS enhances the coverage performance of conventional BS-serve-only mmWave networks, the network communication topology is also complicated at the same time.
? To maintain the mmWave links, extended Kalman filter is utilized for beam tracking, whose performance highly relies on the precise modeling of user motion.
|