Department of Electrical, Electronics and Systems Engineering, Faculty of Engineering and Built Environment, Universiti Kebangsaan Malaysia, Bangi 43600, Malaysia.
Sensors (Basel). 2022 Jul 24;22(15):5522. doi: 10.3390/s22155522.
The unmanned aerial vehicle (UAV) industry is moving toward beyond visual line of sight (BVLOS) operations to unlock future internet of drones applications, including unmanned environmental monitoring and long-range delivery services. A reliable and ubiquitous mobile communication link plays a vital role in ensuring flight safety. Cellular networks are considered one of the main enablers of BVLOS operations. However, the existing cellular networks are designed and optimized for terrestrial use cases. To investigate the reliability of provided aerial coverage by the terrestrial cellular base stations (BSs), this article proposes six machine learning-based models to predict reference signal received power (RSRP) and reference signal received quality (RSRQ) based on the multiple linear regression, polynomial, and logarithmic methods. In this regard, first, a UAV-to-BS measurement campaign was conducted in a 4G LTE network within a suburban environment. Then, the aerial coverage was statistically analyzed and the prediction methods were developed as a function of distance and elevation angle. The results reveal the capability of terrestrial BSs in providing aerial coverage under some circumstances, which mainly depends on the distance between the UAV and BS and flight height. The performance evaluation shows that the proposed RSRP and RSRQ models achieved RMSE of 4.37 dBm and 2.71 dB for testing samples, respectively.
无人机 (UAV) 行业正在向超视距 (BVLOS) 操作迈进，以解锁未来的无人机应用，包括无人环境监测和远程交付服务。可靠和无处不在的移动通信链路在确保飞行安全方面起着至关重要的作用。蜂窝网络被认为是 BVLOS 操作的主要推动者之一。然而，现有的蜂窝网络是为地面用例设计和优化的。为了研究地面蜂窝基站 (BS) 提供的空中覆盖的可靠性，本文提出了六个基于机器学习的模型，这些模型基于多元线性回归、多项式和对数方法来预测参考信号接收功率 (RSRP) 和参考信号接收质量 (RSRQ)。在这方面，首先在郊区环境中的 4G LTE 网络中进行了 UAV 与 BS 的测量活动。然后，对空中覆盖范围进行了统计分析，并开发了预测方法，作为距离和仰角的函数。结果表明，在某些情况下，地面 BS 有能力提供空中覆盖，这主要取决于无人机和 BS 之间的距离和飞行高度。性能评估表明，所提出的 RSRP 和 RSRQ 模型在测试样本上分别实现了 4.37 dBm 和 2.71 dB 的 RMSE。
