Dual-Band Non-Stationary Channel Modeling for the Air-Ground Channel

Multiple air-to-ground (AG) radio propagation channels are experimentally characterized for two frequency bands, C-band and L-band. These characterizations are aimed to support the specification of the control and non-payload communication (CNPC) links being designed for civil unmanned aircraft systems (UAS). The use of UAS is expected to grow dramatically in the coming decades. In the United States, UAS will be monitored and guided in their operation within the national airspace system (NAS) via the CNPC link. The specifications of the CNPC link are being designed by government, industries, academia and standards bodies such as the Radio Technical Commission for Aeronautics (RTCA). Two bands have been allocated for the CNPC applications: from 5030 to 5091 MHz in C-band and a portion of the aeronautical L-band from 960 to 1215 MHz. The project under which this work was conducted is entitled “Unmanned Aircraft Systems Research: The AG Channel, Robust Waveforms, and Aeronautical Network Simulations”, and this is a sub-project of a NASA project entitled “Unmanned Aircraft Systems Integration in the National Airspace System.” Measurements and modeling for radio propagation channels play an essential role in wireless communication system design and performance evaluation; such models estimate attenuation, delay dispersion, and antenna diversity in wireless channels. The AG channel differs significantly from classic cellular, ground-to-satellite, and other terrestrial wireless channels, particularly in terms of antenna heights and velocity. The previous studies about the AG channels are reviewed and the significant gaps are indicated. NASA Glenn Research Center has conducted an AG channel measurement campaign for multiple ground station local environments, including over sea, over freshwater, desert, suburban, near urban, hilly and mountainous settings. In this campaign, over 316 million power delay profiles (PDPs) or channel impulse responses (CIRs), over 82 flight tracks, have been collected. The measurement equipment was a dual-band single-input multiple-output (SIMO) wideband channel sounding system with bandwidth of 50 MHz in C-band and 5 MHz in L-band. Given the dynamic nature of the AG environments, the channels are statistically non-stationary, meaning that the channel’s statistical parameters change over time/space. We have estimated, via two distinct methods, that the stationarity distance is approximately 15 m—this is the distance over which the channel characteristics can be assumed to be wide sense stationary. The AG channel attenuation is considered as a combination of large scale path loss, small scale fading, and airframe shadowing. The large scale path loss is modeled by both the log-distance model and two-ray models. The theoretical flat earth and curved earth two-ray models are presented, along with their limitations, boundaries and some enhancements. Numerous propagation effects in the AG channels are discussed, and this includes earth spherical divergence, atmospheric refraction, atmospheric gas and hydrometeor attenuations, and ducting. The small scale fading is described by the Ricean distribution, which for unit-energy normalizations are completely characterized by Ricean K-factors; these K-factors are approximately 28.7 dB in C-band and 13.1 dB in L-band. The line-of-sight (LOS) signal can be obstructed by the airplane itself in some specific maneuvers, and this is termed airframe shadowing. For the specific flights and NASA aircraft used in our measurements, the shadowing duration was found to be on average 30 seconds, and the shadowing loss can be as large as 40 dB. The statistics, models and simulation algorithm for the airframe shadowing are provided. The wideband characteristics of the AG channel are quantified using root-mean-square delay spread (RMS-DS), and illustrated by sequences of PDPs. Tapped delay line (TDL) models are also provided. Doppler effects for over water channels are also addressed. Given the sparsity of the diffuse multipath components (MPCs) in the AG channels and generally short lifetime of these MPCs, the CIRs are modeled by the two-ray model plus intermittent 3rd, 4th or 5th “rays.” Models for intermittent ray probability of occurrence, duration, relative power, phase, and excess delay are provided. The channels at C-band and L-band were found to be essentially uncorrelated; this conclusion holds for the specific antenna locations used in our experiments (the aircraft underside), but is not expected to change for arbitrary antenna locations. For the aircraft antenna locations employed, intra-band signals are highly correlated, and this is as expected for channels with a dominant LOS component; analytical correlation computations show interesting two-ray effects that also appear in measurements. Multiple aircraft antennas and carefully selected locations are recommended for mitigating airframe shadowing for the CNPC link. Future work for AG channel modeling includes characterization of L-band delay dispersion and L-band TDL models, estimation of building and/or tree shadowing for small UAS that fly at very low altitudes, evaluation of multiple ground site(s) antenna diversity, and AG channel modeling via geometric techniques, e.g., ray-tracing