As civilian use of unmanned aircraft systems (UASs) increases, their safe operation while preventing collisions with either humans or ground structures has become a significant concern. To perform autonomous UAS missions beyond visual line of sight (BVLOS) or in low-altitude airspace safely, achieving high accuracy and reliability of navigation solutions is necessary. This motivates the development of a cost-effective local-area UAS network that utilizes a Local-Area Differential Global Navigation Satellite System (LAD-GNSS) navigation solution. LAD-GNSS achieves a level of integrity comparable to that of Ground-Based Augmentation System (GBAS) Category I operations by monitoring navigation faults at the ground station and by broadcasting integrity information to the UAS. The architecture of this system involves space-conserving hardware configurations and several simplified GBAS integrity monitoring algorithms to reduce both the cost and the complexity of the system. In previous works, only the overall concept of the system had been proposed. Therefore, a full analysis of the system was lacking, and the system architecture had not been implemented. Consequently, in this thesis, we aim to design a comprehensive LAD-GNSS architecture for UASs, and implement it in an actual test bed. Following the implementation, analysis of the system was carried out to assess its performance.
This thesis investigates major issues in fully establishing LAD-GNSS and analyzes their performance. First, we develop a methodology for determining the requirements of BVLOS UAS operation in the low-altitude airspace. We begin by defining the concept and requirements of UAS operation supported by LAD-GNSS for BVLOS UAS operation in the low-altitude airspace, including segregation of UAS operation coverage in low-altitude airspaces, and the alert limits (ALs) for each operational coverage. The required safety levels of UAS operation are analyzed on the basis of the safety risk assessment, and then we determine the ALs that meet the required safety level. Second, we design a LAD-GNSS architecture for UAS by simplifying the hardware and monitoring the algorithms of both the ground facility and the onboard module, using the well-established GBAS as a starting point. We perform theoretical performance evaluations comparing position uncertainty bounds, which are represented by protection levels (PLs), to the corresponding navigation requirements for each coverage. We compare PLs and ALs under both nominal and malfunction cases. Third, we develop methodologies for improving the performance of LAD-GNSS for UASs. To reduce PLs under the nominal case, we first derive the PLs for excessive acceleration and code-carrier divergence fault scenarios which are bounded by using a maximum-allowable error in range. Using these PLs, integrity and continuity allocations are ideally and dynamically assigned to each single-fault hypothesis to obtain optimized PLs that are identical for all fault scenarios. We find that vertical protection levels (VPLs) are reduced by approximately 11% when implementing the optimal allocation method. In addition, we develop and apply the ionospheric threat model for the Korean region to reduce the PLs in the malfunction case. The simulation results show that the VPLs in the malfunction case are reduced by approximately 50% when the Korea threat model is applied. Finally, we explain the development of a prototype of LAD-GNSS and the performance evaluation for both accuracy and integrity performance. The LAD-GNSS prototype is composed of two parts: a ground module and an onboard module. The ground module operates in a manner similar to a GBAS ground facility. Most of the computations regarding integrity monitoring are performed in the ground module. The onboard module computes position solutions using the corrections broadcast from the ground module. The position solutions are then fed into a flight controller. Flight tests are conducted to evaluate the performance of this LAD-GNSS prototype. The results obtained from autonomous landing tests show that the 95% accuracy of the LAD-GNSS landing is 1.70 m. In the path-following test, the results show that the 95% vertical position error was less than 1 m and that the VPLs are much lower than the VAL.
Finally, we propose a methodology for estimating safe separation distances between UASs sharing the same source of guidance and local-area differential corrections. Only the uncorrelated component of the position error between UASs in the same network contributes to potential separation violations. Therefore, to estimate the separation distances, models of uncorrelated navigation system errors and flight technical errors (FTEs) are developed both theoretically and experimentally. Generalized models of ionospheric and tropospheric errors are developed theoretically. Airborne pseudorange errors and FTE components, which depend on the hardware, the environment, and operational conditions, are determined though UAS flight tests. By applying risks of separation violations derived from the safety risk assessment methodology to the proposed error models, the safe separation distances were estimated. In the simulation results for a 24-satellite Global Positioning System (GPS) constellation, the horizontal separations between UASs in the same network were mostly below 3.5 m, while the vertical separations ranged between 3.5 and 6.5 m.