Return to Session A6

Session A6: GNSS Integrity and Augmentation

Wide Area Augmentation System Performance Margin Characterization
Nathan Vary, Patricia Morrison, Federal Aviation Administration; Karl Shallberg, Swen Ericson, Zeta Associates Inc.; Eric Altshuler, Zachary Allen, Lucas Laske, Sequoia Research Corporation
Location: Beacon B

The Wide Area Augmentation System (WAAS) will soon reach its 20-year milestone for providing aircraft with navigation services over the entirety of North America. The services WAAS provides to an aircraft are categorized as Non-Precision Approach (NPA) and Approach with Vertical guidance (APV). The user can determine what level of WAAS service is available at the user location by comparing the computed horizontal and vertical protection levels (HPL, VPL) against the horizontal and vertical alert limits (HAL, VAL) defined for each class of service. Based on the alert limits, APV services are further delineated as LNAV/VNAV, LPV and LPV200. For this paper, the services of most interest are LPV and LPV200 where LPV requires the HPL to be within a HAL of 40 m and the VPL to be within a VAL of 50 m, and LPV200 specifies the same horizontal limit but has a lower VAL of 35 m, which is consistent with aircraft precision approach operations.
It is well known from monitoring WAAS performance over its service life that positioning accuracy typically is far better than the computed protection levels would suggest, with the protection levels many times greater than the actual position errors. This margin between the position errors and protection limits stems from the stringent safety-of-life integrity requirements where WAAS ensures that protection levels computed by the user bound the user position error at a probability of 99.99999% per approach operation, equivalent to underbounding only 1 in 10 million approach operations. To achieve such high confidence in the protection levels, the various system threats such as ionospheric disturbances, ground system equipment biases, satellite failures, etc., are accounted for in a conservative way that assures even unobserved threats are mitigated. In this way, the WAAS safety-of-life integrity requirement is always satisfied. However, should a user have non-safety-of-life applications or the ability to mitigate some of these threats on their own, the performance margin normally present in WAAS could be exploited in favor of higher availability of service for the given application or operation. The purpose of this paper is to rigorously characterize this performance margin and qualitatively describe the threats that cause the protection levels to be so inflated.
The approach taken for demonstrating WAAS performance margin consists first of identifying days of interest from past service history that stressed WAAS performance, and second of developing an analysis capability to investigate every possible position solution from the network of WAAS reference stations. The stressed days are expected to show significantly higher position errors than usual and are primarily associated with ionospheric events, which is understandable given the challenges of estimating ionospheric delays and confidence bounds for a grid of points covering the WAAS service area. Some days with interesting ionospheric behavior were provided from monitoring and analysis conducted by WAAS 2nd level engineering at the Mike Monroney Aeronautical Center in Oklahoma City. Other stressed days were taken from daily WAAS and GPS performance monitoring conducted at the FAA William J. Hughes Technical Center (WJHTC). In addition to the 50-day stressed data set identified, a 30-day nominal data set was defined to contrast its level of performance. Therefore, in all, 80 days of data were compiled to conduct this WAAS performance characterization.
The analysis capability was developed to evaluate every possible position solution with its associated protection level against LPV and LPV200 alert limits for all days of interest. Protection levels were computed in the same manner as a user with User Differential Range Error (UDRE) bounds for satellite clock and ephemeris and Grid Ionosphere Vertical Error (GIVE) bounds for ionosphere. The UDRE’s, GIVE’s, range measurements, weighting and geometry matrices were output from the WAAS prototype for satellite subset analysis. This subset analysis computed and saved position errors and normalized errors by their protection levels for every satellite geometry meeting LPV and LPV200 HAL/VAL using 1 Hz data from all 38 WAAS reference stations. For LPV200 alone, this equates to approximately 1.6 trillion position solutions being evaluated.
The results from this position domain analysis, along with a related range domain analysis, were quantified using various statistical measures as well as a Gaussian overbound methodology. The intent of this characterization is to develop and propose a new WAAS performance commitment that users can potentially utilize for non-safety of life applications or where a user augments their solution with additional monitoring capabilities. The paper will also provide a qualitative review of WAAS threats the ground system considers in its integrity analysis to provide some insight into the monitoring a user needs to consider when leveraging this new performance commitment.



Return to Session A6