1. Avoiding Improper Modeling in SBAS Ionospheric Correction
ION ITM 2018
Reston, VA
Jan. 29-Feb. 1, 2018
Avoiding Improper Modeling
in SBAS Ionospheric Correction
with Shrunk Observations
Takeyasu Sakai
Electronic Navigation Research Institute
National Institute of Maritime, Port and Aviation Technology, Japan

2. Introduction Japanese SBAS: The MSAS program
ION ITM Jan Slide 1
Introduction
Japanese SBAS: The MSAS program
MSAS, Japanese version of WAAS, has been operational since Sept
Still serving horizontal navigation only.
Transition to the new system in 2020, as a part of QZSS program; Planning update for vertical guidance in 2023 with two GEOs.
The ionospheric effect is a major error source for Single-Frequency SBAS:
The ionospheric term is the dominant factor of degradation of availability.
Necessary to develop improved ionosphere algorithms to introduce vertical guidance capability to MSAS.
Proposal: Reduction of false alerts of the current ionosphere algorithm.
The storm detector makes a lot of false trips for the low-latitude ionosphere.
Set the maximum uncertainty to the estimation even there is no storm.
It is possible to improve system availability by reducing such false trips.
Basic idea: Reduce observations for fit to keep the storm detector quiet.

3. Current Status MSAS: Japanese SBAS in operation.
ION ITM Jan Slide 2
Current Status
MSAS: Japanese SBAS in operation.
MTSAT Satellite-based Augmentation System.
Operational since Sept. 27, 2007.
Continues operation with 2 signals via 1 GEO.
MTSAT-1R decommissioned in Dec (MTSAT-2 also in 2020).
Hawaii and Australia MRS sites decommissioned in Feb
Service for Air Navigation
Providing GPS augmentation information.
Available for RNAV, from En-route through NPA (Non Precision Approach), i.e. RNP 0.3.
Within Fukuoka FIR.
Only horizontal navigation due to ionosphere activities.
NOTAM is available to MSAS users.
Alert for service interruption.
Alert for predicted service outage.

4. Current Configuration
ION ITM Jan Slide 3
Current Configuration
Ranging Signals
GPS Satellites
Ground
Network
Naha GMS
Fukuoka GMS
Tokyo
GMS
Sapporo
Hitachi-Ota MCS
(and GMS)
Kobe MCS
Users
MTSAT-2
Augmentation Signals
6 GMS
in Japan
PRN137
PRN129
MSAS Monitor Stations
1 GEO, 2 MCS (Master Control Station), and 4 GMS (Ground Monitor Station);
MCS also has GMS function; MSAS has 6 domestic monitor stations.
MTSAT-2 is broadcasting 2 signals from 2 MCS (PRN129 and PRN137).

5. Performance of MSAS GPS only GPS only MSAS PRN129 MSAS PRN137
ION ITM Jan Slide 4
Performance of MSAS
GPS only
GPS only
Horizontal
0.722m RMS
Horizontal
0.717m RMS
MSAS PRN129
MSAS PRN137
GEONET (Takayama) 2016/8/8-12 (5 days)
PRN129 and PRN137 Broadcast Signal

6. Continuous Operation Replacement in 2020: MSAS V2
ION ITM Jan Slide 5
Continuous Operation
Replacement in 2020: MSAS V2
MSAS continues operation with 1 GEO (MTSAT-2) and 6 GMS until 2020.
In 2020, MSAS will continue operation with a new GEO of the QZSS.
QZSS (Quasi-Zenith Satellite System): Japanese regional satellite navigation system with IGSO and GEO satellites.
The L1Sb signal of QZS-3 (GEO) will be used for MSAS service.
MCS equipment will also be fully replaced at the same time.
7 GMS will be added: Totally 13 GMS domestic.
The performance will be similar with the current MSAS: Horizontal only.
Supporting vertical guidance in 2023: MSAS V3
Vertical guidance: LPV and LPV-200 operation.
Needs software upgrade:
Adding GMS cannot overcome ionospheric effects.
Needs development of the improved algorithms for ionospheric correction.
Will be supported in accordance with introduction of the 2nd GEO in 2023.

7. Toward Vertical Guidance
ION ITM Jan Slide 6
Toward Vertical Guidance
The current MSAS is built on the IOC WAAS:
Achieves 100% availability for Enroute to NPA flight modes.
The primary purpose is providing horizontal navigation means to aviation users.
No vertical guidance; Ionospheric corrections may not be used.
As the first satellite navigation system developed in Japan, the design tends to be conservative, especially for ionospheric corrections..
The major concern for vertical guidance is ionosphere:
The ionospheric term (uncertainty of estimation) is dominant factor of degradation of system availability.
Necessary to reduce ionospheric terms to provide vertical guidance with reasonable availability.

8. Ionosphere Estimation: Planar Fit
ION ITM Jan Slide 7
Ionosphere Estimation: Planar Fit
Cutoff Radius
Vertical Delay
Fit Plane
IPP
IGP
Fitting Radius, Rfit
Developed for WAAS; MSAS employs the same algorithm.
Assumes ionospheric vertical delay can be modeled as a plane.
Model parameters are estimated by the least square fit.
Output:
Vertical delay estimation at the IGP.
GIVE (Grid Ionosphere Vertical Error): Uncertainty of the estimation.
GIVE Equation
Formal Sigma
Spatial Threat
Temporal Threat

9. Storm Detector ^ c2 = ( Iv,IPP -Iv,IPP )T W ( Iv,IPP -Iv,IPP )
ION ITM Jan Slide 8
Storm Detector
The current storm detector algorithm is ‘Chi-Square Test’:
Chi-square statistics is the sum of squared residuals:
If Chi-square < Threshold (Storm Detector does not trip),
The ionosphere model is valid to use.
GIVE is computed from the GIVE equation.
If Chi-square > Threshold (Storm Detector trips),
The ionosphere model is judged not valid.
The assumption that the distribution of residual errors is the normal (Gaussian) may not be valid.
In such a case, the associate IGP is determined as storm condition.
Set the maximum GIVE (Storm Detector ‘trips’).
This action leads larger protection levels degrading availability of the system.
False trips should be reduced.
c2 = ( Iv,IPP -Iv,IPP )T W ( Iv,IPP -Iv,IPP ) ^

10. Example Detector Response
ION ITM Jan Slide 9
Example Detector Response
Severe storm condition with the largest Kp index of 9-.
Chi-square metric is chi-square statistics divided by the associate threshold.
Inflation factor Rirreg applied.
Normalized residual means residual error divided by sGIVE; It must be within 5.33 for integrity.
We can observe many false trips; These trips lower the availability of MSAS.
True Trips
5.33
False Trips 1
No Trips
Storm Detector Trips

11. Possible Situation Fit with N IPPs Storm Detector trips due to anomaly
ION ITM Jan Slide 10
Possible Situation
Fit with N IPPs
Storm Detector trips due to anomaly
Rfit
Anomaly
IPP
for fit
not for fit
Fit with N-DN IPPs
Storm Detector does not trip
Anomaly
IPP
for fit
not for fit
Rfit
Reduce IPPs for Fit
Normally, larger Rfit is preferable to have enough N, the number of IPPs.
Due to curvature of the real ionosphere, large Rfit (means large N) sometimes reaches an anomalous region where ionosphere is not on the planar model.
In such a condition, storm detector trips and sets GIVE to the maximum.
Basic Idea: How about decreasing N to avoid observations within anomaly ?

12. ION ITM Jan Slide 11
Shrink Observations
Storm Avoidance Algorithm: If Storm Detector trips, shrink observations by:
Radius Method: Decrease Rfit until Storm Detector does not trip.
Southernmost Method: Invalidate some IPPs at the southernmost.
Note: Keeps spatial threat not reduced.
For conservatism, employs the spatial threat model once computed before these adjustment of Rfit.
This means the adjustment of Rfit affects to fitting and Storm Detector only and does not reduce the spatial threat model.
(1) Decrease Rfit while Storm Detector trips
(2) Stop if DRfit = DRmax, DN=DNmax, or N = Nmin
Invalidate the southernmost IPP while Storm Detector trips
(2) Stop if DRfit = DRmax, DN=DNmax, or N = Nmin

13. Simulation for MSAS Simulates MSAS V3
ION ITM Jan Slide 12
Simulation for MSAS
Simulates MSAS V3
SBAS simulator developed by the ENRI, with some modifications for proposed algorithm.
13 GMS locations after 2020 (Red Triangles).
Test periods
For each period, check:
GIVEI distribution in the message stream
Availability of LPV flight mode
Period
Max Kp
Remark
(1) 2011/5/16 4-
Moderate Storm
(2) 2011/11/19 3
Quiet
Note: GIVE Filter and IPP Filter do not apply.

14. Histogram for GIVEI in the Message Stream
ION ITM Jan Slide 13
GIVEI: Radius Method
Histogram for GIVEI in the Message Stream
GIVEI Reduction
GIVEI=
Applies Storm Avoidance with Radius Method for Period (2).
Replaces GIVEI of 14 conditions with lower GIVEI.
Improves navigation availability with lower protection levels.

15. GIVEI: Southernmost Method
ION ITM Jan Slide 14
GIVEI: Southernmost Method
Histogram for GIVEI in the Message Stream
GIVEI Reduction
GIVEI=
Applies Storm Avoidance with Southernmost Method for Period (2).
Replaces GIVEI of 14 conditions with lower GIVEI for more IGPs.
Improves navigation availability with lower protection levels.

16. LPV Availability: Period (1)
ION ITM Jan Slide 15
LPV Availability: Period (1)
95%
99.9%
Baseline MSAS
Southernmost Method
Availability of LPV flight mode with HAL=40m and VAL=50m.
Proposed method improves availability to reasonable level.
Other additional algorithms can be applied for more availability improvement.

17. LPV Availability: Period (2)
ION ITM Jan Slide 16
LPV Availability: Period (2)
80%
99%
Baseline MSAS
Southernmost Method
Availability of LPV flight mode with HAL=40m and VAL=50m.
Proposed method improves availability to reasonable level.
Other additional algorithms can be applied for more availability improvement.

18. Conclusion MSAS will have a new GEO and new ground facilities in 2020.
ION ITM Jan Slide 17
Conclusion
MSAS will have a new GEO and new ground facilities in 2020.
Alters the current GEO and ground facilities: MSAS V2.
Update for MSAS V3 with vertical guidance capability is also planned in with the 2nd GEO.
Development of ionospheric algorithms for MSAS vertical guidance:
Storm Avoidance Algorithm: Decrease Rfit or N to keep the storm detector quiet.
With this algorithm, Invalidating southernmost IPPs is much better method.
The simulation confirmed improvement of system availability.
Further activities:
Verification with more historical storm data archives.
Testing combined use with other algorithms.
Contact for more information:
Dr. Takeyasu Sakai
Electronic Navigation Research Institute
National Institute of Maritime, Port and Aviation Technology, Japan