Describe how you would perform an outage analysis to determine how long the integrated navigation solution remains within a given position error bound during GNSS loss.

• A. Model the holdover performance of the INS with process noise, propagate the error covariance over time, and compute the time until the error bound is exceeded, then assess against requirements.
• B. Use a static estimation with no dynamics.
• C. Wait for GNSS to return without modeling holdover.
• D. Use only horizontal position.

Answer: (A)

When GNSS is lost, the navigation solution must rely on the inertial measurement unit, and the uncertainty grows as the INS drifts due to sensor noise and biases. The test here is about predicting how long the integrated solution can stay within a specified position error bound during that outage. The best way to do this is to model the INS holdover with the error-state dynamics and propagate the error covariance over time using the known process noise characteristics. By starting from the current error covariance at outage onset and advancing it with the system’s dynamics, you obtain P(t), the evolving uncertainty. From P(t) you can determine the time at which the error bound would be exceeded—for example, by translating the covariance into a bound on position error (such as a 95% confidence ellipse) and finding when that bound is surpassed. This approach directly captures how biases, random walk, and sensor noise accumulate and interact, providing a realistic estimate of holdover duration and whether it meets the requirements.

Static estimation without dynamics would ignore how errors grow during outage, giving an unrealistically optimistic view. Waiting for GNSS to return without modeling holdover doesn’t quantify the outage duration or risk of exceeding the bound. Focusing only on horizontal position neglects the vertical component and the full 3D bound that may be specified, leading to an incomplete assessment.