Future open Galileo signals on E1, E5a and E5b bands will provide measurements with reduced tracking errors thanks to their signal structure. Moreover, because they are located on different bands, they also allow for efficient ionospheric delay estimation. The consequence is an accurate positioning capability, even based on code-based pseudorange measurements. This article investigates the event of a triple-frequency Galileo receiver losing the Galileo E1 signal. In this case, ionospheric delay estimation becomes difficult since the E5a and E5b frequency bands are very close. Several methods are investigated for this purpose. It is seen that the use of a vertical Total Electron Content model to represent the ionosphere delay for all visible satellites is very promising. When included in a Kalman filter, it offers an ionospheric delay estimation accuracy at the sub-meter level with a convergence time of only a few minutes. An extension is then provided to the single-frequency case where only E5b is available. A slight degradation in accuracy and convergence time is obtained compared to the dual-frequency E5a/E5b case is observed, although the results are also promising.