Integrity of GBAS relies on the protection levels based on the broadcast error models and on-ground monitors employing comparisons of multiple reference receivers. The former and latter were considered for an Istanbul reference station under real, independently marked code jumps from the GPS. In total, 60,269 smoothed test epochs were considered from a 23.8 km station pair where the fault-free vertical protection level according to the ICAO Annex 10 of the error model data tails covered the error at each epoch; with the use of standard error models, the protection level was violated at 20 epochs for the weighted least squares but the error did not exceed the 10 m alert limit. Under the two to four simulated reference receivers, the B-value monitor was able to detect almost all 20 m single-receiver jumps, hence user-side exclusion and robust estimation were mostly relevant for the case of single-receiver stations and unmonitored faults. Incorporation of Galileo resulted in the reduction of the median VDOP at Istanbul Airport from 1.48 to 1.10 and in the improvement of correct exclusion of two simultaneous faults inserted in the real smoothed code from 59% to 95%. Also, the rare Galileo E1 code deviations of about one chip length at signal acquisition were registered by the reference station; the exclusion kept the vertical effect of the deviation under 4 m, while without exclusion the impact was about 100 m.