Radalt Failures
The radio altimteter was designed as a landing aid, and has been adapted for uses beyond, such as GPWS. The signal bounced from the ground gets weaker with the square of the distance, so funny things can happen. Other aircraft, and even heavy rain clouds can reflect the signal, giving a false lockon with a perfectly operating system. Ice without a wet surface is nearly invisible to it, however. Since you are using a computer, you are conversant with speeds/frequencies. The radalt sends and receives on 4.35 GHz, +/- 100 MHz, Frequency Modulated, Continuous Wave.
Most radalt installation failures are due to such simple things as corrosion between the antennas and the fuselage, and coaxial connector fittings at the back of the transceiver mounting tray. Symptoms are almost always a lock onto a false signal at altitude, where the ground return signal is weak, and therefore the transceiver reception is at full volume. Back below 300', most quirks disappear, making them difficult for Sparky to confirm.
I spent a month one week in Auk chasing authrottle retards as 250' on a fleet of 747-200. The pilots were concerned, of course... Lack of electrical bonding between the radalt antennas and the flame sprayed belly panels, and a signal switching inside the radalt at 250' conspired to cause one radalt to go to near 0 feet briefly, triggering A/T retard.
Every radalt anomaly should be reported to maintenance, before it gets worse. Since it is a vital part of a Cat IIIb Autoland system, any altitude indication higher than correct should be reported to authorities.
For dual autoland, of course, at least two radalts must be operating and indicating near identical altitude. Glideslope signal is washed out of the autoland equation by 50', as the radalt indication has been brought in. Below about 40', in widebodies at least, the approach is ballistic.
GB