And again a seamless continuation from part 4 on 3GPP’s Non-Terrestrial Networks specification extension in Release 17 in which I’d like to come back to a particular requirement: The mobile device has to be aware of its location, which means it has to get a GPS fix before it can contact the network via a satellite. This is an interesting and quite ‘uncommon’ requirement compared to how satellite communication works today. 3GPP describes this as follows in TS 36.300:
"The UE shall have valid GNSS position as well as the ephemeris and common TA (*) before connecting to an NTN cell. To achieve synchronisation, before and during connection to a cell, the UE shall pre-compensate the Timing Advance (T TA, see TS 36.211  clause 8.1), see Figure 220.127.116.11-1, by considering the common TA, UE position and the NTN payload position through the ephemeris."
It’s a lengthy sentence but it makes it quite clear that the mobile device shall pre-compensate for the delay of the transmission to the satellite and not just perform a random access and wait for the network to adjust the initial timing advance as it would be done over the terrestrial LTE or 5G air interface. I’m particularly interested in this because today’s satellite messengers like the Garmin inReach Mini do not require a GPS fix before communicating with the Iridium network. So why is this required in NTN?
My guess is that this was perhaps necessary to keep the air interface modifications to a minimum. Or perhaps it was done to minimize the capacity required for the random access channel? I was intrigued, so I searched a bit and found out that there was actually a debate in 3GPP in 2020 on whether the mobile requires a location fix for deriving its distance to the satellite or if the cell should just broadcast a crude initial timing advance in the system information on the broadcast channel for the mobile to blindly use. In the end, the more complicated solution has prevailed to also cover large beam (i.e. cell) scenarios where the transmission delay in the center of the cell is significantly different from the delay experienced at the fringe of the beam, where the distance to the satellite is significantly longer. Have a look at here and here for some parts of that discussion.
This also answers why getting a GPS fix is not required, for example, in the Iridium system: Here, the approximate altitude of the satellites and the size of the beams are well known, so the system was designed exactly around narrowly defined parameters with no flexibility required. NTN, on the other hand covers everything from low flying LEOs to geostationary satellites so the air interface delay can vary widely, making an adaptation necessary.
3GPP also makes it clear that the GPS fix is only required for connection establishment. Once connected, the timing advanced is update by conventional means:
"In connected mode, the UE shall continuously update the Timing Advance and frequency pre-compensation, but the UE is not expected to perform GNSS acquisition."
(*) The quote also mentions a “common TA”. This is the delay of the feeder link, i.e. the delay between the satellite and the ground station. It is ‘common’, because it is the same for all devices served by the satellite.