Does it mean Tracking RS or TCI-RS set in this context?
trs-info it is mentioned only 3 times in 38.331
I couldn’t find more details about it…
trs indicates tracking reference signal.
We have only TCI (transmission configuration indicator).
Never seen TCI-RS… Is it new addition in 3GPP?
I m not aware…
You are right, it is Tracking Ref only.
So much information in contradiction here, don’t know what to take for granted or not. 
Yes correct @RFSpecialist. Actually none of docs clearly mentioned beam switching part.
You can get info for beam sweeping, beam mobility, beam failure, beam measurement and reporting but beam switching within same cell.
Tricky one .
Let’s try to dig more…
Let’s add beam part in more detail
1 beam sweeping
2 beam measurement and reporting
3 beam switching: need more clarification
4 beam mobility
5 beam failure and recovery
Please add if missed any beam part for 5G NR, so we can try to search in detail.
Whether csi beams or ssb beams are used for beam switching it is based on reportquantity IE inside RRC Connection reconfiguration.
UE may or may not be informed about beam switching.
I searched inside logs and found reportquantity for an example like CRI-RI-PMI-CQI so in this case it is based on CSI resource indicator.
It’s part of beam measurement and reporting.
Beam mobility & switching I thing same topic here.
We can add Beam Refinement with traffic beam.
Depends on interpretation.
I think any PCI-SSBid combination will be used for both cases ie within same cell or different cell with different PCI.
L1 message is like in the example posted above, and MAC CE is not in rrc reconfig message. MAC CE is layer 2 RRC reconfig is layer3.
But the tci stateid allocated by MAC is present in RRC reconfig, right?
Nope, not mandatory, this is the trick. 
When not present in TCI means UE is not informed that beams are switched.
But beams are switchiod without UE knowledge only based on UE priodic reporting.
For example CSI-beams are generated periodically every 20 msec.
UE measures them and report 4 strongest beams among all possible 64 beams, for example.
ss beam switching may be without UE agreement since idle mode.
But CS-beam switch requires handshake, right?
Intra-cell beam switching can be both transparent and seamless from the UE perspective in 5G, so UE doesn’t always have to know that beam was changed.
So he moves inside the cells beams will follow him.
I mean another beam from grid of beams will be selected for PDSCH, to replace the old beam.
Really nice explanation here:
http://www.5g-bullets.com/5G%20in%20Bullets%20-%20Beam_Refinement_Sample.pdf
Ok, anyway I can see 64 stateids in RRC reconfig message.
Need to track that drm and look for other messages to connect the dots…
MSG_tci-StatesToAddModList.txt (31.2 KB)
Yes, there can be 64 TCI states (beams) for PDCCH and 128 TCi states (beams) for PDSCH.
I guess that’s the upper limit.
This is for PDSCH.
Each of those beams points in a different direction to cover the grid of beams, i.e. the footprint on the ground.
It’s common to switch beams without informing UE.
But how can UE update the new beam used by gNB?