Hi Ali,

I don't think it has been completely standardized yet. I doubt the purpose of it is for slotless transmission only.

"However, how mini-slot transmission works, if it does not follow the frame structure " ?

During RRC_ACTIVE, also most everyone including mini-slots has to follow the frame structure. For example, mini-slots also improve packing efficiency. For some applications, IoT liked applications or VoIP liked applications, running in a wideband system, e.g., 100MHz or wider, sometimes it is hard for NB/UEs to fill up 7 symbols for each transmission. This can be waste, can't it?

Mini-slot transmission in the Time domain is now standardized in Release 15 of 3GPP. The purpose is to support URLLC and to allow the 5G network to support heterogenous traffic with different data sizes.

For more details follow the below-mentioned link:

https://www.etsi.org/deliver/etsi_tr/138900_138999/138912/15.00.00_60/tr_138912v150000p.pdf

I don't know how detailed it is in the 3GPP spec but I do know that it is standardized now. As you said it is intended for URLLC mission critical services where low latency is required such that the network cannot wait for the end of the slot - - in fact, it turns out to be unnecessary.

Again I don't know the details in the spec but for a guess: Since it is just the scheduler task, gNB must be scheduling the corresponding UEs. One guess is, even if the number of symbols do not add up, network may still be following 10ms frame structure which I think is the case. This may be provided by adding guard symbols during mini-slot transmission.

Thank you for sharing Noman Aftab