SSB Frequency Location SA and NSA

Hello, Humans of NR.

I’ve observed that for NSA scenarios some vendors use absoluteFrequencySSB which don’t meet the GSCN criteria. Example for n78 629988 and n28 155456.
Question 1: Is there any advantage of not using GSCN when on NSA?

I understand that for NSA we use absoluteFrequencySSB which indicates the direct ARFCN thus not properly needing to be GSCN, while using GSCN on SA is key part of the cell search procedure.
Question 2: how would the UE behave if at SA the carrier is configured with a SSB location not-GSCN? Have you ever performed tests under such scenario?

Cheers!

For question 1, I guess this is done to prevent SA UEs from initiating the access attempts on the NR cell which is meant to to be used only for NSA acess. Since SA UE searches only GSCN freqs, it will never find that NSA NR cell whose SSB is scheduled on non-GSCN Freq. The other ways of doing this is, enabling the barring bit in SSB of NSA NR cell and not scheduling the RMSI completely etc.

For question 2, I think that kind of configuration is incorrect. SSB of SA carrier must be at GSCN Freq otherwise it will never be found by SA UE during initial access.

1 Like

@marcengo Good questions

A1: ideally, you can have higher peak throughput if you properly allocate the SSB. For example, if resource allocation type 1 is used, it is better to have SSB at the edge of the channel (maximize consecutive RBs). In case of resource allocation type 0, you can try to adjust SSB not to partition more than 1 RBG

A2: I doubt vendor will allow to being SA cell up with SSB not on GSCN. However, in case it is allowed, SA UE cannot find the cell defined by this SSB

1 Like

Hi, Pradeep. Thanks for your comments!

For Question 1: in this scenario you mentioned, SA UE would not access it since the NSA cell does not broadcast SIB1, regardless of the SSB frequency location. My question is more centered in the performance point of view, what would be the performance benefits of such implementation.

For Question 2: I agree it is incorrect from the “literature” point of view (specs, bullets, etc…) but it is not uncommon to UEs implement non-3GPP algorithms to handle some NW issues, that`s why I was wondering if anyone did really test this scenario.

Hi, O.B.

A1: I´ll be honest with you, in my internal conversations with folks from the operator, this is the only item I could imagine it would bring some impact. In this case I was looking at, the two SSBs frequency are located 900kHz apart from the other, which could overlap with some RGB group. I need to do the math.

I read on Bullets about the the tradeoffs of SSBs location, for example, centered in the middle of the band could be more representative since it would represent an “average” in terms of fidelity to interference (the good old, lower the frequency, better propagation) across the whole channel bandwidth, especially for larger BW`s sizes, for RSRP and RSRQ measurements.

Another point it brings is the consideration about the CSI-RS reporting, which follows the same logic as PRBs allocation you mentioned… in the middle may impact how you setup the CSI-RS as it would “break” the reporting in average.

For question 2, the vendor indicates clearly it should be GSCN for SA, so I agree with you. It is more a very “scientific” question to see what would happen :slight_smile:

Thanks your your replies!

1 Like

For CSI-RS, it could be time division multiplexed with SSB either slot or symbol level