You forgot something: once PRB Load Balancing is activated there is no difference between carriers in terms of PRB load.
Why would you want to shift users on a specific carrier?
I have managed highly crowded events load bakancing by IMLB only.
How load balancing can be achived without moving UEs from one carrier to another?
By Connected Mode PRB Load Balancing.
So user shifting will happen in this case also.
User shifting required signalling, etc.
PRB load balancing in Connected Mode makes sense.
I don’t find it reasonable the Idle Mode one.
Same can be achived in Idle LB, but with less efficiency compared to MLB.
Actually i disagree.
I have implemented Idle LB in wide network, and it’s pretty good.
@RFSpecialist, just small feedback from my side:
In Huawei, we have Idle Mode MLB, Connected Mode and MBOCS (enhance connected mode based on not only UE number/SE, it will consider UE usage to make it good balance).
Idle MLB is handle based on PRB usage/UE number but UE transfer will be after RRC release with priority.
As only Idle MLB is not much efficient due to less RRC inactivity release UE in busy hour.
We need connected mode MLB based on A4 event.
Connected mode MLB is having complex formula to calculate PRB/ SE/ UE count to transfer UE by handover.
In heavy event, i found Idle MLB is not much efficient due to very less UE release RRC (10 sec) so connected MLB is giving very good gain.
So we must have Idle MLB and connected MLB both to get equal load.
1 Like
Idle LB is efficient in layers with contiguous BW as in non contiguos BW UE will always go to low utilized which is normaly low BW layer as IMMC will override SIB5 and in IMMC we mostly get low BW as high priority for non CA supported handsets.
This also what I want to say, thanks 
Perfectly right.
So in heavy events, we have feature that dynamically change inactivity timer to lower value.
And we get good gain.
Right.
I don’t have much info about dynamically change of inactivity timer.
But how are you managing signalling surge on RRC/ERAB if inactivity timer became 1,2,3 sec?
Secondly, do you know any MLB mechnism which can work base on Uplink load/ uplink user count/ uplink SE?
As in heavy event, 90 % user activity because of uplink.
Static inactivity timer is 10s for us.
And we do not allow it to reduce less than 5s.
In between it can be changed as per curreny rrc users in a cell.
Secondly only IMLB is not sufficient, MLB is must.
But IMBL can reduce burden on MLB.
We have definition of Load.
Which is MAX of DL load and UL load.
We have multiplying parameters by which we can make DL load as zero.
So finaly MLB work based on UL load only.
Secondly, we also ensure that most of the NGBR traffic should be carried by carrier having high UL Capacity. i.e. in High BW FDD.
And not in low BW FDD or TDD.
if you still are not convinced let me tell you that even with offsets we are balancing 3 or 4 layers… and those offsets are only valid in idle mode (as priorities are the same)
now, regarding mlb algorithm (idle, interfreq, prb or user, or whatever) still didnt have a positive result yet. …