Mp load
mp load control mechanism (new in L13A)
Main contributors (signaling) for MP Load
• RRC Connection Setups
• Incoming Handovers
• Paging Requests
Mp load
1.1 RRC Connection setup intensity
(due to UE releases by inactivity)
• The time a UE can be inactive before it is released is controlled
by tInactivityTimer
• Current setting for tInactivityTimer is 10s, requirement from UE
vendor
• Recommend to change to 30s to reduce signaling load for high
capacity cells
For tInactivityTimer set to 30s,
• E-RAB drop rate will increase due to the E-RAB drops only consider the abnormal E-RAB Releases
if there was data in buffer (i.e. active).
• Number of UEs stay in RRC_Connected state will be larger, thus lowering the number of available
Connected Users License for new users
Mp load
1.2 RRC Connection setup intensity
(due to UE releases by rLC Supervision) – cont’
• As part of Radio Connection Supervision, RLC failures are caused by
– T310 expiry
– Max RLC re-transmissions reached
• T310 is controlled by
– After receiving N310 consecutive “out-of-sync” indications, start timer T310
– After receiving N311 consecutive “in-sync” indications, stop timer T310
– After T310 expired Radio Link Failure
– Current settings:
• N310 = 20 (recommended, no change)
• N311 = 1 (recommended, no change)
• T310 = 2s (recommended, no change)
UE is released after T310 (2s) + T311 (3s)
Mp load
1.2 RRC Connection setup intensity
(due to UE releases by rLC Supervision)
• RLC re-transmissions are controlled by
– Max # RLC re-transmissions before stopping and indicating to RRC max RLC re-tx reached
• SRB/DRB :: dlMaxRetxThreshold (Current setting: 8/8, recommended, no change)
• SRB/DRB :: ulMaxRetxThreshold (Current setting: 8/8, recommended, no change)
– Poll timer for RLC re-Tx (if no status report received)
• SRB/DRB :: tPollRetransmitDl (Current: 45ms/50ms, recommend 80ms/80ms)
• SRB/DRB :: tPollRetransmitUl (Current: 45ms/50ms, recommend 80ms/80ms)
Re-Tx count
Increase Poll timer may reduce
SE/TTI usage for re-Tx and free
SE/TTI for traffic, but re-Tx time is
extended from 360-400ms to 640ms
#1 (80ms x 8)
#8 RLC Failure
Mp load
2. incoming handovers intensity
• Current setting for handover is
– ReportConfigEUtraBestCell::a3offset = 3dB
– ReportConfigEUtraBestCell:: hysteresisA3 = 1dB
– ReportConfigEUtraBestCell:: timeToTriggerA3 = 40ms
• Propose to change a3offset to 5dB (handover margin increased to
6dB) for high capacity cells to reduce signaling load. a3offsetAnrDelta
remains unchanged
Increase handover margin reduces
handovers and may impact
retainability
Will review feasibility to use
alternative approach on ‘UE Level
Oscillating Handover Minimization’
feature
Mp load
3. hardware expansion
• MP capacity for DUS is higher than MP capacity
on DUL
Mp load
observability
• pmProcessorLoad
– The average CPU load on the whole MP system (all the CPU cores).
• pmProcessorLoadDistr
– A distribution of the CPU load of the MP corresponding to the definition of the CPU load used by MP Load
Management.
– This counter also includes CPU load consumption from execution on priorities below traffic priority. This includes CPU
load from performance management and some O&M tasks.
• pmPagS1Received
– This counter counts the number of received S1AP paging messages in the RBS.
• pmPagS1Discarded
– The number of discarded S1AP paging messages not routed to any cell due to high load (MP Load Control).
• pmRrcConnEstabFailHighLoad
– The total number of failed RRC connection establishments due to high load (MP Load Control).
• pmRrcConnEstabFailOverload
– The total number of failed RRC connection establishments due to overload (Procedure Latency Supervision).
• pmHoPrepRejInHighLoad
– The number of Handover Preparation Failure messages sent by the target eNodeB due to high load (MP Load Ccontrol).
• pmHoPrepRejInOverload
– The number of Handover Preparation Failure messages sent by the target eNodeB due to overload (Procedure Latency
Supervision).
•
