Introduction
Handover is a key technique to ensure that UEs can freely move in networks while enjoying
high-quality communication services. MRO takes handover performance statistics for different
scenarios, identifies abnormal handover scenarios, and optimizes the handover-related
parameter settings. The optimization achieves better resource utilization and satisfies user
experience requirements.
Benefits
With the development of mobile telecommunications technologies, networks are growing and
incorporate multiple radio access technologies (RATs), resulting in complicated network
maintenance. To simplify maintenance, the LTE system is required to support the SON. MRO is
used in the SON as one of the self-optimization functionalities.
MRO architecture
When the MRO period approaches its end, the eNodeB modifies its parameter settings
based on the number of abnormal handovers.
3. Result monitoring
After modifying the parameters, the eNodeB monitors handover-related performance
indicators.
If handover performance improves, the eNodeB retains the parameter settings in the next
MRO period.
If handover performance deteriorates, the eNodeB rolls back to the previous parameter
settings in the next MRO period.
Intra-frequency handovers are triggered by event A3, and inter-frequency handovers are
triggered by event A3, A4, or A2. The parameters to be adjusted for MRO are the cell
individual offset (CIO) for event A3, CIO for event A4, and threshold for event A2.
For details about the CIOs and threshold, see Mobility Management in Connected Mode
Feature Parameter Description.
Figure 3-2 shows premature and delayed handovers, assuming that a UE is handed over
from cell A to cell B.
A2-related
A2-related delayed inter-frequency handovers occur because the threshold for event A2
is set too low.
The source cell does not attempt to or fails to deliver a gap-assisted measurement
configuration to a UE, and the UE moves out of the source cell and an RLF occurs.
The RRC connection is then reestablished to another cell (called target cell).
The source cell receives an RLF INDICATION message from the target cell over an
X2 interface and detects that it has the context identified in the message.
This type of delayed handover can be further classified into A3- and A4-oriented
because the RSRP threshold for A3-oriented inter-frequency event A2 is different from
that for A4-oriented inter-frequency event A2. For details about the events, see Mobility
Management in Connected Mode Feature Parameter Description.
A3- or A4-related
A3- or A4-related delayed inter-frequency handovers occur because the CIO for event
A3 or A4 is set too small.
When an MRO period approaches its end, if the proportion of coverage-induced RLFs to all
premature and delayed handovers reaches the value of CoverAbnormalThd, the eNodeB
does not perform MRO in this period.
Type 1: A UE receives a handover command. After the handover from cell 1 to cell 2,
the UE camps on cell 2 for only 3s and then experiences an RLF. The RRC connection
is then reestablished to cell 3 for the UE. In this type of handover, cell 2 was
inappropriately selected as the target cell because there is a high probability of a
handover to cell 2.
Type 2: A UE receives a handover command. During the handover from cell 1 to cell 2,
the UE experiences an RLF. The RRC connection is then reestablished to cell 3 for the
UE. In this type of handover, cell 2 was inappropriately selected as the target cell
because the signal quality of cell 2 is unstable and there is a low probability of a
handover to cell 3.
In the early phase of network deployment, the number of handovers within an MRO period
cannot reach the preceding threshold in many cells; however, RLFs frequently occur in
these cells. In this situation, these cells cannot enter the MRO procedure. If users change
the value of OptPeriod or StatNumThd to enable these cells to enter the MRO procedure,
the statistics are not useful. To solve this problem, Huawei provides the following solution:
12. If the number of handovers between a pair of neighboring cells reaches the threshold
within the first MRO period, these cells enter the MRO procedure.
13. If the number of handovers between a pair of neighboring cells does not reach the
threshold within the first MRO period, the eNodeB retains the number of handovers.
Within 30 MRO periods, these cells enter the MRO procedure if the cumulative number
of handovers reaches the threshold at the end of any of the 30 periods.
14. If the cumulative number of handovers within 30 MRO periods does not reach the
threshold, the eNodeB clears the number and these cells do not enter the MRO
procedure.
Huawei eNodeB takes premature and delayed handovers together into consideration during
MRO procedures: The eNodeB identifies premature and delayed handovers and records the
number of premature handovers and the number of delayed handovers for the
corresponding cell pairs in NRTs. Based on the proportion of premature or delayed
handovers, the eNodeB determines how to modify parameters for MRO to minimize the
number of RLFs caused by premature or delayed handovers.
The eNodeB records parameter adjustments, if any, for each MRO period.
NOTE:
If the proportion of premature handovers to the total number of abnormal RLFs during
handovers exceeds the abnormal RLF threshold (70%), the eNodeB decreases the CIO
for intra-frequency event A3 by one step.
If the proportion of delayed handovers to the total number of abnormal RLFs during
handovers exceeds the abnormal RLF threshold (70%), the eNodeB increases the CIO
for intra-frequency event A3 by one step.
The eNodeB does not perform MRO in an MRO period during which users have manually
performed any of the following modifications online: adjusting the CIO or other handover-
related parameters (such as the hysteresis, threshold, offset, time-to-trigger, and filtering
coefficient), adding a cell to the blacklist, or removing a cell from the blacklist. In the next
MRO period, the eNodeB will perform MRO based on the manual modifications. In addition,
during MRO evaluation, the eNodeB does not consider how the RLF proportions fluctuate
between MRO periods.
For details about the CIO value range, see 3.2.4 CIO Value Range Constraints.
Inter-Frequency MRO
The MRO feature of Huawei eNodeBs for inter-frequency neighboring cells is controlled by a
switch under the MroSwitch parameter. Delayed inter-frequency handovers are classified
into A2-related delayed handovers and non-A2-related delayed handovers. Therefore, in
addition to the conditions for intra-frequency MRO, inter-frequency MRO has the following
triggering conditions:
The MRO method for A3-configuration-triggering event A2 is the same as that for A4-
configuration-triggering event A2.
If the proportion of A2-related delayed handovers is less than the specified threshold,
the eNodeB performs MRO against premature handovers or non-A2-related delayed
handovers. If premature handovers or non-A2-related delayed handovers occur, the
eNodeB adjusts the CIO value for event A3 or event A4. The modification principles and
the parameters involved are the same as those of MRO for intra-frequency neighboring
cells.
NOTE:
The preceding specified threshold has a fixed value of 1/20. It is not configurable.
Event A2 is used to start inter-frequency measurement, and event A1 is used to stop inter-
frequency measurement. The threshold for event A2 must be lower than the threshold for
event A1. The threshold for event A1 must be set appropriately. If the threshold for event A1
is set excessively low, the eNodeB cannot resolve A2-related delayed handovers by
adjusting the threshold for event A2. If the threshold for event A1 is set excessively high, the
number of unnecessary measurements increases.
The eNodeB does not perform MRO in an MRO period during which users have manually
performed any of the following modifications online: adjusting the CIO or other handover-
related parameters (such as the hysteresis, threshold, offset, time-to-trigger, and filtering
coefficient), adding a cell to the blacklist, or removing a cell from the blacklist. In the next
MRO period, the eNodeB will perform MRO based on the manual modifications. In addition,
during MRO evaluation, the eNodeB does not consider how the RLF proportions fluctuate
between MRO periods.
For details about the CIO value range, see 3.2.4 CIO Value Range Constraints.
Cell-Level MRO
If an eNodeB has performed MRO against premature or delayed handovers between a local
cell and a neighboring cell indicated in an NRT within an MRO period, the eNodeB does not
perform MRO against ping-pong handovers between the cells in this period. If the eNodeB
has not performed MRO against premature or delayed handovers, the eNodeB checks the
proportion of ping-pong handovers between the cells in this period.
If the proportion is greater than the value of PingpongRatioThd, the eNodeB decreases the
CIO by one step. The proportion is calculated as follows:
Intra- and inter-frequency MRO against ping-pong handovers follow the same mechanism,
except that intra-frequency MRO adjusts the CIO for intra-frequency event A3 and inter-
frequency MRO adjusts the CIO for inter-frequency event A4 or A3.
UE-Level MRO
To decrease the number of ping-pong handovers, the eNodeB identifies ping-pong UEs and
delivers CIO values to these UEs.
Assuming that the UE enters cell A X consecutive times according to the UE History
Information IE, the UePingPongNumThd parameter is set to N, and
the PingpongTimeThd parameter is set to M, the eNodeB delivers the CIO to the UE by
adhering to the following principles:
If X is less than N, the eNodeB does not consider the UE as a ping-pong UE and
delivers the configured cell-specific CIO to the UE.
If X is greater than or equal to N and average stay time 1 is less than M, the eNodeB
considers the UE as a ping-pong UE.
When X is greater than or equal to N + 1 and average stay time 2 is greater than or
equal to M, the eNodeB decreases the configured cell-specific CIO by one step for
the UE and delivers the result to the UE.
When X is greater than or equal to N + 1 and average stay time 2 is less than M, the
eNodeB decreases the configured cell-specific CIO by two steps for the UE and
delivers the result to the UE.
NOTE:
Average stay time 1 = Total time of stay in cell B for N consecutive times/N
Average stay time 2 = Total time of stay in cell B for (N + 1) consecutive times/(N + 1)
The eNodeB counts the number of ping-pong handovers according to the latest UE History
Information IE, regardless of whether the eNodeB has considered this UE as a ping-pong
UE during the UE-level MRO period.
A large CIO value adjustment may result in a high service-drop probability. This 2 dB limit
reduces the probability of a service drop caused by low reference signal (RS) signal to
interference plus noise ratio (SINR) in the source cell of a handover. Service drops have a
negative impact on user experience. Therefore, the UE-specific CIO value can be
decreased by a maximum of 2 dB based on the cell-level CIO.
The eNodeB takes special actions for UE-level MRO in the following scenarios:
If a UE that has experienced a handover failure has its RRC connection reestablished
with the source cell, the eNodeB considers that the handover failure was caused by an
abnormal RLF, and then does not treat this UE as a ping-pong UE or perform UE-level
MRO.
If a UE handed over to a cell meets the ping-pong UE requirement, the eNodeB delivers
the dedicated CIO value to the UE. If the UE using this CIO value experiences a
successful RRC connection reestablishment, the eNodeB retains this CIO value.
When a UE-level MRO period (which is permanently 4 hours) approaches its end, the
eNodeB postpones UE-level MRO by 50 seconds (fixed value) to prevent MRO conflicts
if the eNodeB has adjusted parameter settings for cell-level MRO.
UE-level MRO brings relatively higher gains in the following scenario: A UE camps on the
edges of two cells, where signal fluctuations may result in relatively more ping-pong
handovers. For example, a stationary UE continuously performs services in the handover
area between two cells.
However, UE-level MRO may be ineffective in certain test scenarios. For example:
Generally, a neighboring cell can provide continuous services only when Mn is higher than -
110 dBm. Therefore, Huawei eNodeB calculates the upper limit of the CIO value range for
event A4 according to the following function:
In summary, the CIO value range for inter-frequency event A4 is [-24,Min(Thresh + 110 -
Ofn + Hys)].
For details about the parameters for inter-frequency events A3 and A4, see Mobility
Management in Connected Mode Feature Parameter Description.
In the MRO process for intra-frequency neighboring cells in an LTE network, the eNodeB
does not check whether the parameter rollback conditions are met.
Cell-Level Penalty
Ping-pong adjustments of parameter settings may occur between MRO periods. Huawei
eNodeB monitors the latest three parameter adjustments during MRO periods. If the last
value is identical with the first value, the eNodeB assumes that a ping-pong parameter
adjustment occurs. As a penalty, the eNodeB will not perform MRO throughout the next two
MRO periods, each specified by OptPeriod.
UE-Level Penalty
When a UE-level ping-pong handover MRO period (which is permanently 4 hours)
approaches its end, the eNodeB calculates the proportion of RLFs due to delayed
handovers as follows:
If the eNodeB has delivered the adjusted CIO values to some UEs before imposing a UE-
level penalty, the eNodeB does not change the CIO values back.
4 Inter-RAT MRO
Inter-RAT MRO is a process to optimize the parameter settings related to handovers from
an E-UTRAN cell to an inter-RAT cell.
Currently, inter-RAT MRO can be used only when an E-UTRAN covers the same area as a
UTRAN or GERAN (not both) that provides better coverage. This is because inter-RAT
MRO has not been defined appropriately in 3GPP specifications: If a delayed handover
occurs, the eNodeB cannot determine which system the RRC connection is reestablished
for the UE that has experienced an RLF or determine whether the UE enters a coverage
hole.
The number of delayed inter-RAT handovers is equal to the total number of A2- and B1-
related delayed handovers.
An A2-related delayed handover occurs if all of the following conditions are met:
An B1-related delayed handover occurs if all of the following conditions are met:
The eNodeB receives an A2 measurement report and successfully delivers a gap-
assisted measurement configuration, but it does not or fails to deliver a handover
command.
The eNodeB does not receive a premature or delayed intra-RAT handover indication.
The source cell has at least one valid inter-RAT neighboring cell, and the inter-RAT
neighboring cell supports the service on the UE.
The source cell deletes the UE context when the relevant timer expires.
The method that the eNodeB uses to measure the number of B1-related delayed handovers
varies with the following scenarios:
The eNodeB does not receive an inter-RAT measurement report. In this scenario, the
eNodeB does not know the target RAT and therefore cannot measure the number of
delayed handovers to that RAT. If there is an inter-RAT neighboring cell and the UE
supports that RAT, Huawei eNodeB increases the QCI-specific number of B1-related
delayed handovers by 1 for that RAT.
The eNodeB receives an inter-RAT measurement report. In this scenario, the eNodeB
increases the QCI-specific number of B1-related delayed handovers by 1 for the inter-
RAT system to which the best cell indicated in the measurement report belongs.
For Huawei eNodeBs, MRO for mobility to UTRAN and MRO for mobility to GERAN are
controlled by UtranMroSwitch and GeranMroSwitch, respectively.
When one of the preceding switches is turned on, the eNodeB identifies and measures
abnormal handovers to the corresponding RAT and then modifies parameters related to
handovers to that RAT. When an MRO period approaches its end, the eNodeB triggers
MRO only if all the following conditions are met:
NOTE:
The abnormal handover threshold and proportion fluctuation threshold have fixed values of
1/20 and 0, respectively. They are not configurable.
The eNodeB does not perform MRO in an MRO period during which users have manually
adjusted the CIO or other handover-related parameters (such as the hysteresis, threshold,
offset, time-to-trigger, and filtering coefficient) online. In the next MRO period, the eNodeB
will perform MRO based on the manual modifications. In addition, during MRO evaluation,
the eNodeB does not consider how the abnormal handover proportions fluctuate between
MRO periods.
NOTE:
The threshold for A2-related delayed handover proportion has a fixed value of 1/20. It is
not configurable.
NOTE:
Event A2 is used to start inter-RAT measurement, and event A1 is used to stop inter-RAT
measurement. The threshold for event A2 must be lower than the threshold for event A1.
The threshold for event A1 must be set appropriately. If the threshold for event A1 is set
excessively low, the eNodeB cannot resolve A2-related delayed handovers by adjusting the
threshold for event A2. If the threshold for event A1 is set excessively high, the number of
unnecessary measurements increases.
Network Performance
Intra-RAT MRO and inter-RAT MRO for mobility from E-UTRAN to GERAN/UTRAN
minimize the number of premature handovers, delayed handovers, and ping-pong
handovers in the network.
7 Engineering Guidelines
7.1 When to Use MRO
MRO can be enabled only if an X2 interface is available between eNodeBs. If the X2
interface is unavailable between eNodeBs, RLF indication messages cannot be transmitted
over the X2 interface, and the eNodeBs cannot count the number of premature or delayed
handovers. In this case, the MRO algorithm cannot make a correct parameter adjustment.
Intra-RAT MRO
Use intra-RAT MRO when one of the following conditions is met:
Inter-RAT MRO
Use inter-RAT MRO when the inter-RAT handover success rate is lower than expected.
Inter-RAT MRO
Collect the following information for inter-RAT MRO:
7.3 Planning
RF Planning
MRO optimizes handover-related parameters and has the following coverage-related RF
planning requirements:
No coverage holes
No cross-cell coverage
No pilot pollution
No imbalance between uplink and downlink