introduction to the automatic nodeb and cell allocation in the bsc6910 feature v1.0

8/10/2019 Introduction to the Automatic NodeB and Cell Allocation in the BSC6910 Feature V1.0

1/21

www.huawei.com

Copyright 2010 Huawei Technologies Co., Ltd. All rights reserved.

Chen Qucai (employee ID: 00160385)

July, 2012

Introduction to the AutomaticNodeB and Cell Allocation in

the BSC6910 Feature


2/21

HUAWEI TECHNOLOGIES CO., LTD. Huawei Confidential

Training Objectives

After this training, you are supposed to:

Understand the value and application scenario of the Automatic

NodeB and Cell Allocation in the RNC feature.

Understand the specifications of the feature.

Explain basic principles of the feature.

Be able to activate, configure, commission, and verify the feature

according to the related manuals.

Page3


3/21


Contents

Background

Benefits

Solution

Dependency on Other Features

Feature Deployment

Operation and Maintenance

Reference Documents

Acronyms and Abbreviations

Page4


4/21


BackgroundCurrent network conditions (BSC6900):

When configuring NodeBs, cells, NCPs, CCPs, or transmission links (SCTP/SAAL/M3ULNK), users need to specify the boards or subsystems to which the service

objects are to be bound.

The cell resource management load cannot be shared among different subsystems. When there are hotspot base stations, unnecessary flow control is triggered, and

users need to reallocate NodeBs or cells.

When capacity is expanded, additional configuration is always required. Users may need to reallocate NodeBs or cells after adding a board.

Dynamic

allocation and

reallocation of

a cell

Simplify the configuration.Users do not need to specify the board or subsystem for a new cell.

No configuration is required during

capacity expansion.Users do not need to reallocate a NodeB after adding a

board.

This feature helps meet the many-core

development trend in the future.Customers do not need to know the changes in the number

of cores and processes to avoid NodeB reallocations after

an upgrade.

Heterogenous boards need to be used at

the same time in the future.Customers do not know the cell capacity and traffic capacity of

different types of boards.

Automatic adjustment of the resource

proportions for the CP and UPUsers do not need to adjust the resource proportions for the two

planes when the capacity required by one plane, for example, theCP,Is increased.

.

The CP resources work in a resource

pool mode.If the load of a subsystem is high, cell reallocation can be easilyperformed to fully utilize the resources in a pool.

However, manually specifying the boards or subsystems for NodeBs, cells, NCPs, CCPs, or transmission links

complicates the configuration and maintenance. This mode is criticized by operators and cannot meet the future

demands.


5/21


Benefits

This feature provides the following benefits:

Configuration free

-When configuring NodeBs, cells, NCPs, CCPs, or transmission links through commands, users do not need

to specify the boards or subsystems to which the service objects are to be bound.

-When a board is added or removed, the system automatically reallocates NodeBs, cells, NCPs, CCPs, and

transmission links, which simplifies user configuration during capacity expansion or downsizing.

Automatic allocation and reallocation of NodeBs or cells

NodeBs, cells, NCPs, or CCPs are automatically reallocated based on the load.

-Immediate allocation: Used to balance the heavy load of a certain subsystem due to burst. Services are

slightly affected during the allocation. (Call drops do not occur, but new services cannot access the network

for 10s.) Immediate allocation avoids the KPI loss caused by CPU overload flow control.

-Scheduled allocation: Used to balance the load generated due to resource management. The radio

resources include NodeBs, cells, NCPs, and CCPs. Users do not need to allocate a NodeB manually,

reducing the operating expense (OPEX).


6/21


Solution

UP POOL

C

P

C

P

C

P

CP POOL

U

P

U

P

U

P

Homing

STEP 2

Add Object

such as

NodeB/Cell

STEP 1

Add Boards

STEP 3

Mapping Object to

Computing Resource

STEP 1: Run the ADD BRD command to add GPU board

resources. In this step, set LGCAPPTYPEto UCUP.

The GPU boards whose logical type is UCUP form a CP

resource pool and a UP resource pool.

STEP 2: Run the ADD UNODEB, ADD UCELL, ADD

UNCP, and ADD UCCP commands to add the following

service objects, respectively: NodeB, cell, NCP, and CCP.

When adding service objects, you do not need to specify

the physical boards or subsystems to which the service

objects are to be bound. This reduces configuration and

maintenance workloads.

STEP3:The BSC6910 automatically allocates the service

objects to the CP subsystems on the GPU boards based on

the board specifications and cell load.

Note:

- Immediate allocation: Immediate allocation is performed

based on the traffic volume in the latest 10 minutes.

-Scheduled allocation: Scheduled allocation is performed

based on the traffic volume in the latest 24 hours.

Homing Principle

- RNC Initial: object number on CP subsystem is balanced.

- Add NodeB online: added object is prior to allocate to low

load CP subsystem(whose load of radio resource

management is the lowest).

Initial configuration


7/21


Solution

Load-based dynamic

allocation

The system dynamically adjusts the binding relationship between NodeBs/cells, NCPs/CCPs and CP subsystems based on

the load of CP subsystems.

Automatic allocation

Manual allocationReviewed and

executed by customers

Note: Scheduled allocation (automatic) interrupts services, which has a significant impact on the network. Scheduled allocation

(manual) is recommended if the customer intends to accurately control the allocation, for example, when to allocate the service

objects, which NodeBs/cells/NCPs/CCPs are to be allocated, and which subsystems the service objects will be allocated to.

Dynamic allocation

Immediate allocation

(Performed during busyhours, and services are

slightly affected.)

Scheduled allocation

(Performed during idlehours, and services are

slightly affected.)

Call drops do not occur, but new services cannot

access the network for 10s.

The reallocated object is reestablished and isunavailable. Call drops occur, and the network willbe out-of-service. The out-of-service time is lessthan or equal to 30s.


8/21


SolutionScheduled Allocation (Automatic) Scheduled Allocation (Manual) Immediate Allocation

Function Switch The SchedHomingSwitchparameter is set to ON and theSchedHomingMode parameter is set to AUTO in the SET

UCELLAUTOHOMING command.

The SchedHomingSwitch parameter is set to ON andthe SchedHomingMode parameter is set to MANUAL

in the SET UCELLAUTOHOMING command.

The ImmdHomingSwitchparameter is set to ON.

Application

Scenario

Used to balance the load generated due to radio resource management. The radio resources include NodeBs, cells, NCPs,

and CCPs.

Used to balance the heavy load of a certain subsystem due to burst traffic in

hotspots. The load is generated due to cell resource management.

Burst traffic generally occurs during major assemblies or holidays.

Allocation Time It is recommended that the allocation be performed in the early

morning when traffic is the lightest. The allocation time is set inthe SET URSCADJTIME command.

The customer can decide the allocation time as required.

It is recommended that the allocation be performed inthe early morning when traffic is the lightest.

Immediate

Allocation Object NodeB, cell, NCP, and CCP Cell

Impact on

Services

The allocation has the following impact on services:

The allocated service objects will be removed and then reestablished.

Online circuit switched (CS) subscribers will experience call drops.

Online packet switched (PS) subscribers will experience service drops, but services will be automatically reconnected.

The network will be out-of-service. For a NodeB with 12 cells, the out-of-service time is less than or equal to 20s.

Note:

During scheduled allocation, service objects will be removed and then reestablished. Therefore, the values of the

performance counters for monitoring the service objects may be incredible and are for your reference only.

The allocation (not more than 10s) has the following impact on services:

New subscribers cannot access the network.

Online subscribers will not experience call drops or service drops, but nonew signaling procedure will be allowed.

Note:

During immediate allocation, service objects will be removed and then

reestablished. Therefore, the values of the performance counters formonitoring the service objects may be incredible and are for your reference

only.

Trigger

Condition

If the average CPU load of a certain subsystem is greater than SchedAssignOutCpuThd for 30 consecutive minutes,service objects may be allocated from this subsystem to another.

If the average CPU load of a certain subsystem is less than SchedAssignInCpuThdfor 30 consecutive minutes, serviceobjects may be allocated to this subsystem from another.

SchedAssignOutCpuThd and SchedAssignInCpuThd can be set in the SET UCELLAUTOHOMING command.

If the average CPU load of a certain subsystem is greater thanImmdAssignOutCpuThd for 10 consecutive minutes, service objects may be

allocated from this subsystem to another.

If the average CPU load of a certain subsystem is less thanImmdAssignInCpuThd for 10 consecutive minutes, service objects may beallocated to this subsystem from another.

ImmdAssignOutCpuThd and ImmdAssignInCpuThd can be set in the SET

UCELLAUTOHOMING command.

Observation

Method

A file named in the format of SchMovHist_20120716_21_21.txt isgenerated in the Ftp\OptMml directory on the OMU.

View

VS.AutoHoming.AttIubPort/VS.AutoHoming.SuccIubPort/VS.AutoHoming.AttNodeB/VS.AutoHoming.SuccNodeB/VS.AutoHoming.

AttCell/VS.AutoHoming.SuccCell

ALM-22240 Inappropriate NodeB Deployment isreported.

A file named in the format of iImmMovHist_20120711_11_58_00.txt isgenerated in the Ftp\OptMml directory on the OMU.

View:

VS.AutoHoming.AttIubPort/VS.AutoHoming.SuccIubPort/VS.AutoHoming.AttNodeB/VS.AutoHoming.SuccNodeB/VS.AutoHoming.AttCell/VS.AutoHoming.S

uccCell


9/21


Solution: Parameter Settings

SN Parameter IDParameter

NameDescription

1

SchedHomingS

witch

Schedule

Homing Switch

Whether to enable the schedule automatically homing algorithm. When this switch is turned on, cells,

NodeBs, and NCPs/CCPs are dynamically adjusted between subsystems based on the system load duringthe busy hour. This achieves load balancing in the RNC.

2SchedHoming

Mode

Schedule

Homing Mode

Mode in which cells, NodeBs, and NCPs/CCPs are schedule automatically homing. If this parameter is set

to MANUAL, cells, NodeBs, and NCPs/CCPs are manually homing. If ALM-22240 Advice for cell auto-

homing is reported, perform the following operations: In the ftp/optmml directory of the active workspace of

the OMU, check the optmml_yyyymmdd.txt file for recommended homing method; Download the file to a

local personal computer(PC), then following two methods can be used: upload the f ile to the

bam/version_x/ftp directory, run the RUN BATCHFILE command to execute the file; or copy the content of

the file to LMT, then use BATCH to execute the file. If this parameter is set to AUTO, cells, NodeBs, and

NCPs/CCPs are automatically homing based on the detected system load.

3ImmdHomingS

witch

Immediate

Homing Switch

Whether to enable the immediate automatically homing algorithm. When this switch is turned on, if load in

a subsystem increases rapidly due to a large amount of traffic volume in a cell, the cell is immediately

automatically homing to another lightly loaded subsystem to achieve load balancing in the RNC.

4SchedAssignO

utCpuThd

CPU Threshold

for Schedule

Assign Out

CPU usage threshold for triggering schedule automatically homing from this CP subsystem to another

subsystem. If the average CPU usage of a CP subsystem during the past 24 hours is above this threshold,

cells, NodeBs, and NCPs/CCPs in this subsystem can be assigned to another subsystem.

5SchedAssignIn

CpuThd

CPU Threshold

for Schedule

Assign In

CPU usage threshold for triggering schedule automatically homing to this CP subsystem. If the average

CPU usage of a CP subsystem during the past 24 hours is below this threshold, cells, NodeBs, and

NCPs/CCPs in another subsystem can be assigned to this subsystem.

6ImmdAssignOu

tCpuThd

CPU Threshold

for Immediate

Assign Out

CPU usage threshold for triggering immediate automatically homing from this CP subsystem to another

subsystem. If the average CPU usage of a CP subsystem in the past 10 minutes is above this threshold,

cells in this CP subsystem can be assigned to another subsystem.

7ImmdAssignInC

puThd

CPU Threshold

for Immediate

Assign In

CPU usage threshold for triggering immediate automatically homing to this CP subsystem. If the average

CPU usage of a CP subsystem in the past 10 minutes is below this threshold, cells in another subsystem

can be assigned to the subsystem.

SET UCELLAUTOHOMING:


10/21


Dependency on Other Features

Required Features

None

Mutually Exclusive Features

None

Affected Features

It is recommended that this feature be used with the WRFD-141201 RNC User Plane and Control

Plane Dynamic Sharing feature to improve hardware efficiency.

This feature is used to balance the CPU load among the subsystems in the CP resource pool.

The WRFD-141201 RNC User Plane and Control Plane Dynamic Sharing feature is used to

balance the CPU load between the CP and UP resource pools.


11/21


Feature Deployment --- Network

Configuration

Makea

networkingplan

(Optional) Addboards

Add a NodeB

Start theNodeB

Add a NodeB

(Optional) Calculate the number of boards to be added based on the board specifications.

Assume that the number is N.

Plan the NodeB data and ID of the cells under the NodeB based on the network plan.

Run the following command to add boards:

ADD BRD: SRN=?, BRDCLASS=GPU, BRDTYPE=EGPUa, LGCAPPTYPE=UCUP, SN=?;

Run the following commands to add NodeBs, cells, NCPs, and CCPs. (You do not need to

specify the physical boards or subsystems to which the NodeBs, cells, NCPs, and CCPs are to

be bound.)

ADD UNODEB: NodeBName="NodeB1", NodeBId=100, TnlBearerType=ATM_TRANS,

HostType=SINGLEHOST, SharingType=DEDICATED;

ADD UNCP: IDTYPE=BYNAME, NodeBName="NodeB1", CARRYLNKT=SCTP, SCTPLNKID=100;

ADD UCCP: IDTYPE=BYNAME, NodeBName="NodeB1", PN=0, CARRYLNKT=SCTP,

SCTPLNKID=101;

ADD UCELLQUICKSETUP: CellId=100, CellName="cell100",;

Run the following command to activate cells:

ACT UCELL: CellId=100;

F D l All i M d


12/21


Feature Deployment --- Allocation Mode

SelectionSwitch Status Problem Solved or Not

Load Imbalance Due to Cell ResourceManagement in Burst Traffic Scenarios

Load Imbalance Due to Radio ResourceManagement in Normal Traffic Scenarios

Both the scheduled allocation and immediate

allocation switches are turned on.

Both the scheduled allocation and immediate

allocation switches are turned off.x x

The scheduled allocation switch is turned off,

and the immediate allocation switch is turnedon.

x

The scheduled allocation switch is turned on,

and the immediate allocation switch is turned

off.

X

If dynamic NodeB and cell allocation is not enabled, load imbalance due to radio resource management can be

solved only by manual NodeB allocation. It is recommended that both the scheduled allocation and immediate

allocation switches be turned on.

If you intend to accurately control the allocation of NodeBs and cells among CP subsystems, consider manual

allocation.

F t D l t I di t All ti


13/21


Feature Deployment: Immediate Allocation

Collect the network

topologyinformation

Activate the feature(Configure the data)

Verify the feature

Turn on the immediate

allocation switch

N/A (Information collection is not required.)

Run the following command to turn on the immediate allocation switch:

SET UCELLAUTOHOMING: ImmdHomingSwitch=ON, ImmdAssignOutCpuThd=70,

ImmdAssignInCpuThd=60;

View the NodeB, cell, NCP, and CCP allocation records in the Ftp\OptMml

directory on the OMU one day later. Check whether there is a file named in a

format similar to ImmMovHist_20120711_11_58_00.txt.

The year and date are used as examples.

-If there is such a file, the immediate allocation has been effective.

-If there is not such a file, run the LST UCELLAUTOHOMING command to view the switch

setting.

F t D l t S h d l d All ti


14/21


Feature Deployment: Scheduled Allocation

(Automatic)

Collect the network

topologyinformation

Activate the feature(Configure the data

Verify the feature

Turn on the scheduled

allocation switch (automatic).

Obtain the off-peak hours. Off-peak hours are the hours during which the number of online UEs

is the smallest.

Set the allocation time.

Run the LST URSCADJTIME command to query the time period during which allocation is

allowed. Check whether the time period falls within the off-peak hours.

If the time period is not within the off-peak hours, run the following command to change it:

SET UCELLAUTOHOMING: StartTime=??, EndTime=???;

Run the following command to turn on the scheduled allocation switch (automatic):

SET UCELLAUTOHOMING: SchedHomingSwitch=ON, SchedHomingMode=AUTO,

SchedAssignOutCpuThd=65, SchedAssignInCpuThd=55;

View the NodeB, cell, NCP, and CCP allocation records in the Ftp\OptMml directory on the

OMU one day later. Check whether there is a file named in a format similar to

SchMovHist_20120711_11_58_00.txt. The year and date are used as examples.

-If there is such a file, the scheduled allocation has been effective.

-If there is not such a file, run the LST UCELLAUTOHOMING command to query the switch

setting.

F t D l t S h d l d All ti


15/21


Feature Deployment: Scheduled Allocation

(Manual)

Collect the network

topologyinformation

Activate the feature

(Configure the data)

Verify the feature

Turn on the scheduled allocation

switch (manual).

Obtain the off-peak hours.

Set the allocation time.

Run the LST URSCADJTIME command to query the time period during which allocation is

allowed. Check whether the time period falls within the off-peak hours.

If the time period is not within the off-peak hours, run the following command to change it:

SET UCELLAUTOHOMING: StartTime=??, EndTime=???;

Run the following command to turn on the scheduled allocation switch (manual):

SET UCELLAUTOHOMING: SchedHomingSwitch=ON, SchedHomingMode=MANUAL,

SchedAssignOutCpuThd=65, SchedAssignInCpuThd=55;

If ALM-22240 Inappropriate NodeB Deployment is reported one day later, the

scheduled allocation has been effective. If the alarm is not reported, run the LST

UCELLAUTOHOMING command to query the switch setting.

Operation and Maintenance Sched led


16/21


Operation and Maintenance: Scheduled

Allocation (Manual)

Follow the following procedure to perform maintenance when scheduled allocation

(manual) is adopted:

1.ALM-22240 Inappropriate NodeB Deployment is reported during the maintenance.

2.Download the configuration script of allocation recommendations from the Ftp\OptMml

directory on the OMU. The configuration script is named in the format of

OptMmlYearMonthDate.txt.

3.The customer confirms the configuration script of allocation recommendations. (The customer

can modify the configuration script.)

4.The customer uploads the configuration script to the ftpfolder on the OMU.

5.The customer runs the RUN BATCHFILEcommand to execute the configuration script.

Note:

In manual allocation, the customer can prepare their own configuration script without using the

generated one.

Operation and Maintenance: Immediate


17/21



Allocation/Scheduled Allocation (Automatic)

The following table lists the counters that indicate the numbers of NodeB, cell,

NCP, and CCP allocations under an RNC.

Counter ID Counter Name

VS.AutoHoming.AttIubcp Number of NCP/CCP allocation attempts

VS.AutoHoming.SuccIubcp Number of successful NCP/CCP allocations

VS.AutoHoming.AttNodeB Number of NodeB allocation attempts

VS.AutoHoming.SuccNodeB Number of successful NodeB allocations

VS.AutoHoming.AttCell Number of cell allocation attempts

VS.AutoHoming.SuccCell Number of successful cell allocations



18/21



Allocation/Scheduled Allocation (Automatic)

Users can view the NodeB, cell, NCP, and CCP allocation records in the Ftp\OptMml directory on theOMU.

If NodeBs, cells, NCPs, or CCPs experience ping-pong allocations between two subsystems, increase the

difference between the allocate-in and allocate-out thresholds. The details are as follows:

Scheduled allocation

Assume that the values of the SchedAssignOutCpuThdand SchedAssignInCpuThdparameters are

65% and 55%, respectively. If ping-pong allocations occur, run the SET UCELLAUTOHOMINGcommand

to increase the difference between the values to 15%. Below is an example:SET UCELLAUTOHOMING: SchedAssignOutCpuThd=65, SchedAssignInCpuThd=50;

Immediate allocation

Assume that the values of the ImmdAssignOutCpuThd and ImmdAssignInCpuThd parameters are

70% and 60%, respectively. If ping-pong allocations occur, run the SET UCELLAUTOHOMINGcommand

to increase the difference between the values to 20%. Below is an example:

SET UCELLAUTOHOMING: ImmdAssignOutCpuThd=70, ImmdAssignInCpuThd=50;

Feature Performance Optimization

Reference Documents


19/21


Section 4.1 "CP Load Balancing" in the WCDMA RAN Con tro l ler Resou rce Shar ing

Feature Parameter Descript i on

Page20

Reference Documents



20/21

HUAWEI TECHNOLOGIES CO., LTD. Huawei Confidential Page21

Acronym and Abbreviation Full SpellingCA Cell Automation

NCP NodeB Control Port

CCP Communication Control Port

NBM NodeB Manage Module

CP Control Plane

UP User Plane

HWT/HT Hardware Thread



21/21

Thank you!www.huawei.com