network management and traffic analysis for wimax speaker: dedi rahmawan putra (95598052)...

Network Management and Traffic Analysis for Wireless MAN 1

NETWORK MANAGEMENT AND TRAFFIC ANALYSIS FOR WIMAX

Speaker: Dedi Rahmawan Putra (95598052)

Supervisor: Dr. Kai-Wei Ke

July, 28 2008

Network Management and Traffic Analysis for Wireless MAN

2

Outline

July, 28 2008

Simple Network Management Protocol Development of Network Management

Station (NMS) Network Management Deployment on

IEEE 802.16 system Traffic Analysis Closing Remarks


SNMPSIMPLE NETWORK MANAGEMENT PROTOCOL

Principle Operation

Standards

Object Addressing

July, 28 2008

4

July, 28 2008Network Management and Traffic Analysis for Wireless

MAN

SMI (Structure of Management Information)Define how management information may look

like SMIv2 (currently used)

MIBs (Management Information Bases)Tell what management information exists MIB-II (currently used)

SNMP (Simple Network Management Protocol)Define how information is exchanged SNMPv1, SNMPv2c, SNMPv3

Standards

5


MAN

NMS

Agent

MIB

poll

poll

poll

TRAPS

GET/SET

variable

table

MIBMIB

AgentAgent

Principle Operation

6


MAN

Ensuring that each variable

gets a unique name

for interne

t people

for company

or university

Object Addressing (1)

OID: 1.3.6.1.2.1

7


MAN

OID: 1.3.6.1.2.1 wmanIfMib is defined under

this branch

Object Addressing (2)

OID: 1.3.6.1.2.1.1.1

e.g.Scalar Instance: 1.3.6.1.2.1.1.1.0Table Instance: 1.3.6.1.2.1.2.1.1.1

GET 1.3.6.1.2.1.1.1.0 = “MP.16 BS v.130” Column number

Index Value


THE DEVELOPMENT OF NETWORK MANAGEMENT STATION (NMS)

Challenges

System Context Diagram

System Architecture

Front Page

Simple NMS Workflow

NMS DEMONSTRATION

July, 28 2008


9

July, 28 2008

Remote Accessibility Continuous Monitoring Cross Platform Widely Applicable for SNMP-Capable Device User friendly tools and menu Configurable Objects parameters Satisfying Requirements: Operation,

Administration, Maintenance, Provisioning, Functions

Robust Feature Extendable

Challenges

10


MAN

MIB

MIB

MIB

User

Presentation Logic

Control Logic

Database Driver

SNMP Tools

LAN/MAN/WAN

Network Management

Station

Database Server

Agents

WiMAX BS

WiMAX SS Access Point Printer

SNMP

SNMP

SNMP

SNMP SNMPPC

MIBMIB

System Context Diagram

11


MAN

Database

Session Beans

Entity Beans

ServletsJSP

Pages

Java EE Server

Web Container

EJB Container

Client

Browser

Applets

Class Library

System Architecture

12


MAN

Front Page

13


MAN

Object Registra

tion

Agent Registrati

on

Agent-Object

Mapping

Object Monitorin

g

Simple NMS Workflow


NMS DEMONSTRATION

July, 28 2008


NETWORK MANAGEMENT DEPLOYMENT ON IEEE 802.16 SYSTEM

Network Management System Deployment

Challenges

Monitoring Parameters

Monitoring Scenario

Monitoring Result Snapshots

July, 28 2008

16


MAN

WiMAX BS140.124.183.210

WiMAX SS-1140.124.183.211

AP-1140.124.183.213

Internet

WiMAX SS-2140.124.183.212

AP-2140.124.183.214

NMS140.124.183.224

User Client

Clients

Clients

Gateway

Wired or Wireless

Network Management System Deployment

17


MAN

Wireless Environment Unstable WiMAX Test-Bed Network

System Limited MIB Agents Time Constrain Collecting Adequate Data

Challenges

18

Monitoring Parameters


MAN

QoS Service Class: UGS, rtPS, nrtPS, Best Effort

Modulation Type: BPSK, QPSK, 16QAM, 64QAM Traffic Intensity: light load, heavy load Data flow: Downlink (DL), Uplink (UL) Data Traffic:

UGS: VoIP rtPS: MPEG video, web-TV nrtPS: file transfer (FTP) BE: internet application(web browsing), email


19

Monitoring Scenario

July, 28 2008

WiMAX BS140.124.183.210

WiMAX SS-1140.124.183.211

WiMAX SS-2140.124.183.212

WiMAX Wireless Mediumeth0 eth0

eth0

OUT

IN OUT

IN

OUT

IN

wmanIfBsSsPacketCounterTableNOT

EXIST

ifTable on MIB-II StandardifInOctets

ifInUcastPktsifOutOctetsifOutUcastPkts

• May introduce inaccurate result• Traffic Pattern can be observed


20

Monitoring Result Snapshots

July, 28 2008

e.g. Best Effort Heavy Load on SS-1


TRAFFIC ANALYSIS

Self-Similar vs Poisson Traffic Model

Challenges

UGS Service Class

rtPS Service Class

nrtPS Service Class

Best Effort Service Class

Analysis Summary

July, 28 2008

22


MAN

Poisson Model Self-Similar ModelConsiders network arrival as a random process.

Uses autocorrelation and does not consider the network traffic to be random.

Does not scale the bursty traffic properly.

Scales bursty traffic well, because it has similar characteristics on any scale.

In fine scale, bursty traffic appears bursty, while in coarse scale, bursty traffic appears smoothed out and looks like random noise.

Gives more accurate pictures due to Long Range Dependence in the network traffic.

Self-Similarity vs. Poisson Model

23


MAN

Inaccurate Data source Short Duration Data Self-similarity assessment methods:

Pictorial Guess, Variance-Time Plot, Rescaled Adjusted Range (R/S) Plot

Hurst Parameter Estimation

Challenges


24

UGS Service Class (1)

July, 28 2008

0 100 200 300 400 5000

50

100

150ifOutUcastPkts-SS1 (UGS Heavy)

Time Unit = 1000 milliseconds

Val

ue

0 10 20 30 40 500

5000



Val

ue

0 50 100 150 200 2500

1000



Val

ue

0 100 200 300 400 5000

20

40

60ifOutUcastPkts-SS1 (UGS Light)


Val

ue

0 10 20 30 40 50 600

1000

2000



Val

ue

0 50 100 150 200 250 3000

200

400



Val

ue


25

0 0.5 1 1.5 2 2.5 30

1

2

3R/S Plot ifOutUcastPkts-SS1 (UGS-Heavy)

log10(N)

log

10(R

/S)

0 0.5 1 1.5 2 2.5 3-2

-1

0

1

2Variance Time Plot ifOutUcastPkts-SS1 (UGS Heavy)

log10(m)

log

10(V

aria

nce

s)

UGS Service Class (2)

July, 28 2008

0 0.5 1 1.5 2 2.5 3-2

-1

0

1Variance Time Plot ifOutUcastPkts-SS1 (UGS Light)

log10(m)

log

10(V

aria

nce

s)

0 0.5 1 1.5 2 2.5 30

1

2

3R/S Plot ifOutUcastPkts-SS1 (UGS Light)

log10(N)

log

10(r

/s)

β = -1.0329

H = 0.4836

Slope = -1 Slope = -1

β = -0.8488

H = 0.5756

Slope = 0.5

Slope = 1Slope = 0.5

Slope = 1H = 0.6828 H = 0.5316


26

rtPS Service Class (1)

July, 28 2008

0 100 200 300 400 5000

100

200

300ifOutUcastPkts-SS1 (rtPS Heavy)


Val

ue

0 10 20 30 400

5000

10000



Val

ue

0 50 100 150 2000

1000

2000



Val

ue

0 100 200 300 400 5000

100

200

300ifOutUcastPkts-SS1 (rtPS Light)


Val

ue

0 50 100 150 2000

1000



Val

ue

0 5 10 15 20 25 30 350

2000

4000



Val

ue


27

0 0.5 1 1.5 2 2.5 30

1

2

3

4Variance Time Plot ifOutUcastPkts-SS1 (rtPS Heavy)

log10(m)

log

10(V

aria

nce

s)

rtPS Service Class (2)

July, 28 2008

0 0.5 1 1.5 2 2.5 30

1

2

3Variance Time Plot ifOutUcastPkts-SS1 (rtPS Light)

log10(m)

log

10(V

aria

nce

s)

0 0.5 1 1.5 2 2.5 30

0.5

1

1.5

2

2.5

3R/S Plot ifOutUcastPkts-SS1 (rtPS Heavy)

log10(N)

log

10(r

/s)

0 0.5 1 1.5 2 2.5 30

1

2

3R/S Plot ifOutUcastPkts-SS1 (rtPS Light)

log10(N)

log

10(r

/s)

β = -0.4495

H = 0.7752

Slope = -1

Slope = -1

β = -1.1594

H = 0.4203

Slope = 0.5

Slope = 1

Slope = 0.5

Slope = 1H = 0.5507 H = 0.7686


28

nrtPS Service Class

July, 28 2008

0 100 200 300 400 5000

200

400

600ifOutUcastPkts-SS1 (nrtPS Heavy)


Val

ue

0 50 100 150 200 2500

2000

4000



Val

ue

0 10 20 30 40 500

1

2

3x 10



Val

ue

0 100 200 300 400 5000

200

400

600ifOutUcastPkts-SS1 (nrtPS Light)


Val

ue

0 10 20 30 400

1

2x 10



Val

ue

0 50 100 150 2000

2000



Val

ue


29

nrtPS Service Class

July, 28 2008

0 0.5 1 1.5 2 2.5 31

1.5

2

2.5

3

3.5

4Variance Time Plot ifOutUcastPkts-SS1 (nrtPS Heavy)

log10(m)

log

10(V

aria

nce

s)

0 0.5 1 1.5 2 2.5 31

1.5

2

2.5

3

3.5

4Variance Time Plot ifOutUcastPkts-SS1 (nrtPS Light)

log10(m)

log

10(V

aria

nce

s)

0 0.5 1 1.5 2 2.5 30

0.5

1

1.5

2

2.5

3R/S Plot ifOutUcastPkts-SS1 (nrtPS Heavy)

log10(N)

log

10(r

/s)

0 0.5 1 1.5 2 2.5 30

0.5

1

1.5

2

2.5

3R/S Plot ifOutUcastPkts-SS1 (nrtPS Light)

log10(N)

log

10(r

/s)

β = -0.2832 H = 0.8584

Slope = -1Slope = -1

β = -0.5039

H = 0.7480

Slope = 0.5

Slope = 1

Slope = 0.5

Slope = 1

H = 0.8409 H = 0.5933


30

Best Effort Service Class (1)

July, 28 2008

0 100 200 300 400 5000

100

200

300ifOutUcastPkts-SS1 (BE Heavy)


Val

ue

0 20 40 60 800

5000



Val

ue

0 100 200 300 4000

1000

2000



Val

ue

0 100 200 300 400 5000

200

400ifOutUcastPkts-SS1 (BE Light)


Val

ue

0 10 20 30 40 500

5000



Val

ue

0 50 100 150 200 2500

1000

2000



Val

ue


31

Best Effort Service Class (2)

July, 28 2008

0 0.5 1 1.5 2 2.5 30.5

1

1.5

2

2.5

3

3.5Variance Time Plot ifOutUcastPkts-SS1 (BE Heavy)

log10(m)

log

10(V

aria

nce

s)

0 0.5 1 1.5 2 2.5 31

2

3

4Variance Time Plot ifOutUcastPkts-SS1 (BE Light)

log10(m)

log

10(V

aria

nce

s)

0 0.5 1 1.5 2 2.5 30

0.5

1

1.5

2

2.5

3R/S Plot ifOutUcastPkts-SS1 (BE Heavy)

log10(N)

log

10(r

/s)

0 0.5 1 1.5 2 2.5 30

0.5

1

1.5

2

2.5

3R/S Plot ifOutUcastPkts-SS1 (BE Light)

log10(N)

log

10(r

/s)

β = -0.5333

H = 0.7333Slope = -1

Slope = -1

β = -1.1141

H = 0.4429

Slope = 0.5

Slope = 1Slope = 0.5

Slope = 1

H = 0.7958 H = 0.7413


32

Analysis Summary

July, 28 2008

Rank

QoS Category

Traffic Intensi

ty

Pictorial Guess

Hurst variance

plot

Hurst by R/S Plot

Average Hurst

1 nrtPS Heavy Self-Similar 0.8584 0.8409 0.8497

2 Best Effort Heavy Self-

Similar 0.7333 0.7958 0.7646

3 nrtPS Light Self-Similar 0.7480 0.5933 0.6707

4 rtPS Heavy Poisson 0.7752 0.5507 0.6630

5 rtPS Light Poisson 0.4203 0.7686 0.5945

6 Best Effort Light Self-

Similar 0.4429 0.7413 0.5921

7 UGS Heavy Uniform 0.4836 0.6828 0.5832

8 UGS Light Uniform 0.5756 0.5306 0.5531


CLOSING REMARKS

Conclusions

Future Works

July, 28 2008

34

Conclusions


MAN

The NMS had been designed, implemented, and deployed successfully.

The monitoring procedure was done on Ethernet interface using the objects of MIB-II standard.

The response time in notifying the network manager much depends on wireless link quality and the traffic intensity.

Different QoS class and traffic intensity generated different traffic behavior and self-similarity degree.

35

Future Works


MAN

Continuing the incremental sequence to NMS Development. Defining additional features to support more functionality

into NMS system, e.g. providing statistical analysis Integrating the NMS with another system, e.g. developing

Java application client. Developing J2ME application is also possible for operation on

mobile device. Establishing additional research on self-similar model

towards WiMAX QoS. Collecting more rigorous data to get more convincing result. Repeatable analysis and measurement once the WiMAX MIB

interface is available.


THANK YOU

July, 28 2008


37

Network Management Requirement Analysis

July, 28 2008

Operation deals with keeping the network up and running smoothly.

Administration deals with keeping track of resources in the network and how they are assigned.

Maintenance regards with performing repairs, upgrades, and corrective and preventive measures to make the managed network run better.

Provisioning is concerned with configuring resources in the network to support a given service.

Functions that are performed as part of network management accordingly.

38

Retrieving Scalar Value


MAN

1.1.0

1.2.1.0 1.2.2.0

Object OID: 1.1

Instance OID: 1.1.0

Value

GET 1.1.0 = 140.124.181.210

39

Retrieving Table Entries Value


MAN

X.C.I = OidOfTable.ColumnNumber.IndexValue

new-MIB.routeTable.next.8 = 3

destination

1.3.2.8 = 3

40


MAN

FeaturesSatisfied Requirements

Operation Administration Maintenance Provisioning Functions

Object Registration x x

Object Summary x

Agent Registration x x

Agent Summary x

Agent object Management x x x

Agent Status Control x x x

Static Monitoring x x x

Dynamic Monitoring x x

Grouped Monitoring x x

Manager System Control x x

Agent Management Structure x x

Admin Login Control x x x

Server Time x x

Ping Tool x x x

Get Tool x x x

Set Tool x x x x

Logs of Activities x x

Feature Refinement

41


MAN

+getId() : Long+setId(in id : Long) : void+getIpAddress() : String+setIpAddress(in ipAddress : String) : void+getName() : String+setName(in name : String) : void

-id : Long-ipAddress : String-name : String-type : String-deviceStatus : Boolean-monitorStatus : Boolean-snmpVersion : int-snmpPort : int-communityWrite : String-communityRead : String

AgentEntity

+getId() : Long+setId(in id : Long) : void+getName() : String+setName(in name : String) : void+getSyntax() : String+setSyntax(in syntax : String) : void

-id : Long-oid : String-name : String-syntax : String-maxaccess : String-description : String-retrievalMethod : String-asDefault : Boolean-intervalRetrieval : int-enumeration : Boolean

ObjectEntity

+getId() : Long+setId(in id : Long) : void+equals(in object : Object) : Boolean

-id : Long

AgentObjectEntity

+getId() : Long+setId(in id : Long) : void+getInstanceValue() : int+setInstanceValue(in value : int) : void+getMeaning() : String+setMeaning(in meaning : String) : void

-id : Long-instanceValue : int-meaning : String

EnumInstanceEntity

+getId() : Long+setId(in id : Long) : void+getMoment() : Long+setMoment(in moment : Long) : void+getInstanceValue() : String+setInstanceValue(in value : String) : void

-id : Long-moment : Long-retrievalValue : String

MonitoringEntity

-agentEntity

0..*1

0..*

-objectEntity

1

1

-gateway0..1

1

-objectEntity0..*

-agentObjectEntity 0..*

1

Implementation Class Diagram

42


MAN

Header

Footer

Content

Content

Menu

Content

Content

Containers and contents are implemented using JSP

GUI Layout

43


MAN

aimed to detect the device status utilizes isReachable() function call

provided by Java. Implemented using Stateless Session

Bean within EJB Container.

Ping Function

44

Get Function


MAN

Used for monitoring feature and simple Get feature.

Implemented using Stateless Session Bean within EJB container.

To perform the value retrieval from a specific object ID in the MIB agent.

Utilizing Snmp4j library to create SNMP PDU and execute Get command.

45


MAN


Used for simple Set tool feature To create SetRequest-PDU and execute Set command, we also utilize Snmp4j library.

Set Function

46


MAN

To schedule a sequence of actions by giving a certain delay in between.


Constructed by 3 sub functions: initiating timer, cancelling timer, timeout actions.

Used for dynamic monitoring and Auto-ping feature.

Timer Interval

47

Chart Generator


MAN

Generates the chart presentations on monitoring procedure.

Utilizing JFreeChart library. Two kinds of charting presentation available: bar

or histogram time series chart, and line time series chart.

Implemented in: Servlet within web container: for Static PNG image Applet within Applet Class Library: for real monitoring

The chart is a time series domain chart drawn against the retrieval value

48


MAN

displaying hidden field

Object Registration UI

49


MAN

Object ID Object Name Syntax: the object data type Max-Access: the maximum possible access Description Make Default for New agent: special field

intended to distinguish which objects should be included in a newly-registered agent by default.

Object Registration Field

50


MAN

meaningless status value for human readable

more understandable meaning

Syntax Enumeration

51


MAN

Once: retrieved once when the monitoring starts. Eg: SystemDescription, systemName.

Continuously-Updated-Value: retrieved and updated regularly. Eg: systemUpTime.

Line-Chart: retrieved regularly, the previously retrieval data will be kept.

Bar-Chart: basically similar with Line-Chart one, but using bar-chart or histogram to illustrate the monitored object.

newly appeared field

Object Retrieval Method

52


MAN

hidden by default

show hidden field

Agent Registration UI

53


MAN

IP Address Agent name Device type: the level position in the WiMAX network

deployment hierarchy: WiMAX-BS, WiMAX-SS, AP, CLIENT

Upper level gateway: which upper gateway the device belongs to.

SNMP version : which compatible SNMP standard the device agent is designed for: v1, v2c, v3

Read Community: community string for read access. Write Community: community string for write access. SNMP Port

Agent Registration Field

54


MAN

Need confirmation: monitoring in progress? restart

Object naming convention: name + index for Table instances

Agent Object Management

55


MAN

•OK•Need Attention•Not Performed

Agent Status Control

56


MAN

Used for: Once and Continuously-Updated-Value

Static Monitoring

57


MAN

Line-Chart

Object NameBar-Chart(Histogram

)

Dynamic Monitoring

58


MAN

wmanIfMib = 1.3.6.1.2.1.10.184

wmanIfMibObjects = 1.3.6.1.2.1.10.184.1

wmanIfBsObjects (1) wmanIfSsObjects (2) wmanIfCommonObjects (3)

wmanIfBsPacketCswmanIfBsCpswmanIfBsPkmObjectswmanIfBsNotificationwmanIfBsPhy

wmanIfSsPkmObjectswmanIfSsNotificationwmanIfSsPhy

wmanIfCmnPacketCswmanIfCmnCpswmanIfCmnPkmObjects

wmanIfMibConformance = 1.3.6.1.2.1.10.184.2

wmanIfMibGroups (1) wmanIfMibCompliances (2)

IEEE Std. 802.16f-2005

59


MAN

Object ID Object Name Syntax Max-Access

1.3.6.1.2.1.2.2.1.2.x1 ifDescr OCTET-STRING READ-ONLY

1.3.6.1.2.1.2.2.1.5.x1 ifSpeed Gauge32 READ-ONLY

1.3.6.1.2.1.2.2.1.10.x1 ifInOctets Counter32 READ-ONLY

1.3.6.1.2.1.2.2.1.11.x1 ifInUcastPkts Counter32 READ-ONLY

1.3.6.1.2.1.2.2.1.12.x1 ifInNUcastPkts Counter32 READ-ONLY

1.3.6.1.2.1.2.2.1.14.x1 ifInErrors Counter32 READ-ONLY

1.3.6.1.2.1.2.2.1.16.x1 ifOutOctets Counter32 READ-ONLY

1.3.6.1.2.1.2.2.1.17.x1 ifOutUcastPkts Counter32 READ-ONLY

1.3.6.1.2.1.2.2.1.18.x1

ifOutNUcastOctets Counter32 READ-ONLY

MIB Extraction

60


MAN

Management Scenario Flowchart


61

WiMAX QoS Classes

July, 28 2008

UGS (Unsolicited Grant Service) Used for real-time constant bit rate service that gives periodic transfer data at similar packet size (burst) and sensitive towards throughput, latency, and jitter.

Real Time Polling Service (rtPS)Used for real-time service flows with periodic variable size data packets which is sensitive toward the throughput and latency, but with loosely tolerance compared to UGS guaranteed rate and delay requirement

Non-Real-Time Polling Service (nrtPS) Used for non real-time services with regular variable size burst that needs intensive throughput with minimal guarantee on its latency

Best Effort (BE) This service is dedicated to the network traffic that doesn’t require guaranteed data rate and delay


62

Equations

July, 28 2008

Variance-Time

Rescaled Adjusted Range (R/S)

Linear Regression

network management and traffic analysis for wimax speaker: dedi rahmawan putra (95598052)...

Documents

wireless man

system traffic analysis

page slide

system architecture

simple nms workflow

system context diagram

university object

branch object