QoS SIG Presentation Summary of QoS Requirements

© Copyright 2010 GISFI. All Rights Reserved.

QoS SIG PresentationSummary of QoS Requirements

Dr. Parag Pruthi NIKSUN

Dr. Ashutosh Dutta NIKSUN

Date: September 27-29, 2011

� Develop QoS framework and requirements to support

Service Oriented Network in rural environment� Network Characteristics

� Ad hoc type networks – scattered groups of users

� Lossy links

� Fluctuating traffic demand

� Low bandwidth

� Network heterogeneity (e.g., LTE and WiFi)

� User mobility pattern (e.g., cell mobility, subnet mobility)

� Application characteristics

� Current Application

�Cellular voice, File transfer (e.g., non-real-time), SMS

� Future NGN Application

�Real-time streaming (e.g., IPTV)

� Interactive (e.g., VoIP) – Delay sensitive

�Distance learning, conferencing, Mobile banking

� (Priority) Services need QoS reservation

Objective

Slide 2© GISFI

�QoE => Perceived Quality of Service

�Network Metrics

�Bit rate, delay, jitter, packet loss rate

�Power consumption

�Application Metrics

�Call setup delay

�Failed calls, dropped calls, retransmission

�MOS (Mean Opinion Score)

�Several standards groups …� ITU-T SG12

� ITU X.902 (IP Telephony)

� IETF – IPPM, DIFFSERV

� 3GPP – 3GPP TS 32.409 (IMS performance)

QoE Metrics

Slide 3© GISFI

QoS in a Service Oriented Network

IP CANCaller (UE1)

diameter

SIP

GETS Call

HSS

GETS-Application

Server

P-CSCF

I-CSCF

S-CSCF

PCRFPCRF

P-CSCF

S-CSCF

Invite

OK

ACK

Called (UE2)

Managed IP

(Multi-Provider Network)

(EPC)

DNS/ENUM

DNS

RTP

Layer 3 control

Wireless

Access

� UMTS

� EvDO

� WiMAX

� LTE

� Satellite

Wireline

Access

� Cable

� DSL

� Fiber

� Ethernet

Wireline

Access

� Cable

� DSL

� Fiber

� Ethernet

Wireless

Access

� UMTS

� EvDO

� WiMAX

� LTE

� Satellite

IMS

RAN

GW

GW

RAN

RAN IP CAN

4

IMS-layer control

4

Layer 2 control

Monitoring

Agent

Multi-layer SON

Controller

Network/

Application

Feedback

RAN

�Performance Standards

�Terminal, Customer Service, Tarriff, application

contents

�KPIs, KQI (Key Quality Indicators)

�Data Performance, KPI, KQI, SLA

�Operator control, non-operator control

�Availability, Retainability, Accessibility, Integrity, Mobility

� IRAT success rate

�Global Metrics

KPIs (AirTel Presentation)

Slide 5© GISFI

�Call setup time

�Call setup success rate

�Call completion success rate

�MOS

�Retainability

�Accessibility

� Latency

� Jitter

�Throughput

3G KPI/KQI benchmarking

Slide 6© GISFI

�FTP, Context,

�Performance benchmarking

3G Services

Slide 7© GISFI

� QoE is driven by measurements

� Which layer, which application, how often

� Require Multi-timescale Measurements

� Require Multi-Layer Measurements

� Network layer, application layer

� Require Session-based Measurements

� E.g., SIP-based multimedia application

� Require DPI for Application Classification

� Classification and Marking

� Policing and Markdown

� Queuing and Dropping

QoE Requirements

Slide 8© GISFI

� Require mechanisms to provide priority services for

signaling and transport of voice services in case of

emergency.

� e.g., SIP signaling for VoIP use Resource Priority Header

(RPH)

� e.g., Diffserv mechanisms to take care of associated media

(e.g., RTP) traffic.

� QoS measurements should consider KPIs at several layers including

� Data link Layer– MAC layer retransmission

� Network Layer - thruput, goodput, network delay, network

retransmission, utilization, packet loss, jitter

� Transport Layer - TCP connections and RTT

� Application layers – Set up delay, dropped calls, failed calls,

retransmission of INVITEs

QoE Requirements

Slide 9© GISFI

� Future NGN application requires cross-layer feedback in

order to provide desired QoS during handover and due to

bandwidth fluctuation

�IEEE 802.21

� Interoperability with Mobility protocols

�SIP, Mobile IP, Mobile IPv6

�Tightly coupled or loosely coupled

� Require correlation between signaling and media

� (e.g., EPC and IMS)

� Require Open API to dynamically control the network element in the network

QoE Requirements

Slide 10© GISFI

KPIs for RTP

KPIs for RTP

Type of KPIs KPIs Remarks

Performance Mos Average MOS score for all VoIP calls on the network

Performance Jitter Average Jitter for all VoIP calls on the network

Performance packetDelay Average Packet Delay for all VoIP calls on the network

Performance packetLoss Average Packet Loss for all VoIP calls on the network

Performance dataVolume Total of all VoIP data traffic on the network

Performance releaseCauseCode Release Cause Code distribution for all VoIP calls

Performance cdrTable List of VoIP CDRs for a given period of time

Performance faxTable List of Fax(T.38) Records for a given period of time

Perfromance rasTable List of RAS Records for a given period of time

Performance callVolumeLeaders Top N VoIP Call volume leaders

KPIs for RTP (cont.)


Performance callDistribution Total of all VoIP calls per given time period

Performance networkVolume Total of all data traffic on the network

Security Abnormally

Terminated Calls

Total of Abnormally Terminated Calls per time period

Security Initiated vs. Rejected

Calls

Total of Initiated vs. Rejected calls per time period

Security Codec Distribution List of all codecs utilized by VoIP calls

Performance Total Duration Time Total aggregated Call Duration Time of VoIP Calls per given

time period

Performance Simultaneous Calls The number of simultaneous calls on the network

Performance Total Setup Time Total aggregated Setup Time per given time period

Security errorDist Error Distribution for VoIP calls

Performance TopApps Top N VoIP protocols on the network, i.e. SIP, H.323,

MGCP/NCS, MEGACO, SCCP, T.38, etc.

KPIs RTP (contd.)


Performance mediaStream List of media streams utilized by a VoIP session

Performance faxStream List of media streams utilized by a Fax(T.38) session

Security bounceFlow Message Flow(bounce diagram contents) for a given VoIP session

Security packetTree List of all packets in the context of a VoIP session

Performance callMos Average MOS score for a given VoIP session

Performance callJitter Average Jitter for a given VoIP session

Performance callPacket

Delay

Average Packet Delay for a given VoIP session

Performance callPacket

Loss

Average Packet Loss for a given VoIP session

KPIs for IMS

KPI NAMES DESCRIPTION LOGICAL FORMULA

WHY INTERESTING?

INTERFACE

List of all SIPmethods

SIP message flows with same Call-ID

Shows the list all SIP messages.

Basic functionality for session view.

Gm

Aggregate IMS signaling packets

Sums all the methods and responses.

Ratio of total count and all IMS messages

Dos Attack Gm

Aggregate totalattempted initial registration b/w UEs and P-CSCF

Total number of registrations

Sum of all SIP Registers from all UE to PCSCF

System Loadindication, VZW Req, Dos Attack

Gm

Aggregate successful initialregistration b/w UE & CSCF

Count the successfulregistrations of UEs.

Count 200 Ok in response to Register from PCSCF.

Informative Gm

Aggregate failed initialauthentication b/w UE & CSCF

Count the number of failed authentication of UEs.

Count SIP 401 errors from PCSCF in response to Register from PCSCF.

Passwordguessing attack

Gm

Aggregate

failed initial

registration

b/w UE &

CSCF

Count the

number of

failed

registrations

of UEs.

Count SIP 4XX

(except 401) errors

from PCSCF in

response to

Register.

Informative Gm

Mean IMS

initial

registration

set-up time

Time taken

between

register and

200 OK

Total time of getting

an response after

the registration

request is sent.

System

Performanc

e indication

Gm

Aggregate

unauthenticat

ed initial

registration b/w UE & P-CSCF

Identify the

unauthenticat

ed

registrations

Count Registers

without nonces that

get 200(2XX)

responses without

401(4XX) challenges.

Security

risk, allows

impersonati

on attack,

VZW req.

Gm

Aggregate total attempted re-registration between UE & P-CSCF

Registers with previously seen Call-IDs.

Counts the totalnumber of re-registered IDs.

System Load Indication

Gm

Aggregate successful re-registration between UE & P-CSCF

Registers with previously seen Call-IDs.

Counts the totalnumber of 200 Ok’s to

Informative Gm

Aggregate Failed re-registration between UE & P-CSCF

SIP 4xx errorsin response to registration request.

Count SIP 4XX errors Password guessing attack

Gm

Mean IMS re-registration set-up time.

Time to set-up the registration

Time between register and 200 OK

System performance indication

Gm

Aggregate Failed re-authentication between UE & P-CSCF

Calculate the failed number of re-authentication.

Count SIP 401 errors in response to Register

Total unauthenticated REGs

Gm

Aggregate unauthenticated re-registration between UE & P-CSCF

Total number of unauthenticatedregistrations.

Count re-register msgs without nonces that get 200 without 401 challenges

Security risk, allows impersonation attack

Gm

Aggregate successful de-registration between UE & P-CSCF

Successful de-registrations.

Count Register with expiry==0 or contact: expiry==0

System load indication

Gm

Aggregate Failed de-registration between UE &PCSCF

All the failed de-registration.

Count SIP 4XX errors in response to above register.

Informative Gm

Aggregate total attempted de-registration b/w UE & PCSCF

Total ofattempted de-registration

SIP Passwordguessing attack

Gm

Mean IMS de-registration time

Mean of de-registration time

Mean of the time. System performance indication

Gm

Aggregate failed authentication for de-registration

Total number of failed authentication de-registrations

Count SIP 401 errors in response to register

Total unauthenticated REGs

Gm

Aggregate unauthenticatedde-registration

Total number of unauthenticated de-registration

Count de register msgs with 200 and without 401 challenges.

Security risk, allows impersonation

Gm

Attempted session establishments

How many attempts have been made to establish the session.

#initial Invite sessions with #INV==1

System load indication, VZW req.

Gm

Successful session establishments

Tracking all successful sessions we have.

Count #180 ringing, not #200 OK, when #INV==1


Gm

Answered session establishment

Tracking all successful answered sessions we have.

Count #2XXringing, should be 200 but 2XX is finewhen #INV==1


Gm

Failed session establishments

All the failed session establishments.

Count 4XXmessages


Gm

Mean time for successful session establishment

Time taken to have a success in establishment.

Time b/w INVITE and 180 response.


Gm

Attempted session re-establishment.

Number of attempts to have re-establishedsession

#INVITE where a 180 has already been seen, or #INV > 1

System tuning aid Gm

Aggregate session de-establishments

Total de-establishments.

Count BYEs. Security risk, allows impersonation attack, VZW req.

Gm

Number of attempted subscription procedures

Number of attempted procedures requests.

PCSCF receives a SIP subscribe

System load indication, VZW Req.

Gm

Number of successfulsubscription procedures

Number of successfulprocedures forsubscription.

PCSCF sends a 200 OK


Gm

Number of failedsubscription procedures

Number of failed subscription procedures

PCSCF send a 4XX


Gm

Number of unauthenticated subscribers

Total number of unauthenticated subscribers

Number of subscribers without 401 challenges.

Security risk, allows impersonation attack, VZW req.

Gm

Number of emitted notify procedures

Total count of SIP notifymessages.

PCSCF sends a SIP Notify


Gm

Number of successful notify procedures

Total count of SIP notifymessages with OK response.

PCSCF receives a SIP 200 OK

System load indication.

Gm

Number of failed notify procedures


PCSCF receives a SIP 4XX

System performance indication.

Gm

Number of unauthenticatedNotify’s


Number of Notify without a 401 Challenge

Security risk, allows impersonation attack, VZW req.

Gm

Number of spoofed notify procedures

Total count of Spoofed notifymessages.

Count notifys from src!=PCSCF


Gm

Attempted immediate Messaging procedures

Procedures attempted to haveimmediate messaging.

PCSCF receives a SIP message


Gm

Successfulimmediate Messaging procedures

Successful procedures attempted to have immediate messaging.

PCSCF sends a SIP 200 OK message


Gm

Failed immediate Messaging procedures

Failed procedures attempted to haveimmediate messaging.

PCSCF sends a SIP 4XX message in response


Gm

Number of unauthenticatedmessages

Number of all unauthenticated messages .

Messages without 401 challenges

Security risk ,allows impersonation attack

Gm

NAIs of top 5 originators of IMS signaling traffic

Display associated packets, bytes, exclude CSCF as originator.

From: and #bytes

System load indication, Dos attack

Gm

NAIs of top 5 receivers of IMS signaling traffic

Display associated packets, bytes, exclude CSCF as receiver.

To: and #bytes


Gm

NAIs of top 5 senders of Instant messages

Display associated packets, bytes, exclude CSCF as originator.

#bytes System load indication, Dos attack

Gm

NAIs of top 5 receivers of 4XX errors

Display associated packets, bytes, exclude CSCF as receiver.

#bytes System load indication

Gm

Top 5 visited n/w IDs in IMS signaling traffic

Display associated packets and bytes


Gm

Top 5 access n/w Infos in IMS signaling traffic



Gm

Top 5 senderdomains in IMS signaling traffic



Gm

Receiver domains in IMS signaling traffic



Gm

NAI of top 5 originators of register request

Display associated packets ,bytesexclude CSCF as originator(PCSCF �callee)


Gm

NAI of top 5 originators of INVITE request

Display associated packets ,bytesexclude CSCF as originator


Gm

NAI of top 5 originators of INSTANTmessage request

Display associated packets ,bytesexclude CSCF as originator(DUP)


Gm

NAI of top 5 originators of subscribe requests

Display associated packets ,bytes exclude CSCF as originator

#bytes Systemload indication

Gm

IP Addresses of top 5 originators of register requests

Display associated packets, bytes exclude PCSCF as originator.


Gm

IP Addresses of top 5 originators of INVITE requests



Gm

IP Addresses of top 5 originators of INSTANT message requests



Gm

IP Addresses of top 5 originators of subscribe requests



Gm

Top methods & response types sent. Drill down to send UE.

Calculate the number of UEs

Rate and total counts for UE

Informative Gm

KPIs FOR LTE

KPI Names Description Logical Formulae KPI

object

Why interesting? Interfa

ce

Attempted

EPS attach

procedures

Provides the

number of

attempted EPS

attach procedure

within the MME.

Sum of all attempts

at attach EPS.

Valid for

PS

Receipt of “ATTACH

REQUEST” message

with “ATTACH TYPE”

information element

indicating EPS attach

from the MS.

Successful

EPS attach

procedures

Number of

successfully

performed EPS

attach within

MME.

Sum of all

successful attaches.

Valid for

PSTransmission of a "ATTACH ACCEPT" message to the MS but retransmission doesn’t increase the counter.

Failed EPS

attach

procedures

Number of failed

EPS attach

procedures . It is

split into sub

counters per the

reject cause.

Sum of all failed

attach procedures.

Valid for

PSTransmission by the SGSN of the ATTACH REJECT message to the MS, in response to a "ATTACHREQUEST"

KPI Names Description Logical Formulae KPIobject

Why interesting?

Interface

Attempted EPS detach procedure

Number of attempted EPS detach procedures initiated by UE within the MME.

Receipt of “DETACH REQUEST” message with “detach type” information element from UE

Valid for PS

Gives the total number of attempted detaches.

Successful EPS detach procedure

Number of successful EPS detach procedures initiated by UE within the MME.

Transmission of “DETACH ACCEPT” from MME

Valid for PS

Gives the total number successfully of attempted detaches.

Attempted EPS detach procedure by MME

Number of attempted detach procedures by MME.

Transmission of “DETACH REQUEST” message by UE from the MME.

Valid forPS

All the attempts of detach to calculate success


Why interesting?

Interface

Successful EPS detach procedure by MME

Number of successful detach procedures by MME.

Receipt of “DETACH ACCEPT” message by UE from the MME.

Valid forPS

Calculate success rate

Attempted EPS detach procedure by HSS

Number of attempted detach procedures by HSS.

Receipt of “CANCELLOCATION ” message by MME from the HSS with “cancel type” information element indicating ”delete user”.

Valid forPS

All the attempts of detach by HSS and uses RADIUS.

Successful EPS detach procedure by HSS

Number of successful detach procedures by HSS.

Receipt of “CANCELLOCATION ACK” message from MME by HSS.

Valid forPS

All the successful attempts to detach by MME.


Why interesting?

Interface

Attempted tracking area update procedure with SGW change

Number of attempted tracking area update procedure with SGW change initiated with MME.

Receipt of “TRACKINGAREA UPDATE REQUEST” message from the MS, with last registered TAI information element indicating to the MME that it wishes to be served by new SGW.

Valid forPS

Calculate AREA UPDATE attempts.

Successful tracking area update procedure with SGW change

Successful number of attempted tracking area update procedure with SGW change initiated with MME.

Transmission of “TRACKING AREA UPDATE ACCEPT” message from the MS,with last registered TAI information element indicating to the MME that it wishes to be served by new SGW

Valid forPS

Success rate of Area updates.


Why interesting

?

Interface

Failed tracking area update procedure with SGW change

Number of failedtracking area update procedure with SGW change initiated with MME.

Transmission of “TRACKINGAREA UPDATE REJECT” message from the MS, with last registered TAI information element indicating to the MME that it wishes to be served by new SGW. Causes are also defined in the message.

Valid forPS

Causes of failure.

Attempted tracking area update procedure without SGW change

Number of attempted tracking area update procedure without SGW change initiated with MME.

Receipt of “TRACKINGAREA UPDATE REQUEST” message from the MS, with last registered TAI information element indicating to the MME that it wishes to be served by same SGW.

Valid forPS

Calculate AREA UPDATE attempts. Without SGW change.


Why interesti

ng?

Interface

Successful tracking area update procedure without SGW change

Successful number of attempted tracking area update procedure without SGW change initiated with MME.

Transmission of “TRACKINGAREA UPDATE ACCEPT” message from the MS, with last registered TAI information element indicating to the MME that it wishes to be served by same SGW

Valid forPS

Success rate of Area updates.

Failed tracking area update procedure without SGW change

Number of failed tracking area update procedure without SGW change initiated with MME.

Transmission of “TRACKINGAREA UPDATE REJECT” message from the MS, with last registered TAI information element indicating to the MME that it wishes to be served by same SGW. Causes are also defined in the message.

Valid forPS

Causes of failure in case of same SGW.

KPINames

Description Logical Formulae KPIobject

Why interesti

ng?

Interface

AttemptedEPS paging procedures

Number of attempted PS paging proceduresinitiated at the MME.

Incremented when an EPS paging procedure is started i.e.at the transmission of first “Paging ” message from MME to the eNodeB, which are counted when paging area is smaller than or equal to one TA.

Valid for PS.

Initial as well as repeated paging arecounted

Successful EPS paging procedures

Number of successful PS paging procedures initiated at the MME.

Receipt of “SERVICE REQUEST” message with Service Type = Paging Response from the UE.

Valid for PS

Initial and repeated are both counted.

Failed EPS paging procedures

Number of failed PS paging procedures initiated at the MME.

Increments when an EPS PS paging procedure times out, which are counted when paging area is smaller than or equal to one TA.

Valid for PS

Initial and repeated are both counted.


object

Why

interesting?

Interfa

ce

Number of

dedicated

EPS bearers

in active

mode(Mean)

Mean

number of

dedicated

EPS

bearers.

Sampling at a pre-defined

interval, sum of dedicated

EPS bearer established by

MME in active mode and

then taking the arithmetic

mean.

Valid

for

PS

The total

active

mode

bearers

would give

the load.

Number of

dedicated

EPS bearers

in active

mode(Maxim

um)

Maximum

number of

dedicated

EPS

bearers.

Sampling at a pre-defined

interval, sum of dedicated

EPS bearer established by

MME in active mode and

then taking the maximum.

Valid

for

PS

The total

active

mode

bearers

would give

the

maximum

load.


object

Why

interestin

g?

Interfa

ce

Dedicated

bearer set-up

time by

MME(Mean)

Valid time per

dedicated

bearer set-up

procedure by

MME.

Obtained by accumulating the time

intervals for every successful

dedicated bearer setup by MME B/W

“ACTIVATE DEDICATED EPS

BEARER CONTEXT REQUEST” and

corresponding “ ACTIVATE

DEDICATED EPS BEARER

CONTEXT ACCEPT” message receipt

by the MME. This time when divide by

the total number of successful

dedicated bearer gives the mean.

Valid for

PS

This would

let us

calculate

the mean

of set-up

time.

Attempted

dedicated

bearer

activation

procedure by

MME

Number of

attempted

dedicated

bearer

activation

procedure by

MME.

Transmission of “ACTIVATE


CONTEXT REQUEST” message by

MME.

Valid

FOR PS

Total

number of

attempts.

KPI Names Description Logical Formulae KPI object Why

interesting

?

Interfa

ce

Successful

dedicated

bearer

activation

procedure by

MME

Number of

successful

dedicated bearer

activation procedure

by MME.

Receipt of “ACTIVATE


CONTEXT ACCEPT”

message by MME.

Valid FOR

PS

Total

number of

successful

attempts.

Failed

dedicated

bearer

activation

procedure by

MME

Number of failed

dedicated bearer

activation procedure

by MME.

Transmission of “ACTIVATE


CONTEXT REJECT”

message by MME from UE

with “ESM cause” indicating

the cause of failure..

Valid FOR

PS

Failure

detection.

Attempted

dedicated

bearer de

activation

procedure by

MME

Number of

attempted dedicated

bearer deactivation

procedure by MME.

Transmission of “

DEACTIVATE EPS BEARER

CONTEXT REQUEST”

message by MME.

Valid FOR

PS

Total

number of

attempts to

de activate

KPI Names Description Logical Formulae KPIobje

ct

Why interesting

?

Interface

Successful dedicated bearer de activation procedure by MME

Number of successful dedicated bearer de activation procedure by MME.

Receipt of “DEACTIVATE EPS BEARER CONTEXT ACCEPT” message by MME.

Valid FOR PS

Total number ofsuccessful attempts to deactivate.

AttemptedEPS bearer modification procedure by MME

Number of attempted bearer modificationprocedure by MME.

Transmission of “MODIFY EPS BEARER CONTEXT REQUEST” message by MME.

Valid FOR PS

Total number ofattempts to modify

Successful EPS bearer modification procedure by MME

Number of successful EPS bearer modification procedure by MME.

Receipt of “MODIFYEPS BEARER CONTEXT ACCEPT” message by MME.

Valid FOR PS

Total number ofsuccessful attempts to MODIFY.

KPI Names Description Logical Formulae KPIobje

ct

Why interesti

ng?

Interface

Failed EPS bearer modification procedures by MME

Number of failed EPS bearer modification procedure by MME.

Receipt of “MODIFY EPS BEARER CONTEXT REJECT” message by MME from UE. With “ESM Cause”.

Valid for PS.

Used to calculate the success rate.

Attemptednumber of PDNGW initiated dedicated bearer creation

Number of attempted PDN-GW containing multiple bearer IDs, each cumulated to the counter.

Total Number of PDNGW created.

Valid for PS

Success rate calculation.

Successful number of PDNGW initiated dedicated bearer creation

Number of successful PDN-GW containing multiple bearer IDs, each cumulated to the counter.


Valid for PS


KPI Names Description Logical Formulae KPIobjec

t

Why interestin

g?

Interfac

e

Failed number of PDNGW initiated dedicated bearer creation

Failed PDN-GW containing multiple bearer IDs, each cumulated to the counter.


Valid for PS

Failurerate and cause of failure.

Attempted number of PDNGW initiated dedicated bearer deletion.

Number of attempted PDN-GW initiated Dedicated Bearer Deletion.

Transmission of “Delete Bearer REQUEST” message may contain multiple bearer ID, each bearer shall be cumulated to the counter.

Valid for PS

Attempts of number of deletion.

Successful number of PDNGW Dedicated bearer deletion

Number of successfully PDN-GW initiatedDedicated bearer deletion.

Receipt of “Delete Bearer Response” message by PDNGW where cause and receives a n “ACCEPTNCE RESPONSE”

Valid for PS



objec

t

Why

interestin

g?

Inte

rfac

e

Failed number

of PDNGW

initiated

dedicated

bearer deletion

Failed PDN-GW

initiated dedicated

bearer deletion.

Split into sub

counters per

reject cause.

Receipt of “Deletion

Bearer response”

MESSAGE where

“cause” is indicated.

Valid

for

PS

Failure

rate and

cause of

failure.

Attempted

number of

PDNGW

initiated

modification

with QoS

update.

Number of

attempted PDN-

GW initiated

Dedicated Bearer

modification with

QoS update

Transmission of

“Update Bearer

Request” message

from PDNGW with

“Bearer Level QoS”

containing all bearer

IDs.

Valid

for

PS

Attempt

s of

number

of QoS

updation

.

Successful

number of

PDNGW

Dedicated

bearer modification

with QoS update

Successfully

PDN-GW initiated

Dedicated bearer

modification with QoS update.

Response to the

request sent.

Valid

for

PS

Success

rate

calculati

on.


objec

t

Why

interestin

g?

Inte

rfac

e

Failed PDNGW

initiated

dedicated

bearer with

QoS update

Failed PDN-GW

initiated dedicated

bearer modification

with QoS. Split into

sub counters per

reject cause.

Receipt of “Update

Bearer response”

MESSAGE where

“cause” is

indicated.

Valid

for

PS

Failure

rate and

cause of

failure.

Attempted

number of

PDN-GW

initiated

Dedicated

Bearer

Modification

without QoS

update.

Number of attempted

PDN-GW initiated

Dedicated Bearer

Modification without

QoS update.

Transmission of

"Update Bearer

REQUEST"

message From

PDN-GW without

“Bearer Level

QoS” IE, this

message may

contains multiple

Bearer IDs

Valid

for

PS

Attempt

s of

number

of

updation

without

QoS ..


object

Why

interes

ting?

Inte

rfac

e

Successful number

of PDN-GW initiated

Dedicated Bearer

Modification

without QoS update.

Number of successfully

performed PDN-GW

initiated Dedicated Bearer

Modification without QoS

update.

Response to the request

sent.

Valid for

PS

Succes

s rate

calcula

tion.

Failed number of

PDN-GW initiated

Dedicated Bearer

Modification without

QoS update

Number of failed PDN-GW

initiated Dedicated Bearer

Modification without QoS

update. The measurement

is split into sub counters

per the reject cause

Receipt of “Update Bearer

Response” message by

PDN-GW without “Bearer

Level QoS” IE in the

“Update

Bearer Request” message

which contains the same

EPS Bearer ID and where

“Cause” IE identifies a

failed

bearer handling with

“Rejection Response

Valid for

PS

Failure

rate

and

cause.

Feedback-assisted Dynamic Network Provisioning

Motivation and application scenario

� Use of monitoring agent to provide feedback about the network

condition and dynamically control the router and assign priority to

application traffic under network congestion

EF

BE

NetDetector

controller

System architecture of testbed

Automatic configuration of router

� EDI: enhanced device interface

� Provides a comprehensive management interface for various

Cisco devices with a standard programmatic interface

Two diffserv queues:

Expedited forwarding (EF)

� match DSCP 46

� assured minimum bandwidth

- Best effort

� match DSCP 0

� set maximum bandwidth usage

Commands can be read in the form of CLI and are

encapsulated in XML format when issuing edit-config

operation of NETCONF

Query NetDetector and mark traffic

� Controller polls NetDetector periodically (~ 20 sec)

� Alternatively, Netdetector can send alarms

� Two kinds of query in form of NIKSUN URI API

� Check application type

- /ngen/srvc/dashboardData?recorder=<recorder>&iface=<iface>&startTime=-

2%20min&endTime=now&layer=Application&dataType=application&customDT=dataFields

- trigger Marking traffic to EF class once application appears

� Get application bit rate

- /ngen/srvc/dashboardData?recorder=<recorder>&iface=<iface>&startTime=-

2%20min&endTime=now&layer=Application&dataType=time,bitRate&customDT=dataField

s&window=10

- Trigger router configuration when cross traffic exceeds upper threshold

� Traffic marking is done by issuing “iptables” command at application

source host

Before marking

After marking

Packet Classification

50

Before marking

After marking

Deliver streaming media and generate cross traffic

� VLC media player works as both streaming media

server and client

� Server: RTP unicast is delivered to destination host address

using port 1234 (MPEG2)

� Client: receive the RTP stream and save it as MPEG2 file in

local host

� Iperf is used to generate cross traffic

� Client: sends UDP packets at 10Mb/s

� Server: receives the UDP packets and provides statistics on

bandwidth, loss rate, and time jitter

Experiment results (1)

�Snapshots of video

Before congestion

During

Congestion

After Dynamic

Control


� Change of quality of video & audio over time

MOS

Packet loss

No cross traffic cross trafficPriority Queue (E/F)

No cross trafficPriority Queue (E/F)


� Change of traffic received at destination host

� Change of bandwidth and time jitter for cross traffic

byte

s

time

Black: streaming media traffic

Red: cross traffic

Response time:

~ 40s


�Effect of polling time interval

20s interval:

~ 40s response time

10s interval:

~ 30s response time

Future work

� Integrating the controlling program into NetDetector

�Reducing the response time� Depends on polling interval and time spent in configuring the

router

� Using interruption/alarms rather than polling can further reduce

response time

�Being more adaptive to various scenarios� Many prerequisite information such as IP address of streaming

media server and client is assumed

� Change of the configuration of the router should not block other

normal traffic in practical situation, AF class is necessary to deal

with these traffic (control traffic)

QoE Metrics

Slide 57© GISFI

Require Session-based

Measurements

QoE metrics – Driven by measurements

Slide 58© GISFI

What To Measure?

Which Layer?

Which Application?

Which User?

Hop-by-hop or End-to-end?

Etc?

QoE Metrics

Slide 59© GISFI

Require Multi-timescale

Measurements

QoE Metrics

Slide 60© GISFI

What is Frequency of Measurement?

Duration of Flow?

Minutes?

Seconds?

Sub-seconds?

Per-packet?

QoE Metrics – Time Effect

Slide 61© GISFI

QoE Metrics

Slide 62© GISFI

Require Multi-Layer

Measurements

QoE Metrics – Network/Transport Layer

Slide 63© GISFI

IP Layer

UDP Layer

QoE Metrics – Application Specific

Slide 64© GISFI

Application HTTP

Application SSH

QoE Metrics – User centric

Slide 65© GISFI

User 1

User 2

QoE Metrics - Application

Slide 66© GISFI

All Traffic Except Facebook

Facebook Only

QoE Metrics

Slide 67© GISFI

Require DPI for Application

Classification

QoE Metrics

Slide 68© GISFI

• Require DPI for Application Classification

• Classification and Marking

• Policing and Markdown

• Queuing and Dropping

Priority QoS Treatment (Diffserv approach) in the Core IP

Transport

DSCP classification and marking at the edge, and class-based priority queuing and assured bandwidth scheduling at the edge and core routers

� EF Treatment: For Emergency SIP signaling and SIP Register transport(Strict-priority forwarding for bounded bandwidth)

� AF Treatment: For Emergency VoIP and Video/IP RTP media (Weighted shared bandwidth scheduling of remaining bandwidth among AF classes)

� BE Treatment: Best-effort traffic forwarding only when bandwidth not in use by higher priority traffic; Drop otherwise

69

7070

Experiencing Quality of Services under congestion

9.5 Mbps Cross Traffic

9 Mbps Cross Traffic



Audio Quality Video Quality

Incre

ase T

raffic

Congestion

71

VoIP Call (10 Mb/s One-way Cross Traffic)

Priority VoIP– 10 Mb/s cross traffic (packet size 1440 bytes)

Non Priority VoIP – 10 Mb/s cross traffic (packet size 1440 bytes)

Non Priority VoIP – 10 Mb/s cross traffic (packet size 400 bytes)

Applying Diffserv-based QoS to VoIP

Measurements - IMS

� Top N Reports – Clients based on

� SIP Registrations/Re-Registrations/De-Registrations

� Number of INVITE/NOTIFY/ SUBSCRIBE messages

� Failed/Successful authentications at P-CSCF/S-CSCF/I-CSCF

depending on

� Authentication type (SIP Digest, IMS AKA)

� Message type (INVITE, NOTIFY, BYE)

�Malformed messages (SIP, RTP, H.248) identified using protocol grammar, rules

�Unauthenticated/Un-authorized messages

�Spoofed messages

�Session ‘minutes’ (service usage)

Slide 72

Measurements – IMS (Contd.)

� Top N Reports – Servers based on

� Excessive Registrations/subscriptions from a given

client/proxy

� Setup requests (e.g., SIP INVITE)

� Unauthenticated/spoofed messages (e.g, SIP BYE/NOTIFY)

� Number of data sessions

� Incoming message/packet rates

�Number of authentication challenges/failure messages

�Malformed signaling/media messages

�Key exchanges during IMS AKA authentication

� Ability to drill down into data sessions - e.g., for given P-CSCF,

find all the messages from client X between time T1 to T2.

Slide 73

Measurements - LTE

� Insight into signaling and user data volumes

� Data volume distribution by � Top SGW

� Top MME

� Top eNodeB

� Top IMSI

� Volume distribution available by� Data Sessions

� Packets

� Bytes

� Setup time

� Session duration

� .. etc

QoE Metrics – QoS due to Mobility

Slide 75© GISFI

• Minimize the interruption in QoS during handover

• Interoperability with

Mobility protocols

QoE Metrics

Slide 76© GISFI

Interoperability with

Mobility protocols

Slide 77© GISFI

What affects QoE for Rural Users

� Network Characteristics

� Ad hoc type networks – scattered group of users

� Lossy links, Fluctuating Traffic demand

� Delay tolerant networks (DTN) - not suitable for interactive traffic

� Low bandwidth –> Network congestion

� Network heterogeneity (e.g., WiMAX and WiFi)

� Frequent user mobility (e.g., cell mobility, subnet mobility)

� Application characteristics

� Current Types of Application

�Cellular voice, File transfer (e.g., non-real-time), SMS

� Future NGN Application requires QoS support from lower layers

�Real-time streaming (e.g., IPTV)

� Interactive (e.g., VoIP) – Delay sensitive

�Distance learning, Mobile banking

�Group-based conferencing (e.g., Multicast)

� (Priority) Services need QoS reservation

�Baseline Document (June )

�Define a set of requirements that is application specific and applicable to Indian rural environment

�Define a set of methodologies to define end-to-end QoS parameters

�Multi-layer QOS

�Network, Application Layer

�Multi-time-scale measurements

�Granularity of measurement

�Define a set of APIs to control the network based on measurement

Next Steps

Slide 78© GISFI

KPIs FOR LTE

80

Documents

QoS SIG Presentation Summary of QoS Requirements