Content uploaded by Perumalraja Rengaraju
Author content
All content in this area was uploaded by Perumalraja Rengaraju on Jun 27, 2016
Content may be subject to copyright.
IEEE Wireless Communications • August 2012 89
1536-1284/12/$25.00 © 2012 IEEE
S3 S4
S5a S5b
Evolved packet core
IASA
MME
UPE
SAE
anchor
3GPP
anchor
SGSN GPRS core
non 3GPP
IP access W
ACCEPTED FROM OPEN CALL
INTRODUCTION
Two emerging technologies, the Worldwide
Inter-operability for Microwave Access
(WiMAX) and Long Term Evolution (LTE)
have been designed with different QoS frame-
works to deliver high performance Internet
applications [1, 2]. WiMAX is based on IEEE
802.16 standards, whereas LTE is based on the
3rd Generation Partnership Project (3GPP).
Both technologies are being considered as candi-
dates for the 4th Generation (4G) broadband
wireless networks. The WiMAX standard uses a
new physical layer radio access technology called
Orthogonal Frequency Division Multiple Access
(OFDMA) for Uplink (UL) and Downlink (DL)
data transmissions; LTE uses OFDMA for DL
and Single carrier Frequency Division Multiple
Access (SFDMA) for UL data transmissions to
mitigate frequency selective fading. Apart from
this, both technologies are using Multiple Input
and Multiple Output (MIMO) technique to
improve the physical layer spectral efficiency.
Adaptive Modulation and Coding (AMC) in
WiMAX and LTE, Sleep mode—idle mode
operations in WiMAX, Discontinuous Reception
and Transmission (DRX/DTX) strategies in
LTE, and also many implementation specific
cross-layer algorithms like opportunistic sched-
ulers, adaptive radio resource management func-
tions etc., are introduced in both WiMAX and
LTE networks. These improvements definitely
improve the QoS performance of the radio
access network, but the user’s satisfaction relies
on both E2E QoS and end user application.
The International Telecommunication Union
(ITU) defines the Quality of Experience (QoE)
as “overall acceptability of an application or par-
ticular service, as perceived subjectively by the end
users” [9]. The QoE is also a consequence of a
user’s internal state, the characteristics of the
designed system, and the environment within
which the service is experienced. It can be mea-
sured by measuring both subjective and objective
measurements. The subjective measurements are
related to how the user experiences the applica-
tion, e.g., the application initiation time, etc.,
and the objective measurements are the QoS
parameters that are related to subjective mea-
surement like E2E delay, etc. Assuring constant
QoE at the customer end is a major task for the
service providers. The parameters that affect
QoE can be classified into four groups:
Quality of audio and video content at the
source: It mainly depends on the codec and its
characteristics (G.7xx for voice; H.26x, Microsoft
codec (WMV), etc. for video).
QoS of the E2E network from source to user
terminal: The QoS parameters that affect the
performance of the E2E applications are band-
width, delay, jitter, and packet loss.
User terminal: Performance of the user
equipment and the codec used.
Human perception: It is usually captured by a
Mean Opinion Score (MOS) that includes expec-
tations and ambiance. The MOS is expressed on
PERUMALRAJA RENGARAJU, CHUNG-HORNG LUNG, AND F. R ICHARD YU, CARLETON UNIVERSITY
ANAND SRINIVASAN, EION INC.
ABSTRACT
From the users and service providers point of
view, the upcoming 4G wireless (WiMAX and
LTE) networks are expected to deliver high per-
formance sensitive applications like live mobile
TV, video calling, mobile video services, etc. The
4G networks are intended to provide an accu-
rate service view of customer-perceived service
quality — their “Quality of Experience” or QoE.
Delivering high QoE depends on factors that
contribute to the user’s perception of the target
services as well as the Quality of Service (QoS)
of the network. Although a better network QoS
in many cases will result in better QoE, fulfilling
all traffic QoS parameters alone may not guar-
antee satisfied users. On the other hand, if the
QoS of the network degrades, the QoE of users’
applications could be affected significantly. This
article presents an integrated view of End-to-
End (E2E) QoS and service assurance support
in WiMAX and LTE networks. The integrated
view also considers the existing cross-layer imple-
mentations to achieve the necessary E2E QoS
and how the QoS and QoE should and can be
monitored over the network in an E2E fashion.
The existing cross-layer implementations ensure
the E2E QoS performance before establishing
the applications. However, even though the net-
work has strong QoS support, the service pro-
viders should still monitor the QoS of the
network and QoE of users’ applications accu-
rately to achieve service quality assurance.
ONQOE MONITORING AND E2E SERVICE
ASSURANCE IN 4G WIRELESS NETWORKS
The authors present
an integrated view
of End-to-End QoS
and service assur-
ance support in
WiMAX and LTE net-
works. The integrat-
ed view also
considers the exist-
ing cross-layer imple-
mentations to
achieve the neces-
sary E2E QoS and
how the QoS and
QoE should and can
be monitored over
the network in an
E2E fashion.
RENGARAJU LAYOUT_Layout 1 8/9/12 9:15 AM Page 89
IEEE Wireless Communications • August 2012
90
a five point scale (ITU-T P.800), where 5 equals
excellent, 4 equals good, 3 equals fair, 2 equals
poor, and 1equals bad. The minimum threshold
for acceptable quality corresponds to a MOS of
3.5 [9].
There is a significant difference between what
a network application experiences as quality at
the network level (i.e., QoS), and what the user
perceives as quality at the application level (i.e.,
QoE). From the applications point of view, the
parameters that are related to the measurement
of QoE are called Key Quality Indicators
(KQIs). KQIs are constructed from Key Perfor-
mance Indicators (KPIs), which are derived from
network performance measurements. The defini-
tions of KQI and KPI and the methodology of
developing KQI for each service are defined in
“wireless service measurements handbook-
GB923” [6]. The development of KPIs and KQIs
is useful for the service providers to define the
Service Level Agreements (SLAs) and for the
network monitoring purposes. There are three
objective methodologies for measuring the QoE
[9]:
1 The No-Reference model has no knowledge of
the original source file (application) and tries
to predict QoE by monitoring several QoS
KPIs in real-time.
2 The Reduced Reference model has some lim-
ited knowledge of the original source and tries
to combine it with QoS KPIs to reach a pre-
diction on the QoE.
3 The Full Reference model has full access to
the source file (possibly the reference, e.g.,
video or audio, files.), then combined with the
measurements conducted in a real-time envi-
ronment.
As the Full Reference model has the knowl-
edge of the original applications, it is possible to
give the best accuracy on the measurement. On
the other hand, the No-Reference model is sim-
ple to measure, but may not always give accurate
results.
This article intends to provide the integrated
view of E2E QoS service assurance support in
both WiMAX and LTE networks and the archi-
tecture for QoE monitoring based on the knowl-
edge of QoS-KPI. For that, we first describe the
QoS support from WiMAX and LTE standards
and how this QoS support can be extended for
E2E networks using cross-layer implementations.
Next, we describe how and where the QoS and
QoE can be monitored in an E2E network to
assure the service assurance.
The rest of the article is organized as follows.
We describe the QoS support in WiMAX and
LTE networks. We discuss the extended cross-
layer works in WiMAX and LTE to support the
E2E QoS for intended user applications. Then
E2E QoS and QoE monitoring in 4G wireless
networks is discussed. Finally, we conclude this
article.
QOS SUPPORT IN
WIMAX AND LTE NETWORKS
QOS SUPPORT IN WIMAX NETWORKS [1]
Figure 1 shows the E2E flat IP architecture of
WiMAX networks. Here, the Access Service
Network (ASN) is maintained by the Network
Access Provider (NAP) to provide the WiMAX
radio access, and the Connectivity Service Net-
work (CSN) is maintained by the Network Ser-
vice Provider (NSP) to provide IP connectivity
and WiMAX services to the subscribers accord-
ing to negotiated SLAs. The logical entities in
the ASN network include Mobile Station (MS),
Base Station (BS), and ASN Gateway (ASN-
GW), while the CSN network consists of AAA
(Authentication, Authorization and Accounting)
server, Operations Support Systems (OSS),
Mobile IP — Home Agent (MIP-HA) and CSN-
GW. The CSN-GW is connected to the applica-
tion service provider network, i.e. Public
Switched Telephone Network, and other non-
WiMAX networks like 3GPP and WiFi net-
works. The Network Working Group (NWG) in
the WiMAX forum is still working on E2E QoS
functionality.
The user’s service flow authorization details
are stored in the AAA server. When a user is
connected to the network, the service flow infor-
mation is temporarily stored in the BS or ASN
gateway. The Radio Resource Management
function (RRM) in ASN-GW, schedulers, UL
and DL bandwidth allocations and admission
Figure 1. WiMAX E2E network architecture.
ASP
BSMS
BSMS
BSMS
ASN
GW
Internet
PSTN
3GPP/
3GPP2
Access
network
IP
network
IP
network
AAA
Gateway
MIP-HA
Connectivity
service network
(CSN)
Access service
network (ASN)
OSS/BSS
The CSN-GW is con-
nected to the appli-
cation service
provider network like
Public Switched Tele-
phone Network, and
other non-WiMAX
networks like 3GPP
and WiFi networks.
The Network Work-
ing Group in WiMAX
forum is still working
on E2E QoS
functionality.
RENGARAJU LAYOUT_Layout 1 8/9/12 9:15 AM Page 90
IEEE Wireless Communications • August 2012 91
controller in BS ensures that the access network
QoS will be based on this user provisioning.
Wireless QoS support is one of the funda-
mental parts of the MAC layer design in IEEE
802.16 standards, and it is based on service
flows. A service flow is a logical unidirectional
flow of packets between the ASN-GW and a MS
with a particular set of QoS attributes. These
SFs may be created, changed, or deleted through
a series of MAC management messages. On the
other hand, the BS temporarily assigns a Con-
nection Identifier for each service flow and for
basic connectivity. The MAC convergence sub-
layer maps the service flow identifier and con-
nection identifier for each UL and DL data
transfer. Similarly, the traffic mapping between
layer-2 and layer-3 QoS for appropriate SFs is
done at the ASN-GW for DL and at the MS for
UL directions, respectively. Between the ASN-
GW and the BS, the QoS of the SFs is support-
ed by backhaul transport QoS. To handle
different applications separately, the IEEE
802.16 standards define five types of service
flows with different QoS requirements as given
below.
• Unsolicited Grant Service (UGS): The UGS is
designed to support real-time service flows
that generate fixed size data packets on a
periodic basis, like Voice over IP (VoIP).
• Real Time Polling Service (rtPS): The rtPS
service is designed to support real-time service
flows that generate variable size packets on a
periodic basis, such as MPEG video.
• Extended Real Time Polling Service (e-rtPS):
The ertPS is designed for real time traffic with
variable data rate such as VoIP service with
silence suppression.
• Non-Real Time Polling Service (nrtPS): This
service is introduced for non-real-time flows,
which require variable size data grants on a
regular basis, such as high bandwidth FTP.
• Best Effort Service (BE): This service is
designed to support best effort traffic and
offers no guarantee.
With the available QoS support and QoS-
based algorithms like scheduler and admission
controller in WiMAX networks, the end user
can get a consistent QoS support.
QOS SUPPORT IN LTE NETWORKS [2]
The LTE standard is established by 3GPP to
compete with WiMAX standards. Figure 2 shows
the high level flat IP architecture of LTE and its
interfaces. Here, the Evolved Radio Access Net-
work (E-RAN) is maintained by the access
provider to provide the LTE radio access for the
users. The logical entities in E-RAN networks
have User Equipment (UE) and evolved Node B
(eNB), also called BS. The evolved packet core
(EPC) is maintained by the LTE service pro-
viders to provide the IP connectivity and other
services according to negotiated SLAs. To assure
the user’s QoS requirement in the LTE network,
SLAs of user profiles are temporarily stored in
the eNB. BW allocations for UL and DL service
flows are based on the SLAs.
In LTE networks, the E2E QoS is established
from UE to the PDN-GW in a core network.
The E2E connectivity between UE and PDN
GW in LTE-SAE networks is established using
bearer service. It provides the QoS level of gran-
ularity in the LTE for different service flows.
Figure 3 shows the E2E QoS support and EPS
bearer establishment in the LTE networks. The
radio bearers are established using the RRC
protocol. While it carries information on the
radio interface, the S1 bearer forwards the infor-
mation between eNB and MME/SGW, and the
S5 bearer transports the packets to the Packet
Data Network — Gateway (PDN-GW). EPS
bearers are established between UE and PDN-
GW. So the UL and DL bearer mapping of an
individual radio bearer and EPS bearer is done
Figure 2. IP-based LTE-SAE network [2].
SGi
Rx+
S1
Iu
Gb
S3 S4
S7
S6
S5a S5b S5
Evolved packet core
IASA
MME
UPE
PDN
GW
SAE
anchor
3GPP
anchor
Op. IP
serv. (IMS,
PSS, etc....)
SGSN GPRS core
PCRF
HSS
GERAN
UTRAN
non 3GPP
IP access WLAN 3GPP
IP access
Evolved
RAN
Figure 3. LTE-SAE bearer establishment.
Radio S1
E-UTRAN
S5/S8
EPC Internet
UE e-NB S-GW P-GW Peer entity
End-to-end service
EPS bearer External bearer
Radio bearer S1 bearer S5/S8 bearer
Gi
RENGARAJU LAYOUT_Layout 1 8/9/12 9:15 AM Page 91
IEEE Wireless Communications • August 2012
92
in UE, eNB, Serving-GW, and PDN-GW. The
EPS bearer QoS profile includes the parameters
QoS class identifier (QCI), Allocation and
Retention Priority (ARP), Guaranteed Bit Rate
(GBR), and Maximum Bit Rate (MBR). Each of
these is highlighted below:
• QCI: It is a scalar that is used as a reference
to control bearer level packet forwarding
treatments (scheduling, admission control,
etc.).
• ARP: The call admission control in the eNB
uses the ARP to decide whether a bearer
establishment or modification request is to be
accepted or rejected.
• GBR and non-GBR: Dedicated network
resources related to a GBR value associated
with the bearer are permanently allocated
when a bearer becomes established. On the
other hand, a non-GBR bearer may experi-
ence congestion-related packet losses. One
EPS default non-GBR bearer is established
when UE connects to the LTE network. A
dedicated bearer can either be a GBR or non-
GBR bearer.
• MBR: The maximum sustained traffic rate the
bearer may not exceed; only valid for GBR
bearers. The QCI mappings for different
applications are shown in Table 1.
An additional QoS attribute, Aggregate
MBR, is used to define the total amount of bit
rate of a group of non-GBR bearers.
Even though the WiMAX and LTE standards
provide strong QoS support, it is necessary to
ensure E2E resources to achieve an intended
QoE for real-time applications. Moreover, the
QoS architecture of the WiMAX standard only
focuses on access networks. For that, cross-layer
works are needed to assure the E2E resources.
The following Section describes some existing
cross-layer works that assure the E2E QoS.
EXTENDED CROSS-LAYER WORKS IN
WIMAX AND LTE TO SUPPORT THE
E2E QOS
As the WiMAX and LTE networks support the
integration with other networks, two types of
E2E QoS requirements are considered for the
analysis: homogeneous E2E QoS and heteroge-
neous E2E QoS.
HOMOGENEOUS E2E QOS
The WiMAX networks provide mechanisms for
QoS support at the MAC level, but E2E QoS
issues are not addressed in the standards. So the
cross-layer work is needed to satisfy both the
network-layer and MAC-layer QoS. A number
of enhancements have been proposed to enable
different levels of QoS in IP networks, including
Integrated Services (IntServ) and the Differenti-
ated Service (DiffServ) [3]. IntServ is imple-
mented by four components: the signaling
protocol (e.g. RSVP), the admission control, the
classifier, and the packet scheduler. Further-
more, some rules are prescribed to classify Diff-
Serv IP packets into different priority queues
based on QoS indication bits in the IP header.
Therefore, the QoS architecture of WiMAX
access networks can support both IntServ and
DiffServ. The mapping rules have been created
between the WiMAX MAC layer and the IP
layer.
In the traditional method, RSVP signaling
messages can be transmitted only in WiMAX
secondary MAC management messages, so the
E2E resource is reserved for each connection to
provide the network QoS. But in the cross-layer
approach, QoS mapping is created and the con-
nection request message, Dynamic Service Addi-
Figure 4. Traffic classification mapping for IntServ services [3].
PATH
Resv
Applications
Applications
PATH
Resv
PATH
Resv
SS-2
SS-1
DSA-REQ
DSA-RSP
DSA-REQ
DSA-RSP
Application
service
provider
Table 1. QCI mapping in LTE.
QCI Resource type Priority Application
1
GBR
2Conversational voice
2 4 Video streaming
3 3 Real-time gaming
4 5 Buffered streaming
5
Non-GBR
1IMS Signaling
6 6 Video & TCP based apps
7 7 Voice, video
8 8 Video & TCP based apps
RENGARAJU LAYOUT_Layout 1 8/9/12 9:15 AM Page 92
IEEE Wireless Communications • August 2012 93
tion (DSA) carries the RSVP-PATH specific
messages. If the BS supports both MAC and
network specific QoS, the BS accepts the con-
nection and informs the mobile user in the DSA
response message. The message exchanges for
the connections can carry the QoS parameters of
IntServ services for E2E resource reservation.
Figure 4 demonstrates the traffic classification
and mapping strategies for IntServ services. The
sender sends a PATH message including Traffic
Specifications (TSpec). The parameters such as
bandwidth, delay, and jitter can be easily mapped
into parameters in DSA messages. According to
the response of the DSA message, the provi-
sioned bandwidth is mapped into Reserved Spec-
ifications (RSpec) of the RESV message.
For DiffServ services, a number of Per-Hop
Behaviors (PHBs) for different classes of aggre-
gated traffic can be mapped into different con-
nections directly. Similar techniques can also be
implemented for LTE networks. In LTE net-
works, E2E application level QoS negotiation
and signaling can be implemented using IP Mul-
timedia Subsystem (IMS) network entities. The
IMS procedures for negotiating multimedia ses-
sion characteristics are specified by the 3GPP
and are based on IETF Session Initiation Proto-
col (SIP), Session Description Protocol (SDP),
and their extensions as required [5]. These QoS
mappings between wireless and wired networks
are important to assure the E2E QoS and ser-
vice assurances.
HETEROGENEOUS E2E QOS
In modern technologies, users should be able to
access their services through different access
technologies, such as WLAN, WiMAX, LTE,
UMTS, technologies from the same or different
network operators, and to seamlessly move
between different networks with active commu-
nications [4]. The recent standardization efforts
in the IETF for a new extensible IP signaling
protocol suite, NSIS describes the network layer
signaling to provide network QoS. The NSIS sig-
naling layer protocol is used to signal the appli-
cation QoS requirements and request the
resource allocation along the full path of data
flows. The NSIS protocol supports the interoper-
ability between different QoS-enabled domains
defining distinct QoS models depending on the
underlying network technologies. The QoS
model defines the QoS parameters, the traffic
descriptors and the methods to provide the
desired QoS through the Resource Management
Function (RMF) specification for the NSIS
nodes of a specific domain. So the recent
research efforts in WiMAX and LTE networks
are trying to integrate the WiMAX E2E archi-
tecture with NSIS signaling architecture and
LTE architecture with NSIS signaling architec-
ture.
The existing cross-layer work described in this
Section is used to assure the E2E resources. On
the other hand, the ISPs should monitor their
E2E network to verify the QoS and for service
verification. The following Section describes how
the ISPs can monitor their E2E network.
QOS AND QOE MONITORING IN
4G WIRELESS NETWORKS
Monitoring only the QoS for an E2E network is
not enough for the ISPs; they also have to moni-
tor their QoE support from gateways to the user
terminals to satisfy their customers. Depending
on the location of root causes, the number of
customers affected in a network will vary. If the
Figure 5. E2E QoS/QoE monitoring.
End
point
UE (user)
Video hub
office
ASN GW/
S-GW
CSN GW/
P-GW
Reference
media
QoE monitoring
tool Source
end
Medias - application
service provider
Access network
Access
network
Core network
Core network
End-to-end
QoE measurement probe/
end point
RENGARAJU LAYOUT_Layout 1 8/9/12 9:15 AM Page 93
IEEE Wireless Communications • August 2012
94
network QoS that is affected is close to the gate-
ways, the QoE of most of the customers will be
affected. Otherwise, the customers’ QoE of that
particular network segment will be affected.
Figure 5 shows the network architecture for
the QoS and QoE monitoring for 4G wireless
networks. As the QoE monitoring for an applica-
tion should start from the head end or possibly
the service provider’s gateway to the user termi-
nal, the QoE monitoring tool should be connect-
ed to the gateway of the service provider’s
network. In the network setup, the measuring
probes are integrated with various network ele-
ments from the gateway to the access network.
These probes are active probes, i.e. they can mon-
itor the network QoS periodically over the net-
work, and if there is any QoS degradation, they
will send an alarm signal to the Network Manage-
ment system (NMS) which is closely integrated
with the QoE monitoring tool. Upon receiving an
alarm signal, the NMS will initiate the QoE moni-
toring tool for the QoE measurement on the
affected segment. Depending on the type of traf-
fic handled on that segment, the QoE monitoring
tool measures the QoE based on the QoS mea-
surements for the No-Reference model or the
Reduced Reference model. If the measurements
are not sufficient to predict the QoE of a specific
application, the service provider can go further on
QoE measurement using the Full Reference
model. With the Full Reference model, the mea-
suring probes measure the QoS-KPI of known
applications and send it back the QoE monitoring
tool. After that, the QoE monitoring tool can pre-
dict the actual QoE for a specific application.
The QoE-KQIs at the application and service
layer, and the KPIs at the transport layer, and
the QoS at the network layer for different appli-
cations, e.g. VoIP, are shown in Fig. 6. The QoE
measurement using the Full Reference model
considers most of the KQIs for constructing the
QoE, as the model knows the original applica-
tion. On the other hand, the Reduced Reference
model considers some of the KQIs, and the No-
Reference model has no knowledge of applica-
tion-specific information; it constructs KQIs only
from the QoS-KPIs to predict the QoE. The
QoE monitoring for different applications is
explained as follows.
VOIP [6, 7]
The traditional call quality testing is only based
on subjective measurements like call availability,
quality, interruption, etc. The subjective QoE
measurement of voice quality defined in the ITU
P.800 is based on the MOS value. Recently, con-
siderable progress has been made on objective
measurements, and the various developments of
standards are:
1 The E-model (ITU G.107).
2 Full Reference model — Perceptual Evalua-
tion of Speech Quality (PESQ — ITU P.862).
3 No-Reference perceptual model — ITU-T
P.563.
Figure 6. QoE-KQI and QoS-KPI for different applications.
Same as video
KQIs +
video
buffering
Same as video
streaming
KQIs
R-factor
Speech distortion
Acoustic echo
Call quality
(SNR) Speech
latency Speech
loudness
PEVQ, frame rate,
pixilation, frozen
frames, skipped
frames, frame loss,
delay variance,
jerkiness, lip-sync,
PSNR, contrast,
color issues,
blurriness
Video streaming
KQIs +
number of
active sessions,
session
throughput
Video streaming
KQIs + channel
start time,
channel change
time, channel
change failure
Same as
video
streaming +
IGMP
signaling
Same as
video
streaming +
mostly SIP
signaling
QoS indicators - throughput, delay and jitter
Call setup success
ratio, call setup
time, call cut-off
ratio, reliability,
availability,
billing
Round trip delay,
jitter,
pkt. loss /
variance,
packet
sequencing SIP
signaling
PID error,
reliability,
service accessibility,
comfort
Round trip delay,
jitter,
pkt. loss / variance,
packet sequencing
IMS SIP signaling,
ISUP/SS7 and ISUP/
SCTP/IP
Video via HTTPVoIP Video streaming IPTV
QoE - KQIs (key quality indicators)
Appllication layerService layer
Transport layer
key performance
indicator (KPIs)
Network
The traditional call
quality testing is
only based on sub-
jective measure-
ments like call
availability, quality,
interruption, etc. The
subjective QoE mea-
surement of voice
quality defined in the
ITU P.800 is based
on the MOS value.
RENGARAJU LAYOUT_Layout 1 8/9/12 9:15 AM Page 94
IEEE Wireless Communications • August 2012 95
4 Speech Processing, Transmission and Quality
Aspects (ETSI TS 102-250).
The E-model, defined in the ITU-T G.107,
analyzes the QoS parameters and quantifies the
voice quality by calculating an Rfactor between
0 (worst) and 100 (excellent). The expression for
the Rfactor (Eq. 1) represents the sum of all
degradation factors in the communication:
R= (Ro– Is) – Id– Ie,eff + A1)
The Signal-to-Noise Ratio [Ro] represents the
subjective quality impairment due to circuit
noise, room noise at sending and receiving sides,
and subscriber line noise. The default value
given in ITU-T Rec. G.107 is 94.77.
The simultaneous impairment factor [Is] rep-
resents the subjective quality impairments due to
loudness, side tone, and quantization distortion.
The delay impairment factor [Id] represents
the subjective quality impairments due to echo
and absolute delay.
The equipment impairment factor [Ie,eff] rep-
resents the subjective quality impairments due to
low bit rate CODEC, packet/cell loss, etc.
The advantage factor [A] allows for compen-
sation of impairment factors to consider advan-
tages of access to the user, e.g. mobile terminals.
Some cost effective open source solutions use
PESQ test methodology to measure the MOS
value. In this test procedure, the SIP protocol is
used for voice transmissions. For the evaluation
of the speech quality, the PESQ algorithm com-
pares the local reference WAV file with the
recorded WAV from the Asterisk (SIP) server.
VIDEO STREAMING AND IPTV [7, 8, 10]
In a broadcast system, each single video program
is described by a Program Map Table which has
a unique Program ID (PID). For instance, a
transport stream used in digital television might
contain three programs to represent three televi-
sion channels. A receiver wishing to decode a
particular “channel” has to decode the payloads
of each PID associated with its program. The
receiver can discard the contents of all other
PIDs.
With respect to the monitoring, the impair-
ments of video can happen at various locations: at
the video head end, some of the common prob-
lems coming from the video provider are improp-
er PID mappings or data table mappings in the
video source, as well as lip synch issues or band-
width configuration issues caused by the
transcoder. After the traffic arrives at the gateway,
these impairments can be monitored at the service
provider’s gateway. The other network impair-
ments can be monitored from different locations
that are shown in Fig. 5. The various standards for
measuring video QoE are given in Table II.
The most accurate approach to evaluating
perceived video quality is the subjective assess-
ment by humans. But the assessment may be dif-
ferent for different people. On the other hand,
the Full Reference model generates closer accu-
racy to the subjective measurements. The Full
Reference model is computed separately within
the context of plane, edge and texture regions of
the scenes. The different measures on the origi-
nal signal in comparison with the reference sig-
nal are Peak SNR (PSNR), Mean Square Error
(MSE), Positive Sobel Difference (PSD), Nega-
tive SD (NSD), and Absolute SD (ASD). These
measures are finally converted into a MOS-V
score. The Video Quality Metric (VQM) soft-
ware, developed by the Institute for Telecommu-
nication Science, follows the ITU-T J.144 and
ITU-R BT.1683 standards. It has high correla-
tion with subjective video quality scores.
IPTV is a system where a digital television
service is delivered by using IP packets over a
broadband network infrastructure. The most
suitable method of QoE assessment for IPTV is
based on the No-Reference model that are Para-
metric models for Non-intrusive Assessment of
Multimedia Streaming (P.NAMS) and Media
Delivery Index (MDI). P.NAMS uses informa-
tion from packet headers (RTP and MPEG2
Transport Stream headers), buffering informa-
tion, information about codecs and encoded
bitrates. P.NAMS does not use any payload
information. The MDI factor indicates the frame
loss, latency, jitter, delay, and channel change
times problems generated on the network inde-
pendent of video encoding. The recommended
maximum acceptable value of Delay Factor (DF)
is 9 – 50 ms; the average media loss rate (MLR)
for HDTV is 0.00005, while for SDTV and VoD,
the MLR is 0.0004 [8].
CONCLUSIONS
This article presented an integrated view of E2E
QoS support in WiMAX and LTE networks, and
how the QoS and QoE can be monitored by the
ISPs. The E2E QoS support includes the existing
QoS and service assurance supports in WiMAX
and LTE networks, and the cross-layer works to
assure the E2E QoS before establishing the
applications. The QoS and QoE monitoring Sec-
tion advocate the need to monitor E2E QoS and
QoE from the gateway to user terminals. In the
same Section, the effective E2E QoS and QoE
monitoring on 4G wireless networks at various
possible locations were described. Here, the
same QoE models can also be applied for other
3GPP wireless networks. The measuring probe
monitors the QoS of the network and, if the net-
work QoS degrades, sends an alarm signal to
NMS. The NMS then initiates the QoE monitor-
ing tool to measure QoE. This approach is use-
ful for the service provider to take a preventive
action and to ensure service assurance.
Table 2. Video QoE standards.
Methodology
Image Resolution
HDTV and SDTV VGA, CIF and QCIF
Subjective ITU-R BT.500, ITU-T J.140 and
ITU-T J.24 ITU-T P.910
Full Reference ITU-T J.144 and ITU-R BT.1683 ITU-T J.247
Reduced Reference ITU-T J.144 and ITU-R BT.1683 ITU-T J.247
No-Reference ITU-T SG12
RENGARAJU LAYOUT_Layout 1 8/9/12 9:15 AM Page 95
IEEE Wireless Communications • August 2012
96
REFERENCES
[1] WiMAX Forum: WiMAX End-to-End Network Systems
Architecture (Stage 3: Detailed Protocols and Proce-
dures), 2008.
[2] 3GPP TS 23.401 v. 8.8.0, “General Packet Radio Service
Enhancements for Evolved Universal Terrestrial Radio
Access Network Access,” 2009.
[3] J. Chen et al., “An Integrated QoS Control Architecture
for IEEE 802.16 Broadband Wireless Access systems,”
Proc. IEEE GLOBCOM’05, Dec. 2005.
[4] S. Sargento et al., “Context-Aware End-to-End QoS
Architecture in Multi-Technology Multi-Interface Envi-
ronments,” Proc. 16th IST Mob. and Wireless Comm.
Summit, July 2007.
[5] S. Choi et al., “A Study on End-to-End QoS Provision for
Multimedia Services in Beyond 3G Convergence Net-
works,” Proc. IEEE VTC’07F, Oct. 2007.
[6] ITU-T, Recommendation G. 107, “The E-Model, A Com-
putational Model for Use in Transmission Planning,”
2002.
[7] DSL Forum, “Triple-play Services Quality of Experience
(QoE) Requirements” Technical Report TR-126, 2006.
[8] Agilent Technologies, “IPTV QoE: Understanding Inter-
preting MDI Values,”
http://cp.literature.agilent.com/litweb/pdf/5989-
5088EN.pdf
[9] F. Kuipers et al., “Techniques for Measuring Quality of
Experience,” Proc. Int’l. Conf. Wired/Wireless Internet
Commun., June 2010.
[10] ITU-T Recommendation J.144R1, “Objective Perceptual
Video Quality Measurement Techniques for Digital
Cable Television in the Presence of A Full Reference,”
2007.
BIOGRAPHIES
PERUMALRAJA RENGARAJU (rpraja@sce.carleton.ca) received his
B.Eng. degree from Bharathidasan University and the
M.Eng. degree in Communication Systems from Anna Uni-
versity, India, in 2002 and 2006. He is currently working
toward the Ph.D., degree in the Department of Systems
and Computer Engineering, Carleton University, Canada.
He was with CDOT-Alcatel research centre in 2006–2007
and with nGIN Technologies in 2007–2009, where he
worked on the research and development of WiMAX tech-
nology. His current research interests include QoS and
Security in 4G wireless networks.
CHUNG-HORNG LUNG (chlung@sce.carleton.ca) received the
B.S. degree in Computer Science and Engineering from
Chung-Yuan University, Taiwan and the M.S. and Ph.D.
degrees in Computer Science and Engineering from Ari-
zona State University. He was with Nortel Networks from
1995 to 2001. In September 2001, he joined the Depart-
ment of Systems and Computer Engineering, Carleton Uni-
versity, Canada, where he is now an associate professor.
His research interests include: Communication Networks,
Wireless Ad Hoc/Sensor Networks, and Software Engineer-
ing.
ANAND SRINIVASAN (anand@eion.com) received his Ph.D.,
and M.Sc. in computer science from the University of Vic-
toria, British Columbia, Canada. He also holds a Master’s
degree in Computing from JNU, New Delhi, and a Bache-
lor’s degree from the University of Delhi, India. He has over
15 years of experience in the system and network design,
and performance for large scale wired, wireless, and satel-
lite networks. He is at present the Vice President for Tech-
nology and Product management in EION Wireless. In
addition to his work with EION, he is an Adjunct Research
Professor in the Department of System and Computer Engi-
neering at Carleton University.
F. RICHARD YU(richard_yu@sce.carleton.ca) received the
Ph.D. degree from the University of British Columbia
(UBC) in 2003. From 2002 to 2004, he was with Ericsson
in Lund, Sweden. From 2005 to 2006, he was with a
start-up in California, USA. He joined Carleton University
in 2007, where he is currently an Associate Professor. He
received the Carleton Research Achievement Award in
2012, the Ontario Early Researcher Award in 2011, and a
number of other awards. He serves on the editorial
boards of several journals and the TPC of numerous con-
ferences.
The most accurate
approach to evaluat-
ing perceived video
quality is the subjec-
tive assessment by
humans. But the
assessment may be
different for different
people. On the other
hand, the Full Refer-
ence model gener-
ates closer accuracy
to the subjective
measurements.
RENGARAJU LAYOUT_Layout 1 8/9/12 9:15 AM Page 96
