This document summarizes a research paper on optimizing quality of service (QoS) and analyzing performance in Next Generation Networks (NGN). It discusses how applying QoS technologies from Huawei can improve capacity and QoS in NGN. The optimized NGN is simulated using different codecs and traffic loads. Simulation results show that G.711 provides better voice quality while G.729 allows higher capacity. Analysis of the simulation matches theoretical models. Overall, the optimized NGN can provide both high voice quality and capacity through adaptive use of codecs.

1. QoS Optimization and Performance Analysis of NGN
Umma Hany1
, A. B. M. Siddique Hossain2
, Pran Kanai Saha3
1
The University of Asia Pacific, Dhaka, Bangladesh
ummahany@gmail.com
2
American International University of Bangladesh, Dhaka, Bangladesh
siddique@aiub.edu
3
Bangladesh University of Engineering and Technology, Dhaka, Bangladesh
sahapk@eee.buet.ac.bd
Abstract---- Next Generation Network (NGN) is a packet based
network which can support the expansion of broadband and
introduction of triple play (Voice +Data +Video) over fixed and
mobile line. Before deployment of NGN, the capacity and Quality
of Service (QoS) over NGN must be ensured. In this paper, the
technologies proposed by Huawei have been applied to ensure
the capacity and QoS over NGN. Then the optimized NGN is
simulated to observe the effect of the quality factors on the voice
performance. The simulation results are then analyzed and
verified by theoretical analysis to determine the maximum
capacity and QoS which is possible to achieve over the proposed
NGN. The simulation results show a good agreement with that of
the theoretical results.
Keywords: NGN, QoS, Voice on internet protocol.
I. INTRODUCTION
QoS can be ensured by reducing the link failures and the
deterioration of quality due to overload. In Next Generation IP
network, IP lacks the traffic engineering (TE) mechanisms
capable of offering a differentiated QoS, while efficiently
allocating the network resource. Factors affecting the QoS
over NGN are speech encoders, delay, jitter, packet loss and
echo. The capacity of NGN is highly dependent on the chosen
speech codec. There are abundant of activities in developing
protocols, speech encoders and optimization services to ensure
the capacity and QoS over NGN. ITU-T developed and
standardized a series of audio codec [1]-[3] ranging in bit rates
from 5.3-64 kbps. Multi-Protocol-Level-Switching (MPLS)
[4] offers a way of incorporating TE mechanisms into IP. The
integration of differentiated services (DiffServ) [5] with
MPLS guarantees the QoS for a broad range of multiservice
traffic. ITU-T defines the QoS control NGN architecture [6] in
which service-related functions is independent of the
transport-related technologies. Hence, the transport stratum is
responsible for admission and resource control based on the
network policy and the resource availability. The service
control function (SCF) is responsible for the application
signaling for the service setup. The open issue of the ITU-T
QoS control is that it is on per-call basis. Thus, the QoS
control in the core network will be a burden. The reliability
and security in the core network is another issue. The QoS
control mechanism can be simplified by using the
performance monitoring information in the core network.
Huawei proposed a service-independent QoS control NGN
architecture [7] in which the network is separated in layers
(service, call control, bearer control) and is based on a unique
packet switched core network for all types of access networks,
services and terminals. Here, QoS assurance technologies are
incorporated in each layer to ensure the QoS [7], [8] enabling
the network to provide efficiency, reliability and security in
the core network. In this paper, the capacity and QoS over
NGN has been optimized by applying the developed
technologies. Then the network is simulated and analyzed to
determine the maximum capacity and QoS over NGN.
II. SIMULATION AND RESULTS
A. Simulation
The QoS control technologies are applied on the following
NGN network of Banglaphone as shown in Fig. 1. Then the
voice over NGN is simulated using the “QoS Analyzer” tool
of Huawei. Five simulation agents have been set for five VoIP
nodes. During simulation, two different codec is used and
incremented throughput test is done through generating traffic
flow with changed packet size. The highest traffic flow is
carried out from 16:00:00 to 18:40:00.
Dotted lines indicate signaling path and solid lines indicate voice path.
Fig. 1 Next Generation Network topology with 5 VoIP nodes
Hence, the Softswitch [7] in the call control layer provide
guaranteed QoS and network efficiency with real-time QoS
monitoring and dynamic flow control (using different codecs
and rejecting high-bandwidth applications). The Universal
Media Gateway (UMG) in the bearer control layer improves
the bandwidth utilization and reduces the effect of the quality
factors by adopting Advanced Voice Quality Assurance
technologies [7]. It is achieved by adopting different
encoding/decoding, Voice Activity Detection (VAD),
Comfort Noise Generation (CNG), dynamic buffering, lost-
packet compensation and echo cancellation technology.
Session Border Controller (SBC) [8] at the edge of the IP
network is used to solve the IMS (IP Multimedia Subsystem)
technical problems as Network security, NAT traversal, QoS
assurance of packets and media streams.
364
6th International Conference on Electrical and Computer Engineering
ICECE 2010, 18-20 December 2010, Dhaka, Bangladesh
978-1-4244-6279-7/10/$26.00 ©2010 IEEE

2. B. Results
1) Packet Loss Ratio
Fig. 2 and 3 show that packet loss using G.729 is more than G.711.
Fig. 2 G.711 Loss Time Graph (on Oct 29, 2008 3:10 PM-10:10 PM)
Fig. 3 G.729 Loss Time Graph (on Oct 30, 2008 2:00 PM to 10:00 PM)
2) Delay
Fig. 4 and 5 show that delay for G.729 is more than G.711.
Fig. 4 G.711 Delay Time Graph (on Oct 29, 2008 3:10 PM-10:10 PM)
Fig. 5 G.729 Delay Time Graph (on Oct 30, 2008 2:00 PM to 10:00 PM)
3) Jitter
Fig. 6 and 7 show that jitter delay for G.729 is more than G.711.
Fig. 6 G.711 Jitter Time Graph (on Oct 29, 2008 3:10 PM-10:10 PM)
Fig. 7 G.729 Jitter Time Graph (on Oct 30, 2008 2:00 PM to 10:00 PM)
4) MOS
Fig. 8 and 9 show that MOS using G.711 is better than G.729.
Fig. 8 G.711 MOS Time Graph (on Oct 29, 2008 3:10 PM-10:10 PM)
Fig. 9 G.729 MOS Time Graph (on Oct 30, 2008 2:00 PM to 10:00 PM)
365

3. 5) Throughput
Fig. 10 and 11 show that the throughput using G.711 is greater than
G.729
Fig. 10 G.711 Throughput Time Graph
Fig. 11 G.729 Throughput Time Graph
III. RESULTS ANALYSIS AND VERIFICATION
A. QoS Analysis
1) Theoretical Analysis
Factors affecting the QoS are encoding mode, network Delay
(ms), Jitter, Packet loss, packet doubling and echo. Different
parameters and equipment impairment factors of the VoIP
codecs are illustrated in Table 1. The Table 2 presents the
equipment impairment factor considering packet loss which is
defined as follows:
Ief = Ie + 30ln(1+ 15e) (1)
Ief: it is related to packet loss, e= packet loss ratio
TABLE I AUDIO CODEC PARAMETERS OF VoIP CODECS [1]-[3],[9]
Para-
meters
Bit
rate
(Kbps)
Framing
interval
(ms)
Payload
(Bytes)
Packets
/s, Np
Equipment
Impairment
Factor, Ie
G.711 64 20 160 50 0
G. 723.1 6.3 30 24 33 15
G.729 8 20 20 50 10 or 11
TABLE II EQIPMENT IMPAIRMENT FACTORS FOR DIFFERENT
CODEC CONSIDERING PACKET LOSS [9]
%packet loss 0 0.5 1 1.5 2 3 4 8 16
G.729a 11 13 15 17 19 23 26 36 49
G.723.1a 15 17 19 22 24 27 32 41 55
The delay impairment factor can be defined as follows:
Id= 0.024d + 0.11 (d-177.3) H (d-177.3) (2)
Id: it is related to end to end delay
d=one-way delay (coding + network + de-jitter delay) [ms]
H(x) =0 for x<0 H(x) =1 for x > 0
The E-model [10] calculates the R from the network QoS
factors. The rating R is computed as follows:
R = R0 – Icodec - Idelay - Ipdv – Ipacketloss (3)
R = 93.2-Id-Ief (4)
Mean opinion score (MOS) [11] is to evaluate the voice
quality according to the scoring standards of ITU-T. MOS is
calculated from R as follows. Table 3 illustrates the ITU-T
voice quality at different network conditions.
MOS = 1 < 1 + (0.035R) + (R(R – 60) (100 – R) 7.0e-06
) < 4.5 (5)
TABLE III ITU-T STANDARD VOICE QUALITY OVER NGN
Parameters and
Service
Good Poor Bad
ITU-T MOS 4.0-5 3.5-4 3-3.5 1.5-3 0-1.5
Standard
Delay ≤40ms ≤100ms ≤400ms
Loss ≤0.1% ≤1% ≤5%
Jitter ≤10ms ≤20ms ≤60ms
Voice
G.711 Excellent Good Fair
G.729 Good Good Poor
G.723.1 Good Almost Good Fair
2) Simulation results analysis
MOS is calculated by replacing the simulation results (delay,
jitter and packet loss) into equation (1)-(5) and compared to
the ITU-T standard to evaluate the voice quality as in Table 4.
TABLE IV SIMULATION RESULTS ANALYSIS TO EVALUATE QoSConsideringthe
packetlossbetween
Agent3–Agent4at
17:35
Simulation results
E-Model
analysis ITU-T
standard
Voice
Quality
Delay
(ms)
Jittter
(ms)
Packet
Loss
(%)
R-
Value
MOS
G.711 6.8 0.3 0 93 4.4 Excellent
G.729 12 0.25 0.17 81 4 Good
Using
aggregated
simulation
results
G.711 6.5 0.3 0 93 4.4 Excellent
G.729 7.5 0.3 0.09 81.6 4 Good
The simulation result analysis shows that due to the QoS
optimization, the effect of the voice quality factors are
reduced and NGN provides good voice quality which is not
significantly degraded due to the utilization of different
codecs.
B. Capacity Analysis
1) Theoretical analysis
Each VoIP packet includes the headers at the various protocol
layers such RTP 12 bytes, UDP 8 bytes, IP 20 bytes, Ethernet
26 bytes and the payload comprising the encoded speech for a
certain duration depends on the codec deployed.
OHhdr = HRTP+ HUDP+ HIP +HMAC (6)
Packet length = OHhdr + payload (7)
Payload = number of payload bits/s Framing Interval (s) (8)
Bandwidth = packet length number of packets per sec (9)
Let n be the maximum number of sessions which is supported
by NGN. n is defined as follows:
n =
Data Rate
Bandwidth Occupied
(10)
Using equation (6)-(9) we can calculate the bandwidth
occupied by various Kinds of Voice Encoding/Decoding as
illustrated in Table 5.
TABLE V BANDWIDTH OCCUPIED USING DIFFERENT CODEC
Parameters G.711 G. 723.1 G.729
Framing interval (ms) 20 30 20
Bandwidth occupied (kbps) 89.78 22.49 33.78
366

4. Peak Hour
Bit Rate
Peak Hour
The capacity using different codec has been calculated using
equation (10) which has been plotted in Fig. 12. Fig. 12 shows
that maximum capacity can be achieved using G.723.1 codec.
Thus NGN provides high capacity using different codec. Fig.
13 shows the comparison of TDM and NGN capacity.
Fig. 12 Maximum simultaneous VoIP nodes
Fig. 13 Comparison of TDM and VoIP capacity for 2 Mbps data rate
2) Simulation (Throughput) analysis
Throughput is the amount of data in bits that is transmitted
over the channel per unit time. VoIP capacity can be
calculated by replacing the maximum throughput (simulation
result for 1 Mbps data channel) into equation (10) as follows:
VoIP Capacity, n =
Maximum Throughput
Bandwidth Occupied
(11)
The simulation results show that using G.711, the average
throughput at the peak hours is 0.92 and using G.729 it is
0.856 Mbps. Replacing the average throughput and the
bandwidth occupied into Equation (11) we get, n =10 for
G.711 and n = 25 for G.729. The analysis shows that as the
bandwidth utilization has been improved using different
codec, NGN is able to provide better capacity.
IV. TRAFFIC MEASUREMENT REPORT ANALYSIS
Analyzing the “Traffic measurement report” of TDM based
DTCL and NGN based Teletalk, the call connected ratio at the
peak hours has been plotted in Fig. 14 and Fig. 15.
Fig. 14 Call connected Ratio diagram at peak hour for TDM based network
Fig. 15 Call Connected Ratio diagram at peak hour for IP based network
The call connected ratio is defined as follows:
Call Connected Ratio =
Number of call attempts
Number of call connected
X 100 (12)
Comparing the plots in Fig. 14 and 15, it is observed that at
the peak hour call connected ratio for DTCL is 59.5%-65%
and the call connected ratio for Teletalk is 96.5%-98.5%. Thus
Call connected ratio of IP based NGN is far better than TDM
based circuit switched network.
V. CONCLUSION
The Optimized NGN improves the bandwidth utilization and
reduces the affect of the quality factors using the developed
technologies. Analyzing the simulation results for G.711 and
G.729 codec, it is observed that better capacity is achieved
using G.729. Thus, optimized NGN is able to provide high
VoIP capacity using different codecs (ranging in bit rates from
5.3-64 kbps) as per the capacity requirement. The result
analysis also shows that excellent voice quality is achieved
using G.711 codec where good voice quality along with high
capacity is achieved using G.729 codec. Thus optimized NGN
is able to provide high voice quality which is not significantly
degraded due to the chosen codec and other quality factors
(delay, jitter and packet loss). Thus it is observed by analyzing
the results that high VoIP capacity and good voice quality is
possible to achieve over the optimized NGN. For the new
services as fixed mobile convergence and IPTV, QoS control
for mobility and the multicast condition must be developed.
Acknowledgement: Authors would like to acknowledge for using
the resources and support provided by Huawei Technologies Ltd,
Banglaphone Ltd, Dhaka Telecom Co. Ltd, Teletalk Bangladesh Ltd.
and the department of EEE, BUET.
REFERENCES
[1] International Telecommunication Union (ITU), “ITU-T
recommendation G.711.1: Wideband embedded extension for ITU-T
G.711 pulse code modulation”, ITU, March, 2008.
[2] International Telecommunication Union (ITU), “ITU-T
recommendation G.723.1: Dual rate speech coder for multimedia
communications transmitting at 5.3 and 6.3 kbit/s”, ITU, May, 2006.
[3] International Telecommunication Union (ITU), “ITU-T
recommendation G.729: Coding of speech at 8 kbit/s using conjugate-
structure algebraic-code-excited linear prediction (CS-ACELP)”, ITU,
Jan, 2007
[4] K. Elsayed and L. Toutain, "MPLS: The Magic Behind the Myths",
IEEE Communications Magazine, January 2000.
[5] S. Blake et al., “An architecture for Differentiated Services,” IETF
RFC 2475, Dec. 1998.
[6] Song, J., Chang, M. Y., Lee, S. S.,”Overview of ITU-T NGN QoS
Control”, IEEE Communications Magazine, 8th
Oct, 2007.
[7] Huawei, “Huawei U-SYS NGN solution”, www.huawei.com, July
2008.
[8] Huawei, “Quidway SessionEngine2300-Feature Description (for
IMS)”, support.huawei.com, 20th
Jan. 2009.
[9] International Telecommunication Union (ITU), “ITU-T
recommendation G.113 Appendix I: Provisional planning values for
the equipment impairment factor Ie”, ITU, 3rd
Dec, 1998
[10] International Telecommunication Union (ITU), “ITU-T
Recommendation G.107. 2003, The E-model, a computational model
for use in transmission planning”, ITU, 30th
Sep, 2003
[11] International Telecommunication Union (ITU), “ITU-T
recommendation G.109: Definition of categories of speech
transmission quality", ITU, 30th
Sep, 1999.
367