Birendra Kumar et al, Int.

Journal of Information Technology & Mechanical Engineering - IJITME,

Generate VOIP Signal over

M.Tech Final Year Student Deptt. of ECE,MMEC, MMU, Mullana Ambala, Haryana, India
[email protected]
2
Associate Professor, Deptt. of ECE, MMEC, MMU, Mullana Ambala, Haryana, India
[email protected]

Abstract- The Aim of this paper is to give a basic introduction on VOIP and describe QOS
problems. To get the best quality in real time. To investigate the performance of speech over VoIP
using commercial 3G wireless network.More specifically, an investigation of IP-based voice
communication with emphasis on the effects of a wireless channel and its conditions on the
quality of the received speech will be investigated. Simulation will be implemented using Matlab
software, and the experiments will be conducted on commercial VoIP networks. The attention is
focused on the quality of service in wireless network.
Keywords: MATLAB, VOIP, 3G, VOICE QUALITY.

I.

INTRODUCTION

A change from analog to digital telephony has well been established in most countries worldwide. Second, there is an
increasing trend towards the use of internet telephony, also known as Voice over Internet Protocol (VoIP). VoIP is the
transmission of voice using a packet switched network, which is usually based on the transmission control
protocol/internet protocol (TCP/IP) suite. First of all, voice transmission over the Internet can be cheaper than that over
traditional telephone networks. Second, it provides a handful of new opportunities and applications to its users; these new
features are almost impossible to implement without the use of a packet switched network. There are many different
ways in which two or more users can be connected to a VoIP network, but the main concept of inter connect both PSTN
and VoIP are used.

13
2016, IJITME All Rights Reserved, www.ijitme.org

Birendra Kumar et al, Int. Journal of Information Technology & Mechanical Engineering - IJITME,
Vol.2 Issue. 6, June- 2016, pg. 13-23
ISSN: 2349-2865

Speech signal

Speech
signal

Analog
to Digital

Speech
compressi
on

Digital to
analog

Packetiza
tion

Speech
decompre
ssion

depacketizat
ion

Transmitter

receiver

Figure 1.1: Basic VoIP communication system [1].

II.

Key challenges for VOIP over 3G wireless networks

Packet delay: Packet delay is of two types: handling and propagation. Handling (packetization) delay is the amount of
time it takes for a speech signal to be processed by the computers hardware before it is transmitted to the medium.
Propagation delay is the amount of time it takes a signal to travel from transmitter to receiver and is dependent upon the
medium used.
Packet delay variation: Packet delay variation or jitter is the variation in delay between consecutive packet inter-arrival
times. It is an effect of packet switched networks and it can be more annoying than the packet delay itself since its effects
vary over time.
Packet loss: Another problem of most packet switched networks is packet loss. It can happen at any point in the network
especially in media that are prone to errors like a wireless medium.
Attenuation and noise: In order to fully recover the signal at the receivers end, two conditions must be met. The signal
strength must be sufficient to be detected by the receivers circuitry and must be higher than noise.
Multipath: In the case of mobile wireless networks, a loss model that accounts for multiple copies of the same signal due
to multipath effects must be considered. The main sources of multipath are diffraction, reflection and scattering. These
multiple copies have varying delays and, in some cases, they might be the only signals received, e.g., in the case of non
line of sight (NLOS) reception.

14
2016, IJITME All Rights Reserved, www.ijitme.org

Birendra Kumar et al, Int. Journal of Information Technology & Mechanical Engineering - IJITME,
Vol.2 Issue. 6, June- 2016, pg. 13-23
ISSN: 2349-2865

Figure 3: Schematic of multipath channel where the signal follows a direct, a reflected and a diffracted path.
VoIP availability: For many years the PSTN companies have been advertising their four 'nines' of availability as their
main advantage over VoIP and cellular telephony. What they really mean is that their network (from PBX to PBX)
availability is close to 99.99%; if the availability of home appliances has to be accounted for, the total availability would
be much lower. In order for VoIP to fully replace the traditional telephony, it has to get closer to its competitors level of
availability. Availability, is given by

III.

Channel

A) Linear predictive coding: In linear predictive coding of speech, the source is represented by the
voiced or unvoiced excitations. In this approach, a linear filter is used to model the vocal tract. The input to
this filter is a random noise or a periodic pulse, depending on whether the excitation is voiced or unvoiced.
The filter can be analytically represented as:

3.1

Where,

is the excitation (also known as prediction error),

is the filter order, is the filter gain and

are

the filter coefficients. The linear predictive filter coefficients are obtained by minimizing the prediction error
power with respect to the filter coefficients

, which is given by
3.2

[ ]

[(

) ]

Where, [ ] is the expectation operator.In practice, in order to compress a segment of speech, it is first divided

15
2016, IJITME All Rights Reserved, www.ijitme.org

Birendra Kumar et al, Int. Journal of Information Technology & Mechanical Engineering - IJITME,
Vol.2 Issue. 6, June- 2016, pg. 13-23
ISSN: 2349-2865
into smaller segments. This segment is then classified as voiced or unvoiced based on its energy and
frequency contents. These parameters are transmitted after obtaining the pitch and vocal tract filter
coefficients. At receiver, the received parameters are then used to reproduce the voice. LPC is preferably used
in applications where compression ratios are of most importance.

B) Code Excited Linear Prediction (CELP): CELP methods improve the voice quality by using
better excitation techniques compared to LPC. The output of the filter is given as
3.3

Where, is the filter gain and

are the filter coefficients. P is the fundamental harmonic period, also known

as pitchand is a scaling factor. The pitch periodicity contribution is

and is calculated every subframe.

Equation 3.3 can be treated as a cascade of two filters. The first filter extracts the pitch, and the second is a
long term formant filter. The excitation is created using the codebook approach so that it is not necessary to
extract voicing patterns or pitch. The codebook is generated offline. In this approach, the synthesized outputs
are compared every time with the predetermined codebook to find the best match. An example of CELP is
shown in Figure 3.2. In this example, the first codebook is called stochastic codebook. This codebook is fixed
and predetermined; however, the second is adaptive. The excitation of each segment is the sum of adaptive
and stochastic codebook outputs. After excitation [n]is produced, a copy of it is fed back to the adaptive
codebook which then adapts to the current segment. In order for the scheme to provide minimum error
between the input and synthesized speech, the codebook indices are scaled using gains.

Figure 3.2: Block diagram of CELP featuring two codebooks: Stochastic and Adaptive

16
2016, IJITME All Rights Reserved, www.ijitme.org

Birendra Kumar et al, Int. Journal of Information Technology & Mechanical Engineering - IJITME,
Vol.2 Issue. 6, June- 2016, pg. 13-23
ISSN: 2349-2865

Fadings:Fading channel has used two types of channel, first are Rayleigh channel and second are rician channels. Rayleigh
channels no paths are consider dominant,in Rayleigh distribute signals have to be different path have different phases and
similar signal strength have receive into produced Rayleigh distribute channel, Rayleigh channel have to be probability
density function are shown as
(

Where are average receive power and r are signals magnitude, in the case of racian channel distributions have to be
dominants line of sight,signals are presents, resultings amplitudes have be modulated with the rician probability density
function,they are given below
(

I0 have two zero orders magnitudes Bessel functions, K are ratio of dominant paths power into remaining paths power,
dominant path power are used to the line of sight, when in addition case if we put k= then the channel has becomes to
an additive white Gaussian noise. In other hand if we put k=0 then the channel has becomes to Rayleigh
channels.measured of fading channel can be used for frequency and time dispersions are obtain into coherence time and
coherence bandwidth, where Tc are the coherence time the decorrelate have two time domains sample are given by

Where Fd are Doppler shifts, Fc are the frequency carrier, Fd as shown in

is the angle of radiations and relative motions, c has speed of light. V has the relative velocity of transmitters

and receivers .:
3.13
Where, is the rms delay spread of the channel.Depending on the amount of time delay spread, channels can be
classified as flat or frequency selective; and depending on the amount of Doppler spread, they can be classified as fast or
slow fading. A discussion on wireless standards is presented below.

IV.

RESULTS

In Figure 4.1, the speech is first recorded through microphones. It is recorded for two sentences by author for three
different intervals and are shown in Figures 4.2 and 4.3. The samples contains two times repetition of the sentences, so in
total are 12 samples. The two sentences are:

17
2016, IJITME All Rights Reserved, www.ijitme.org

Birendra Kumar et al, Int. Journal of Information Technology & Mechanical Engineering - IJITME,
Vol.2 Issue. 6, June- 2016, pg. 13-23
ISSN: 2349-2865

Sentence 1: VoIP are transmission of voice used to a packet switched network.

Sentence 2: There are based on to the transmission control protocols and internet protocols.

Figure 4.2:Speech samples of author corresponding to sentence 1.

Figure 4.3:Speech samples of author corresponding to sentence 2.

18
2016, IJITME All Rights Reserved, www.ijitme.org

Birendra Kumar et al, Int. Journal of Information Technology & Mechanical Engineering - IJITME,
Vol.2 Issue. 6, June- 2016, pg. 13-23
ISSN: 2349-2865

The recorded speech is transmitted through a network which includes at least one wireless link along the end to end path.
This causes distortion to the speech sample due to packet errors, which occurs due to fading caused by wireless channels.
The speech is received at the other end and is distorted and can be completely faded. To estimate the quality of the
speech, the received speech is compared with the original speech through bit error rate, as follows:
(4.1)
VOIP network was simulated in Matlab.digitize, compress, packetize and transmit into receiver, it can be follow to the
reverse procedure. The Rician and Rayleigh fading channels are simulated in Matlab, and their characteristics is shown in
Figure 4.4 and 4.5 respectively.

[a]

[b]

[c]

Fig. 4.4:- The characteristics of Rician channel

19
2016, IJITME All Rights Reserved, www.ijitme.org

Birendra Kumar et al, Int. Journal of Information Technology & Mechanical Engineering - IJITME,
Vol.2 Issue. 6, June- 2016, pg. 13-23
ISSN: 2349-2865

[a]

[b]

[c]
Fig.4.5:- The characteristics of Rayleigh channel.

The Rayleigh fadings channels and Rician channel fadings channel were implemented into Matlab with used to AWGN,
addive white Gaussian noise AWGN with convolutional coding. The recorded signals in Figures 4.2 and 4.3 is passed
through these channels and the received signal is compared with the input signal using BER. The BER performance of
AWGN, Rician and Rayleigh fading channels is shown in Figure 4.6. For a particular value of EbNo, the Rayleigh
fading channel gives highest BER and the AWGN channel is the lowest.
The effect of K factor of Rician channel is shown in Figure 4.7.if K factor are increasing then the BER are also decrease.
At the value close to 0.06.If we put K=0 and down to 10-5 as the value of K factor reaches at 25.In addition case if we put
k=0 then the channel becomes to relay channel, If we put K= then the channel becomes to additive white Gaussian
noise channels, we use the BER high impractical can be get for the transmissions of information. If K factor increase then

20
2016, IJITME All Rights Reserved, www.ijitme.org

Birendra Kumar et al, Int. Journal of Information Technology & Mechanical Engineering - IJITME,
Vol.2 Issue. 6, June- 2016, pg. 13-23
ISSN: 2349-2865
order of magnitude improve, if SNR value is 6dB, then BER value are 0.5 and channel has make into in appropriates, if
SNR value increase from 6 degree to 23 degree, then the BER has decrease and the value closed to 10 -6, if SNR value
between 6 dB to 16dB, then they have no significant improvement in BER, if SNR value more than 16dB then the
channel has improve in BER, if SNR value increases from 22dB to 23dB then BER of transmission on decreases.

Figure 4.6:BER characteristics of AWGN, Rician and Rayleigh channels.

Figure 4.7:K-factor Vs Eb/No of the Rician channel.

21
2016, IJITME All Rights Reserved, www.ijitme.org

Birendra Kumar et al, Int. Journal of Information Technology & Mechanical Engineering - IJITME,
Vol.2 Issue. 6, June- 2016, pg. 13-23
ISSN: 2349-2865

V.

CONCLUSION

The first was Matlab based and examined the effects of wireless channel, compression ratio and recognition quality of the
received speech. The second consisted of experiments on commercial Skype VoIP networks in order to measure speech
recognition and comprehension. The results of these simulations and experiments were reported The increase in the
secondary path delay variation causes an increase in the BER of the signal. As the signal strength of the secondary paths
increases, the BER increases as well .Performance of an out door wireless network was better than that of an indoor
network due to the effect of multipath occurring indoors.

REFERENCES
[1]

ElechiOnyekachi and Eze Elias, Investigating the QoS of Voice over IP using WiMAX Access Networks in a
Campus Network, Computer Engineering and Intelligent Systems, Vol.4, No.5, 2013

[2]

Erik Pertovt, Kemal Alic, Alevigelj and MihaelMohorcic, Performance Evaluation of VoIP Codecs over
Network Coding in Wireless Mesh Networks, Proceedings of the 2013 International Conference on
Electronics and Communication Systems

[3]

Mousavipour, F. and Khoseavipour, M.J. (2012) VoIP quality enhancement with wideband extension method
in broadband networks, IEEE (Revista IEEE America Latina)Latin America Transactions, Vol. 10, No. 1,
pp.11901194

[4]

Alshomrani, S. Qamar, S. Jan, S. Khan I. and Shah, I. A. (2012) QoS of VoIP over WiMAX Access Networks.
International Journal of Computer Science and Telecommunications. 3 (4), p92-98. IJCST

[5]

Mladen T. Koprivica, Mladen M. Ili, Aleksandar M. Nekovi, and Nataa J. Nekovi, An Empirical Study
of the EDCA QoS Mechanism for Voice over WLAN, Telfor Journal, Vol. 3, No. 1, 2011.

[6]

Muntean, V.H. and Otesteanu, M. (2011) Performance evaluation of DQOAS algorithm in case of applications
generating VoIP and video streaming when a new QoS prioritization scheme for LTE is used. Broadband
Multimedia Systems and Broadcasting (BMSB), 2011 IEEE International Symposium on LTE, vol. 5, no. 3, pp.16, 8-10 June 2011.

[7]

Haniyeh Kazemitabar, Sameha Ahmed, KashifNisar, Abas B Said, Halabi B Hasbullah, A Survey on Voice
over IP over Wireless LANs , World Academy of Science, Engineering andTechnology, pp. 71-78, 2010.

[8]

E. Dar and M. Latif, Issues of Implementation of Voice over IP with Quality of Service: An Overview,
International Journal of Computer Science and Telecommunications, Volume 1, Issue 1, November 2010

[9]

Bernardo, V.; Sousa, B. and Curado, M. (2009) VoIP over WiMAX: Quality of experience evaluation,"
Computers and Communications, 2009. ISCC 2009. IEEE Symposium on Communications, vol. 13, no. 5, pp.4247, 5-8 July 2009

[10]

Ansari, N.; Haghani, E.; Shyam Parekh; DoruCalin; and Eunyoung Kim (2009) A Quality-Driven Cross-Layer

22
2016, IJITME All Rights Reserved, www.ijitme.org

Birendra Kumar et al, Int. Journal of Information Technology & Mechanical Engineering - IJITME,
Vol.2 Issue. 6, June- 2016, pg. 13-23
ISSN: 2349-2865
Solution for MPEG Video Streaming Over WiMAX Networks. Multimedia, IEEE Transactions on 4G
Telecommunication, vol.11, no.6, pp.1140-1147, Oct. 2009.
[11]

Garuba, M.; Jiang Li; Zhenqiang Yi; Security in the New Era of Telecommunication: Threats, Risks and Controls
of VoIP. Information Technology: New Generations, ITNG, 2008. Fifth International Conference on, vol., no.,
pp.587-591, 7-9 April 2008

[12]

Hasib, M., Schormans, J. and Timotijevic, T. (2007) Accuracy of packet loss monitoring over networked CPE,
IET Communications, Vol. 1, No. 3, pp.507513

[13]

Chu, T-K. andPheanis, D.C. (2006) QoS evaluation of sender-based loss-recovery techniques for VoIP, IEEE
Network, Vol. 20, No. 6, pp.1422.

[14]

SG. Wang, S., Mai, Z., Xuan, D. and Zhao, W. (2006) Design and implementation of QoS-provisioning system
for voice over IP, IEEE Transactions on Parallel and Distributed Systems, Vol. 17, No. 3, pp.276288

[15]

Ganguly, S., Navda, V., Kim, K., Kashyap, A., Niculescu, D., Izmailov, R., Hong, S. and Das, S. (2006)
Performance optimizations for deploying VoIP services in mesh networks, IEEE Journal on Selected Areas in
Communications, Vol. 24, No. 11, pp.21472158.

23
2016, IJITME All Rights Reserved, www.ijitme.org