See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/280799903

Analysis and Design of E1 over Ethernet Gateway

Article  in  IJCDS Journal · September 2013

DOI: 10.12785/ijcds/020309

CITATIONS READS

0 1,520

3 authors, including:

Abdul-bary Suleiman Dia M. Ali

Ninevah University Ninevah University
17 PUBLICATIONS   77 CITATIONS    8 PUBLICATIONS   28 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

offline NIDS View project

Helthcare View project

All content following this page was uploaded by Abdul-bary Suleiman on 09 January 2018.

The user has requested enhancement of the downloaded file.

Int. J. Com. Dig. Sys. 2, No. 3, 173-180 (2013) 173

International Journal of Computing and Digital Systems

http://dx.doi.org/10.12785/ijcds/020309

Analysis and Design of E1 over Ethernet Gateway

Abdul-Bary Raouf Suleiman, Dia M. Ali and Nashwan Z. Hero

College of Electronics Eng., University of Mosul, Mosul, Iraq

E-mail addresses: suleimana52@uomosul.edu.iq, dia_mohamad@ yahoo.com, nashwan_hero@yahoo.com

Received 30 Jun. 2012, Revised 5 Nov. 2012, Accepted 30 Jan. 2013, Published 1 Sep. 2013

Abstract: The most commonly used technique in networking nowadays is packet switching. It has become one of the
important infrastructure elements in the communication society. Time-division multiplexing (TDM) and Synchronous
Digital Hierarchy (SDH) over Ethernet is regarded as an economical and efficient scheme in the mid-distance of TDM
or (SDH) interconnections. This paper focus on the design and analysis of the E1 to Ethernet protocol converter
(Gateway), it based on the frame granularity scheme which takes into account of miniaturization, power, cost and time
delay of companding reduction. The E1 to new Ethernet (LAN) link protocol converter designed in this paper utilizes
universal programmable devices. A new approach is proposed, modeled and then simulated by using (VHDL) code and
implemented in Spartan 3E kit. In high speed networks, packet processing is relatively expensive while bandwidth is
cheap. The proposed approach considers the most important factor affecting the performance of high speed networks.
These are packet processing time, throughput, and packet efficiency. Packet processing time is an issue in real time
communications. As consequences, real time signal quality (like the voice) might be degraded severely when the
process time exceed its limits. In this paper, the simulation results show that the active mode of the frame payload
effectively improves the performance of the Ethernet protocol. In E1 line, 36% bandwidth is saved at half traffic load of
the maximum.

Keywords: TDMoIP, TDMoE, Ethernet, E1, FPGA.

three years. Note that compatibility principle enforces the

I. INTRODUCTION main industrial companies to keep using the old
At the present time, Ethernet communication products, techniques. One of these old techniques is carrying and
applications and solution are becoming more entrenched combines the voice signals in time slots within a frame of
as the most common standard for data transport (carrier). at a specific time and speed and then transfer these signal
Carriers are reducing costs by eliminating the need for from point (like Private Branch eXchange (PBX)) to
added networks that require different resources for another point (like Public Switched Telephone Network
different service types. The primary goal for carriers is to (PSTN)). One of the available solutions for the
accommodate all of these services on a single network communications service providers and enterprise
where the Ethernet and IP (Internet Protocol) has this customers to deploy voice and leased line services over
ability clearly [1]. efficient network is Voice over IP (VoIP) which requires
Ethernet technology has evolved to the point, where an investment in new network infrastructure and
service providers are beginning to deploy new services to customer premises equipment (CPE). The second
the wide area network. The first task for this technology solution which end user equipment need not be replaced
was to carry a small and slow rate data from the is TDM over Internet protocol TDMoIP, whereby
computer to the peripherals at the mid of the seventies Internet backbone or modern packet switched networks
from the past century. Soon after that, higher data rate (like 2G systems) can be used for transport. TDMoIP was
and bigger bulk of data were easily transferred from node first developed in 1998 by RAD Data Communications
to node or to any peripherals like printers, storage and first deployed in Sweden in 1999 by Utfors [2].
devices, etc. Utfors employed the first generation TDMoIP product
Telephony technology, on the other side, has evolved (known as Ipmux-4) to provide bundled services
rapidly where a new generation is shown up every two or including TDM private lines, TDM leased lines and a
variety of IP and Ethernet services. In 2001, the IETF set

© 2013 UOB SPC, University of Bahrain

174 A. R. Suleiman, D. M. Ali and N. Z. Hero: Analysis and Design of E1…

up the PWE3 working group, and which was chartered to the communication quality. A number of studies
develop architecture for edge-to-edge pseudo wires, and investigated several factors as means for real-time
to produce specifications for various services, including protocol implementation;
TDM. Other standardization forums, including the ITU Yamamoto [5] provided an experimental study of the
and the MPLS - Frame Relay Alliance, are also active in effect of packet-size on speech quality. Based on
producing standards and implementation agreements for subjective experiments, the authors proposed several
pseudo wires. Conventional TDM networks are highly equations to describe the effects of packet size variation
deterministic. The circuit delay through a TDM network and packet loss.
is predictably low and constant throughout the life of a Oouchi et al. [6] evaluated suitable voice-packet length in
connection. Conventional TDM networks are highly IP packets for the adjustment of VoIP network systems.
deterministic with a source device transmitting one or The researchers used a simple test network and evaluated
more octets to a destination device. This occurs by a the effect of changing a voice data length on a packet loss
dedicating-bandwidth channel every 125 μs. TDM rate. Based on his results, the authors concluded that in
delivers timing with the data while tightly controlling most cases, a variable voice packet-length VoIP system
jitter and wander. Timing is delivered along with the would be useful to achieve both high-transmission
data, and the permitted variability (jitter and wander) of effectiveness and stable voice quality.
TDM clocks is tightly defined. In addition, the Ngamwongwattana [7], on the other hand, investigated
infrastructure supports a rich set of user features via a the effect of packet size variation on the end-to-end
vast set of signaling protocols [3]. delay. Due to theoretical studies and simulation he
The communication link is providing a constant data concluded that small voice packet-size is preferred for
stream with a fixed bit rate in TDM network. The higher minimal incurred delay but, because of a large IP-rate
the bit rate, the higher the bandwidth, and the higher the requirement, it has the potential to cause congestion,
amount of data that can be transmitted. Ethernet is a which could result in increasing end-to-end delay. Large
technology-based on package. Data is placed to be voice packet size incurs additional delay due to
transmitted to one or more of the packets and then deliver packetization.
whenever there is available bandwidth, there is a certain Eugene S. Myakotnykh, Richard A. Thompson [8]
randomness introduced. End-to-end delays may not be investigated the effect of voice payload size and
constant. The data may not even be delivered in the same compression variation on VoIP quality under various
packet but chopped up in several packets. The receiving network conditions.
application has to put them back into correct order. There Barani Subbiah[9] described a new method to multiplex
are technologies that address these shortcomings and a number of low bit rate audio streams into a single RTP
provide solutions for “real time services” such as VoIP stream between IP telephony gateways. In this paper,
(voice over IP) and TDMoIP (Time Division they described a new method to multiplex a number of
Multiplexing over IP). Another advantage with Ethernet low bit rate audio streams into a single RTP stream
is the plug-and-play nature of network. A TDM network between IP telephony gateways.
needs to have carefully designed channel allocation to Junius Kim[10] explained the emulated services, such as
optimize the bandwidth usage. Ethernet is largely self CES, allows TDM circuits to be bridged between
configured in this respect [4]. locations by providing a pseudo wire tunnel across a
In this paper the VHDL code used for simulation of the provider’s IP packet switched network.
E1 to Ethernet converter in new manner is proposed. This allows the legacy TDM companies to maintain its
This paper is organized as follows: the next Section existing TDM equipments and technology while using
provides a brief overview of related work in the research advantage of next generation core transport networks,
area. Section 3 describes TDMoIP architectures and thus extending the life of legacy TDM companies with
standards. Section 4 describes TDMoE. Section 5 equipments.
devoted to the analysis of the parameters effect. The
proposed gateway is explained in Sections 6. Section 7 III. OVERVIEW OF TDM OVER PACKET SWITCHING
describes compression operation. Section 8 explains ARCHITECTURES AND STANDARDS
model analysis. The experimental results are described in There are two main architectures for TDM over packet
Section 9. Finally, conclusions are made in Section 10. switching, structured and unstructured TDMoIP [11].
II. RELATED WORK A. Structured mode refers to transport of the active
TDM channels individually or in groups depending on
The TDM data travels through the Internet; an
their destination. Dealing with TDM channels allows
overhead with control information is added to the TDM
individual channel management priority, and quality of
data payload. The header size is 40 bytes: RTP header;
service requirements trade-off with a very large number
UDP header; and IP header which consume big part of
of small packages. TDM channels are collected and sent
bandwidth. To save the bandwidth, it is Possible to
to the same destination, TDM over packet switching can
remove these headers and to use the active mode that
reduce overhead packets regardless of their priorities.
compresses the TDM data payload without any effect on
A. R. Suleiman, D. M. Ali and N. Z. Hero: Analysis and Design of E1… 175

The disadvantage of this approach is generates more (RTP, UDP and IP) headers after remove these
traffic with high priority. headers from Ethernet frame structure.
C. Decreasing the complexity and cost of the
B. Unstructured mode refers to transport of the
Gateway device by decreasing the frame
TDM stream after transparently encapsulated from point-
complexity.
to-point connections over IP network as tube without an
understanding of the signaling and TDM services. Silent- V. FRAME SIZE EFFECTING
suppression can be supported to provide a better
The effect of frame size variation is difficult to
utilization of network resources, but with uncontrollable
describe theoretically because many of the parameters
delay since the TDM signal must be packaged and de-
affecting interactive data quality (delay, loss, and jitter)
packetized every TDM over packet switching network
are not independent and improving one parameter may
hop.
cause a decline in another. Some effects of frame size on
Congestion in the packet switching network will focus on
speech quality are very clear, others are less evident. Four
all strains, which degraded service quality for all calls,
main relationships are identified:
including calls with high priority [11]. Many packet
technologies will soon displace TDM infrastructure. Fig.
1 outlines the packet format for all of the standards that  A loss of a single “long” frame has more
always include network layer headers and the adaptation important negative effect on TDM information
layer [1]. Fig. 2 illustrates communication of multimedia quality than a random loss of several “short”
data on Ethernet LAN environment between end systems. frames.
 Increasing frame size leads to reduce of the IP
rate per call. This possibly will reduce
congestion in the network and get better the
quality of communication.
 In presence of data traffic in the network,
increase of TDM frame size decreases link
utilization, but increases the data-to-TDM traffic
ratio. This can cause additional “instability” in
the network, which may result in higher jitter,
loss or delay. This factor may affect the
resulting TDM information quality, but it might
be not clear how significant the effect. It is seen
Fig. 1. Comparison of Standards [1] that increasing frame size leads to different
effects on the TDM information quality.
 Increasing frame size leads to an increase of
end-to-end delay. If the delay is important, an
additional increase of frame duration may be
noticeable. But, if the delay is not too large, the
direct impact of frame size increase is very
small and not perceptually noticeable [12, 13].
VI. THE PROPOSED GATEWAY ARCHITECTURES
The main architectures of the proposed Gateway
consist of five units as shown in Fig. 3. It operates in two
Fig. 2. Complete gateway block diagram modes these are General TDM data mode and Active
TDM data mode. General mode holds to carry all time
IV. TIME DIVISON MULTIPLEXING OVER ETHERNET slots directly in the payload field. Active mode holds to
(TDMOE) carry the active time slots only in the payload field.
For the proposed method, the TDMoE uses the LAN
as an environment for transport of data. This means that
TDMoE deals with layer 2 functions far from upper layer
such as network, transport and application layer.
So carrying TDM over Ethernet (Not over other layer)
protocol achieves many benefits:
A. Decreasing the processing delay by removing
the process time of RTP, UDP and IP header.
B. Enhancing the packet efficiency by decreasing
the size of header by an amount of 40 bytes for Fig. 3. General block diagram of proposed Gateway
176 A. R. Suleiman, D. M. Ali and N. Z. Hero: Analysis and Design of E1…

The units of the proposed Gateway are: value will be completed. This value will be inserted in the
A. The frame Processor unit, which carries out two CRC field.
operations. The first operation is receiving the After adding physical headers, Ethernet frame is
TDM data from accumulating TDM unit, completed. The final stage of GW TX is the P/S
adding sequence header and inserting them in conversion of the Ethernet frame and loading on the TX
Ethernet payload field. To this payload field Ethernet link when link was idle. It can be noted that the
Ethernet headers are added. These headers are sequence value will increment to be compared with
destination address, source address, and maximum value as shown the flow chart of Fig. 5. The
type/length field. CRC is also added to the frame Ethernet frame format explained in Fig. 4 represents the
after being calculated as a trailer. This frame output of GW TX processes.
containing physical headers (SFD and preamble)
field is transferred to the Ethernet framer unit.
In the second operation the Ethernet frame is
received from the accumulating Ethernet unit.
Matching the destination address and source
address, the CRC will be checked and if valid
the next process of de-capsulation and recovery Fig. 4. Ethernet frame fields
of TDM data is starting. After that, the TDM
data will be sent to the E1 framer unit.
B. Accumulating TDM unit receives data from RX
E1 link to convert the data from serial to
parallel. This unit represents the S/P converter
unit and then the data sent to the unit frame
Processor unit.
C. Accumulating Ethernet Unit receives the frame
serially from the RX Ethernet link to convert it
from serial to parallel. This unit represents the
S/P converter which sends the Ethernet frame to
the unit frame Processor unit.
D. E1 framer Unit receives the E1 frames from a
frame processer Unit loading them serially in
synchronous with the Clock sync. to the TX E1
link.
E. Ethernet framer Unit receives Ethernet frames
from frame processer Unit to be loaded serially
on the TX Ethernet link when the link is idle.

Processing to be applied on serial data bits from RX E1

link to TX Ethernet link is explained as flow chart shown
in Fig. 5. CRC value is calculated partially after reaching
each E1 frame at the frame processor unit, coming from
accumulating TDM data unit. DA, SA, Length, and
sequence headers are encountered in the partial
calculation of CRC value before the first E1 frame only.
On receiving of the remaining seven E1 frames the
calculation of the CRC value will be completed. This
value will be inserted in the CRC field. Partial calculation
of CRC value results in reduction of processing time.
These processes included in General mode.
In active mode the E1 frame is compressed to be
composed of (active frame + control word). At the partial
calculation of CRC value the control word value will be
checked if is equal to zero, the (active frame + control
word) is replaced by one bit (X = 0) and if not the bit
(X = 1) will be added to (active frame + control word) in
the partial calculation of CRC value. On receiving of the
remaining seven E1 frames the calculation of the CRC Fig. 5. G.W. TX at Active M. frame mode
A. R. Suleiman, D. M. Ali and N. Z. Hero: Analysis and Design of E1… 177

In the same manner, processing to be applied on serial VII. COMPERSSION AND DE-COMPERSSION
data bits from RX Ethernet link to TX E1 link as shown OPERATION
in Fig. 6. The Ethernet frame bits is converted from serial The active TDM data mode includes a compression
to parallel by the accumulating unit directly after process for the E1 frame. The compression stage use
removing the physical headers. The CRC value is statistical time-division multiplexing type in which active
calculated after matching each destination address and time slots are dynamically allocated into active frame
source address. If the addresses are valid, the CRC value field to improve Ethernet frame efficiency. In active
will be checked if is equal to zero the payload is used to frame, the number of time slots in each active frame is
recover the E1frames and if not, or addresses not match, less than E1 frame, so the active frame length will be
the Ethernet frame (payload) will be neglected. In general shorter. But, to recover data, the active frame needs the
mode the E1 frames are recovered from payload directly address of the location of all time slots. The control word
without any process. In active mode the E1 frames are (C.W) will represent this address. Each time slot need
recovered from (active frames + control words) by de- one bit to determine the activation and position, so the
compression process. De-compression process will be length of control word is 32 bit for 32 time slot of E1
explained in detail later. By this, E1 frames are frame. In compression stage, there is no fixed
recovered. The final stage of GW RX is the P/S relationship between the inputs and outputs length of
conversion of the E1 frames and loading on the TX E1 frame because there are no pre assigned or reserved slots.
link serially in synchronous with clock sync. For example, if E1 frame at the input is
(098756380021000700549100000017990000800000054
0000084001200001800) hex
then the C.W. and active frame respectively will be:
(75632352), (87563821075491179980054054841218)
This example shows a major difference between E1frame
and active frame and C.W. This compression process is
allows the recovery of the original E1 at de-packetization
stage, which can be applied to all types of information
from TDM data. Compression process is explained in
Fig. 7. By the same manner the original E1 frame will
recovered depending on address field (C.W.) at
de-packetization stage as explained in Fig. 8.

Fig. 6. G.W. RX at Active M. frame mode Fig. 7. Compression process

178 A. R. Suleiman, D. M. Ali and N. Z. Hero: Analysis and Design of E1…

B. Protocol performance analysis

The factors of throughput, delay, loss, transport rate,
frame efficiency, and frame size will limit the
performance level. The proposed technique uses eight E1
frame in payload. In general mode, it gives (1 ms)
transport rate, (2.2 Mbps) throughput, (90%) frame
efficiency, and (275) frame size. The frame size takes
20% only from total length of Ethernet frame but the
benefit was reducing the loss effect by ratio 20:1
comparing with G.711 code. In active mode, it gives
same results with improvement in throughput with value
depending on active ratio of time slots and E1frames
together. Active time slots ratio and active E1 frames
ratio limits the frame size and frame efficiency.
IX. EXPERIMENTAL RESULTS
Large difference length of Ethernet frame is appeared
in the Figs. 10 and 11 at 0% and 100% active ratio. The
Fig. 8. De- compression process process delay is clearly longer at 100% active ratio
comparing with 0% active ration situation.
VIII. MODEL ANALYSIS
This section introduces an analysis for the proposed
gateway model in terms time delay, frame error, and
protocol performance.
A. System Test Bench
One of the proposed test system parts is Real time slot
simulator for E1 link generator explained in Fig. 9. This
block generates the E1 frame with active time slot and
active E1frame ratios controlled by software settings. E1
frame will generates time slots by steps (1-32). M. E1
frame (8 E1 frame) will generates E1 frames by steps (1-
8).
E1 frame received by GW TX model and send it as
Ethernet frame via serial link to GW RX model.
The E1 frames will be received at this stage and the
performance evaluation in terms of time delay and frame
error are performed.
Fig. 10.. Ethernet frame signal at (100%) active ratio

Fig. 9. The proposed test system Fig. 11. Ethernet frame signal at (0%) active ratio
A. R. Suleiman, D. M. Ali and N. Z. Hero: Analysis and Design of E1… 179

Fig. 12 shows the relation between the bandwidth and the [http://www.ciscopress.de/main/main.asp?page=home
no. of active byte (time slot) for general mode and active /bookdetails&ProductID=171749], last date accessed:
mode for different ratios. From the figure, it can be 6/6/2012.
deduced that there is bandwidth saving for active mode. [3] Yaakov (Jonathan) Stein and Brian Stroehlein”
It can be seen that there is benefit of bandwidth saves for Taking an Inside Look at TDMoIP” white paper RAD
all frame active ratios compared with general mode. Only Data Communication Corporation, available at ,
for the case in which the no. of active byte is more than [http://lagercrantz.verktyget.pdf], last date accessed:
25 and frame active ratio 100%, the required bandwidth 6/6/2012.
for active mode is greater than required bandwidth of
[4] Ward, S., Higinbotham, W. and Duvelson, E.,
general mode. “Integrating Current Differential Relaying
Communications into an IP Based Infrastructure”,
Protective Relay Engineers, 2009 62nd Annual
Conference, pp. 382 – 393, March 30 2009-April 2
2009.

[5] Yamamoto, L., Beerends, J., “Impact of network

performance parameters on the end-to-end perceived
speech quality”, Expert ATM Traffic Symposium,
Greece, 1997.

[6] Oouchi, H., Takenaga, T. , Sugawara, H. and Masugi,

M. , “Study on Appropriate Voice Data Length of IP
Packets for VoIP Network Adjustment”, Global
Telecommunications Conference, GLOBECOM '02.
IEEE , vol.2, 17-21 Nov. 2002, pp. 1618 – 1622,
2002.

[7] Ngamwongwattana, B., “Effect of packetization on

Fig. 12. Throughputs vs. Active ratio VoIP performance”, Electrical
Engineering/Electronics, Computer,
X. CONCLUSIONS Telecommunications and Information Technology,
2008. ECTI-CON 2008. 5th International Conference,
In this paper, the applications of E1 over Ethernet
vol.1, pp. 373 – 376, 14-17 May 2008.
protocol convertor are introduced and the features of
Ethernet frame are analyzed. A new E1 over Ethernet [8] Myakotnykh,E., Thompson, R., “Effect of Packet Size
protocol is presented as a gateway matching different and Compression Variation on Quality of VoIP”,
network technologies. The proposed gateway is Security and Communication Networks (IWSCN),
implemented using VHDL. 2009 Proceedings of the 1st International Workshop,
pp. 1 – 6, 20-22 May 2009,
The time delay and the bandwidth reduction of the
designed protocol convertor was tested and [9] Subbiah, B., Sengodan, S., Rajahalme, J., “RTP
analyzed under different load ratios of E1 link. It Payload Multiplexing between IP Telephony
is shown from the results that the required Gateways”, Global Telecommunications Conference,
bandwidth is improved when Ethernet frames ware GLOBECOM '99, 1999, vol.2, pp. 1121 – 1127, 1999.
compressed specially when the E1 frame not fully
[10] Cassing, D., “C6X Solutions for Voice Over IP
loaded (active frame ratio below 87%) . Because the Gateway”, Northcon/98 Conference Proceedings, pp.
accumulating delay, the proposed Gateway has 74 – 85, 21-23 Oct 1998.
some limitation on the maximum Ethernet payload
size such that the process delay not exceed 200 [11] Kure, O.E. Sorteberg, I. Oevsthus, K. NTNU,
ms. Trondheim, Norway “Architecture for TDM circuit
emulation over IP in tactical networks” Military
REFERENCES Communications Conference, MILCOM 2003. IEEE,
Vol.2, pp. 921 – 926, 2003.
[1] Ernhofer, B., Frost, T., “Circuit Emulation Services-
over-Packet and Standards Adoption”, available at [12] The E-model, a computational model for use in
[http://www.analogzone.com/nett1025.pdf], last date transmission planning”, 2000, available at
accessed: 6/6/2012. [http://dev.leurent.eu/voip/MOS/T-REC-G.107-
200503-S!PDF-E.pdf], last date accessed: 6/6/2012.
[2] Grayson, M., Shatzkamer, K., Wainner , S., “IP
Design for Mobile Networks” Cisco Press, Jun 15, [13] Myakotnykh, E., Thompson, R., “Effect of packet size
2009, available at and compression variation on Quality of VoIP
Communication”, Security and Communication
Networks (IWSCN), 2009 Proceedings of the 1st
International Workshop, pp. 1 – 6, 20-22 May 2009.
 180 A. R. Suleiman, D. M. Ali and N. Z. Hero: Analysis and Design of E1…

Dia M. Ali received his BSC in

Elec. Comm. Eng. From Mosul
university in 1992 (Mosul/
Iraq). In 1998 he gets MSC
in Elec. Comm. Eng. Mosul
university (Mosul/ Iraq) and
2007 he recieved a PhD in
Comm. Eng. (Network) from
Mosul university (Mosul/ Iraq).
From 1993 to 2001 he worked
at R&D Center Mosul / Iraq .
Since 2007 join university of Mosul college of
Engineering as lecture. He interesting in Network
Modeling and simulation especially in (OPNET
MODELER), Network Security, Mobile Network
planning, Antenna modeling and System Automation.

Nashwan Hero received his

BSc in Electronic and
communication engineering at
2001. He received an M.Sc. in
communication Engineering
2011. He is now a member in
communication Department
collage of Electronics
Engineering university of
Mosul

Abdul-Bary Raouf Suleiman

B.Sc. in Electronics and
Communications Eng./1976
M.Sc. in Electronics and
Communications Eng./1991
Ph.D. in Computer Eng./
Digital Signal Processing/ 2001
Qualifications:
1) Has Cisco Certifications
from the Cisco Academies
Training Centre (CATC) at
Lebanon in the years 2004 and 2006 in :
a. Computer Networking (CCNA 1-4)
b. Information Technology (IT essentials)
c. Wireless LAN (WLAN).
d. Network Security 1 and Network Security 2
2) Instructor at Mosul University Regional Academy.
3) Legal main contact and Instructor at College of
Electronics Eng. Local Cisco Academy.

View publication stats