hothaifa paper2

8/8/2019 hothaifa paper2

1/15

Abstract- in this paper , a design and

simulation for Asymmetric Digital

Subscriber Line (ADSL)/AsymmetricDigital Subscriber Line 2 (ADSL2)

Initialization Process is presented which

can be applied to different telephone

network. The Initialization process is

designed and simulated under the

MATLAB v7 environment. ADSL channelfaces different types of noise, the most

important types are Background and

Crosstalk noises and Intersymbol

interference (ISI) which can be eliminated

by using Time Domain Equalizer (TEQ).Minimum Mean Square Error (MMSE)

algorithm is implemented as TEQ

algorithm. The initialization process where

tested on American National Standard

Institute (ANSI ) define 8 Carrier Serving

Area (CSA) test loops for ADSL service Asa result , 9.02 Mbps and 9.42 Mbps were

achieved over CSA loop-2 for ADSL and

ADSL2 respectively. The TEQ efficiency

for 8 CSA test loop are calculated where it

achieved 93.1% of zero ISI for CSA loop-

3.a Comparison between Near End Cross-

Talk (NEXT) and Far End Cross-Talk

(FEXT) power shows that NEXT havehigher power and narrow band where

NEXT powers were (-45.593) dBm for allloops and FEXT were ranging from (-

76.353 to -68.261) dBm for the 8 loops.

Finally the results show that ADSL2

outperforms ADSL by about 400Kbps

which about 1.1%.

I. IntroductionBroadband or high-speed Internet access is

provided by a series of technologies that

give users the ability to send and receive

data at volumes and speeds far greater than

current Internet access over traditionaltelephone lines. In addition to offering

speed, broadband access provides a

continuous, always on connection (noneed to dial-up) and a two way

capability, that is, the ability to both receive

(download) and transmit (upload) data at

high speeds. Broadband access, along with

the content and services it might enable, has

the potential to transform the Internet: both

what it offers and how it is used. It is likely

that many of the future applications that

will best exploit the technologicalcapabilities of broadband have yet to be

developed [1].

Telephone line based technologies

provide dedicated access to the individual

users. One of the best solutions is DigitalSubscriber Line (DSL) access, which is

targeted for residential users and has

received much attention by many telephone

companies. The architecture of DSLsystems allows telephone companies to useexisting twisted-pair infrastructures, by

which there is no need to lay extra lines for

new services, for their next-generation

broadband access networks.

The most promising of the xDSL

technologies for integrated Internet access

is Asymmetric Digital Subscriber Line(ADSL) by which ADSL is designed to

Design and Simulation of ADSL/ADSL2 Initialization

Process


2/15

interoperate with the telephone, i.e.

allowing voice and high-speed data to be

sent simultaneously over the same line.

Unlike today's computer dialup modems,the customer may use the telephone while

the computer is connected to the Internet

Service Provider (ISP). The achievable data

rate is significantly higher in ADSL by

almost 100 times compared to today's

fastest 56 Kbps modems [2].A recent development is ADSL2 and

ADSL2+. It defines new applications,

services and deployment scenarios.

The first issue it has to deal with is a

wide range of used cables in telephone loop

plant also a varying length of loops from

few hundred of feet to 18000 feet. Themodem should adept well with these

changes [3].

Other users using ADSL or other

communication service running on the same

binder causes noises, these noises is called

crosstalk noises, where background noiserepresents the thermal and environment

effect. Crosstalk noises have the dominant

effect on ADSL system performance.

Other ADSL impartment whenoperating on long loops is dispersive

behavior of the channel, resulting in a wider

received pulse. This causes a time sample

to spread into the neighboring time slots

that causes InterSymbol Interference (ISI).To eliminate this effect ADSL modem

should shorten the channel impulse to

desirable length.

Cioffe [4], described fundamentals of

MCM and how it is analyzed for channels

with ISI and additive Gaussian noise. Inoue

[5] ,was concerned with an improvement topreviously proposed ADSL echo canceller

using modified conjugate gradient for

adaption schema. Arslan [6], studied

different kinds of equalization technique for

discrete multitone transceiver and

developed a time domain equalizers (TEQ)design method to optimize the channel

capacity at the output of the TEQ. Nadhim

[7] studied MCM technique in DSL

systems using the Fourier transform and

then he used the discreet wavelet transform

(DWT) and studied the systemperformance. DALY [8], dealt with the way

of improving the efficiency of multicarrier

communication on the digital subscriber

loop and also examined bit and power

loading algorithms for multitone systems.

Since ADSL channel does not change

during data transmission and only minornoises level changes occurs, all problems

are handled during ADSL modem start-up

process, which is known as ADSL

initialization process. In this work

ADSL/ADSL2 initialization process in

simulated under MATLAB version 7environment.

II. ADSL/ADSL2 InitializationThe initialization process allows the ATU-

C and ATU-R to establish their

communications. The process allows the

two modems to identify themselves to eachother, determine line conditions available tosupport communication, exchange

parameters that define the request

connection, allocate resources, and begin

normal communication. The process is

divided into four phases as described in

ANSI standard:Activation and Acknowledgment: TheATU-R begins the initialization process by


3/15

transmitting the appropriate tones to the

ATU-C. When this segment of initialization

is complete, the ATU-R and ATU-C have

negotiated the timing method used betweenthem and have determined which device is

the master. At the end of this procedure, the

ATU-R and ATU-C are in state capable of

analyzing the line condition. In this paper,

the ATU-C and ATU-R are assumed to

have negotiated the timing method andperfectly handshaking and all activation and

acknowledgment process is done perfectly.

Transceiver Training: During this process,

the ATU-R and the ATU-C send signals

that allow their partner to determine line

conditions and adjust the equalization of

their transceiver. Transceiver training alsodetermines if ADSL is operating in FDM or

Echo cancellation mode.Channel Analysis: The modem exchange

information on the upstream and

downstream bearer channel required for the

connection the latency paths they will beplaced in, and the bandwidths for each

channel requested. Information about

specific features supported or requested is

also exchanged. The modems then performtest that determine the loop quality and

SNR for each specific 4 kHz DMT tone.

Exchange: Having gathered the

information about quality of the connection

and the requested configuration, themodems configure themselves and

exchange information about their

configuration, the specific bandwidth

allocated to the requested bearer channels is

assigned, the specific DMT tones and the

amount of the data encoded in each tone are

determined and assigned. The connection istested in both directions after which each

modem notifies its peer that it is ready to

enter normal communications. In this

thesis, this phase is represented by bit

loading process.

The ADSL test loop modelingprocess simulate the ANSI CSA loops as

ADSL modem test loop, noise modeling

process simulate AWGN, NEXT and FEXT

noises for ADSL service. Channel analysis

process measure ADSL modem

performance on test loops with theexistence of noises by calculating signal to

noise ratio (SNR) assuming free ISI

channel. Since the used test loop is not free

ISI, therefore ISI elimination process is

required. Channel equalization process

eliminate channel ISI by using Time

Domain Equalizer (TEQ) and measuremodem performance on test loop after

equalization by calculating equalizer output

signal to noise ratio (SNR). Bit loading

processes calculates modem performance in

terms of bit rate and calculate modem bit

loading table using equalizer output (SNR).

A. DSL channelThe transmission characteristics of DSL

loops determine the performance of DSLsystems. DSL loops are based on existing

analog telephone subscriber loops which

were originally developed for voice

communication. A subscriber loop consistsof twisted pair cables that connect a localcentral office to customer premises. Two

subscribers are connected to each other

through central offices.

In order to facilitate the development

of broadband communication on the DSL,

the ANSI and ESTI standardization bodies

specified test loops which would hopefully


4/15

be representative of a large cross-section of

the DSL links encountered in practice.

The ANSI introduced the Carrier

Serving Area (CSA) engineering guidelinesin the early 1980s to shorten subscriber

loop length, which reduces loop

deployment cost and supports all future

digital services. A carrier serving area

(CSA) is a plant administration area

subsection of the main loop plant. [4].These generic lines are referred to as ADSL

CSA test-loops [9].

Figure 1. Transmission Line

Schematic.

The ANSI CSA test loop is modeled and

used as test loops as is shown in Fig 1. Theload impedance shown in Fig. 1 isconsidered to be a real constant (100 inANSI loops). The eight standard CSA loops

are used as test channels in the

ADSL/ADSL2 initialization simulation.

A splitter is used to separate the voice

signal from ADSL signal, hence a highpass

filter with low transition frequency isneeded. For the purposes of modling, a fifth

order Chebyshev high-pass filter with cut-off frequency of 4.8 kHz was used to

simulate splitter effect.

Figure 2. Configuration of the Eight

Standard CSA loops. Number Represent

Length/Thickness in Feet/Gauge. Vertical

Lines Represent Bridge-Taps.

B. Telephone Line ImpairmentsSubscriber loops which connect the

customer premises to a Central switching

Office (CO), were developed and deployed

for voice transmission. The term loop refers

to the twisted copper pair telephone line

from a CO to the customer. This termoriginates from current flow through a

looped circuit from the CO on one wire and

returning on another wire [10].

All signals sent over conventional

pair-cable telephone lines are subject to lineattenuation, dispersion and electrical noise.

Line attenuation and some forms of in-band

noise both increase with frequency [11].


5/15

Intersymbol interference (ISI) is an

unavoidable consequence of both wired and

wireless communication systems. ISI

causes spreading of received signal.As transmissions occur more frequently

(that is, the symbol rate is increased, which

corresponds to decreasing symbol time T

for each successive), then a given channel

shape exhibits increasing overlap of

transmissions and thus more ISI.Equalization methods adaptively configure

the receiver to mitigate ISI [12].

Multicarrier Modulation (MCM)

systems use Cyclic Prefix (CP) to separates

the symbols in time in order to decrease ISI.

As it is well known, the signal going

through the line is linearly convolved withthe impulse response of the line. If the

impulse response is shorter than the

duration of the cyclic prefix, each symbol

can be processed separately, and there is no

ISI [11].

Another impairment of data transmission is

white noise. One special kind of noise is

Additive White Gaussian Noise (AWGN)

which represents background noise asalready specified by its name. It has a flat

spectrum and a Gaussian amplitude

probability. Normally it is characterized by

its power spectral density.

Since AWGN have constant PSD( /2), it is easly built by generating arandom signal process and muliplying it bythe AWGN power, where AWGN power is

the AWGN PSD (Watt/Hz) multiplied by

the used bandwidth.

For simulation, different levels ofAWGN have been used, all moving in a

range between (-170 to -140 dBm/Hz)

single sided power spectral density [13].

For ADSL, however, -140 dBm/Hz seems

to be the most used value for AWGN [14].

The performance of DSL transceivers

can be impaired by interference from. Thetwisting of the wire pairs reduces this

inductive coupling (also referred as

crosstalk), but some signal leakage remains.

Crosstalk is most pronounced at the

segment of cable near the interfering

transmitters. The crosstalk resulting fromother transmission systems in the same

cable (and especially the same binder group

with the cable) is a primary factor limiting

the bit rates and loop reach achievable by

DSLs. There are two types of crosstalk

[12].

Near End Cross-Talk (NEXT) is amajor impairment for systems that share the

same frequency band for upstream and

downstream transmission. NEXT noise is

seen by the receiver located at the same end

of the cable bender of the transmitter that is

the noise source.Far End Cross-Talk (FEXT) is the

noise detected by the receiver located at the

far end of the cable from the transmitter

that is the noise source. FEXT is less severethan NEXT because the FEXT noise is

attenuated by traversing the full length of

the cable [12].

In order to evaluate a NEXT or a

FEXT crosstalk signals, the PSD functionof the interfering service has to be known.

The PSD of downstream and upstream for

ADSL will be discussed in this subsection.

This paper will focus on Standard

T1.413 by ANSI as a reference for the

ADSL service. The ADSL upstream

according to ANSI can be given by


6/15

the is the same for both thedownload and upload stream except for the

differences between the and coefficient. The NEXT forADSL service can be calculated using

where U is the number of disturbers in thecable, is the couplingcoefficient for 49 NEXT disturbers and is the frequency in Hz. The FEXT for

ADSL service can be calculated using

Where

is the channel gain (frequency

response), 8 x 10-20

is the coupling

coefficient for 49 FEXT disturbers and isthe coupling path distance.Generating a crosstalk signal is

conveniently accomplished by defining a

filter transfer function in terms of thedesired power spectral density (PSD). If

the input to the filter is a white noise

process with unit variance, then the PSD is

equal to the magnitude squared of the filter

transfer function. The filter transferfunction is Where is the PSD in Watts/Hz.The total Noise PSD will be

Since there are different PSD origins, each

PSD is can be simulated independently and

the resulted signal is summed to generate

the total noise signal.

C. Equalization for DiscreteMultitone Modulation

With DMT, the problem of fully equalizing

a channel is converted into partitioning the

channel into small subchannels which is

more efficient to implement in high-speed

transmission. However, this does not imply

that equalization is not required in a DMT

system. The spectra of each inverse FFT(IFFT) modulated subchannel is a sampled

sinc function which is not bandlimited.

Demodulation is still possible due to

the orthogonality between the sinc

functions. An ISI channel, destroysorthogonality between subchannels so that

they cannot be separated at the receiver.

One way to reduce ISI with a shorter

cyclic prefix is to use an equalizer. Sincethe length of a DMT symbol is longer thana symbol in single carrier modulation,

equalization is simpler. Also, noise

enhancement by the equalizer is not an

issue because the equalizer does not affect

the SNR in each subchannel, which are the

primary parameters to determine the

performance of a DMT system.

The ADSL standard uses a guardperiod, time-domain equalization, and

frequency-domain equalization. The Time-

Domain Equalizer (TEQ) shortens the

channel to a length of a predetermined but

short guard period. The TEQ can be

implemented as an FIR filter whose filter

coefficients are trained during initialization.

This combination has been standardized

and is implemented in practical systems [4].


7/15

Minimum Mean-Squared Error Impulse

Response Shortening Algorithm

Chow and Cioffi [15] are the first to apply

channel shortening equalization tomulticarrier modulation. They use the

MMSE design method to shorten a given

channel to the length of the cyclic prefix.

The idea behind the MMSE TEQ

design method may be explained by Fig. 3.

The structure consists of an FIR equalizerin cascade with the channel and a parallel

branch that consists of a delay and an FIR

filter with a target impulse response (TIR).

The goal in the MMSE design of the vector

of TEQ taps (w) is to minimize the mean

square of the error between the output of

the equalizer and the output of the TIR.Assume that the error is zero for any given

input signal. That means the impulse

responses of both branches are equal. In

other words, the equalized channel impulse

response (upper branch) would be equal to

a delayed version of the TIR. Setting thenumber of taps of the TIR to a desired

length forces the equalizer channel impulse

response to have the same length.

Figure 3. Block diagram of Minimum

Mean-Squared Error (MMSE) Equalizer[6].

Fig 4. shows a TIR and Shortened

Impulse Response (SIR). The MMSE

design method formulates the square of the

difference between the TIR and SIR as the

error and minimizes it. The method

minimizes the difference between the TIRand SIR both inside and outside the target

window.

In fact, the difference between the

TIR and SIR inside the target window does

not cause ISI. Both the TIR and SIR inside

the target window have higher amplitudes,which mean that difference inside the target

window might contribute more to the MSE

than the difference outside.

Since the MMSE method in general

cannot force the error to become exactly

zero, some residual ISI will remain. To

maximize channel capacity, the residual ISIshould be placed in frequency bands with

high channel noise.

Figure 4. Target Impulse Response (TIR)

and Shortened Impulse Response (SIR)[6].

This ensures that the residual ISI

would be small compared to the noise andthe effect on the SNR would be negligible

[4]. The MMSE design method does not


8/15

have a mechanism to shape the residual ISI

in frequency. MMSE algorithm is

implemented in chapter five and used to

eliminate ISI.

Table 1. Performance Comparison

between Selected TEQ algorithems [16].

Loop

CSA

Bit Rate as a percentageof Match Filter Bound

(MFB)

Rb(MFB)(Mbps)

GeneralMMSE

MSSNR MinISI

1 92% 62% 98% 8.45

2 90% 75% 97% 9.68

3 91% 82% 99% 8.11

4 92% 61% 98% 8.05

5 85% 72% 98% 8.53

6 93% 80% 99% 7.77

7 77% 74.2% 96.0% 7.75

8 56% 71% 99% 6.90

A comparison of selected TEQ algorithmscan be shown in Table 1 as percentage to

Match Filter Bound (MFB) which

represents perfect equalization, the number

of TEQ taps is 17. The Minimum Mean

Squared Error (MMSE) is the mostcommonly used in commercial ADSL

modems. MMSE design methods are

relatively easier to implement with adaptive

algorithms and are efficient in the sense ofcomputational complexity [4]. MMSEalgorithm is implemented in chapter five

and used to shorten test loop impulse

response.

D. Bit LoadingThe process of assigning information and

energy to each of the subchannels is calledbit loading in multichannel transmission

[12]. As was described before, the encoder

takes the data bit stream and encodes it into

N QAM constellation points for each

subchannel. This encoding is doneaccording to the bit loading table which

defines the number of bits carried by each

tone.

The ADSL system has the following rules

for bit loading as given in ANSI T1.413:

Only integer number of bits isallowed, if the resulted is notinteger, it is rounded to the least

integer.

The minimum number of bits thatcan be carried on any sub channel is

2 (4-QAM), so any subchannel that

cannot carry 2 bits is turned off.

The maximum number of bits thateach subchannel can carry is 15 bits,

if the resulted is greater than 15bits, it is replaced by 15 bits

In ADSL2 system, the same rule can be

applied except for the following, asdescribed in ITU G.992.3:

The minimum number of bits thatcan be carried on any sub channel is

1bit (2-QAM), rather than 2 bits for

ADSL, any subchannel that cannot

carry 1 bit is turned off.

E. Channel AnalysisThis section will discuss the process of

calculating ADSL signal to noise ratio

(SNR) which is needed to determine the

channel capacity.

During channel analysis processATU-C send signal know as Medly. This

signal consists of binary random data

transmitted to ATU-R. All subchannels are


9/15

used and each subchannel is modulated

with 4-QAM signals at maximum allowable

power.

On the other side, the ATU-R willreceive the Medly signal but after the signal

was filtered by the channel. The ATU-R

will need also to estimate channel noises.

The received Medly signal power and its

calculated noise are used to calculate each

subchannel SNR value.Fig. 5 shows the ADSL downstream

mask. Using ADSL PSD mask, the total

power can be calculated, as stated

previously only ADSL downstream ATU-C

to ATU-R will be simulated. The used PSD

mask that is the downstream PSD over

POST with overlap spectrum so thebandwidth that will be used for

transmission is from 25.875 KHz to 1104

KHz.

Figure 5. ADSL Downstream PSD Mask.

The maximum transmitted power

PSD is -36.5 dBm/Hz but a margin of 3.5dBm is used as the filter pass band ripple as

was described in ANSI T1.413[17]. Theresulted maximum power will be -40

dBm/Hz. Hence the total power is given by:

Also

where

is the transmitted signal power

spectrum and represent subchannelpower.Using the generated noise and theknown channel impulse response, the SNR

for each of the DMT subchannels can be

calculated. Hence

where

is the SNR for the

subchannel,

is the channel gain

(frequency response) for the subchannel(which can be calculated by taking thesquare of the absolute value of the Fourier

transform of the channel impulse response), is the noise PSD for the subchannel, assuming free ISI channel. The

SNR for a channel without ISI is referred to

as

, where MFB stand for Match

Filter Bound. The average SNR for DMT

channel can be calculated using.

Where term is called the average SNRor geometric SNR (.

The Time domain equalizer (TEQ)

algorithm is not perfect and do not shorten

the channel to (v+1). The residual will

cause ISI and that will affect on the systemperformance. To calculate SNR that include

the residual ISI the following equation can

be used :

where is the equalizer output SNR, is the equalizer gain (frequencyresponse) and

is the ISI gain of the

residual ISI.


10/15

The represents the equalizeroutput SNR which is used to calculate

channel capacity. In order to calculate the

efficiency of the equalizer, the total bit ratefor equalized channel and Match Filter

Bound channel using the geometric SNR of

the equalizer output and thegeometric SNR of the free ISI channel are used:

To calculate the number of bits that each

subchannel can carry, the default bit-loading algorithm is used to calculate thebit loading table as explained in subsection

D where :

Here, is the number of bits that can becarried on the subchannel and is themodulation gap.

For DSL application the targeted bit

error rate is , the modulationgap for QAM ( ) is 9.8 dB as definedin ANSI, and thus, the for a DSL system

is

where is the DSL margin, for standardADSL system the margin is 6 dB, is thecoding gain for the used error correcting

system, the code gain for ADSL system

with RS code is 4.2 dB and for ADSL2where Trills code is used the code gain is

5.5 dB as defined in section 4.5. The total

bit rate () can be calculated by using: where is the total number of transmittedbits in one frame and is the frame period.

III. Simulation of ADSL/ADSL2Process

The CSA loops modeled with impulseresponses consist of 512 samples and

sampled at a rate of 2.208 MHz using

Linemod[16], the ADSL channel noises, the

MMSE TEQ algorithm were all modeled

using MATLAB then ADSL channelanalysis is calculated finally the bit-loading

table is built, The MATLAB code were

assembled to create a program that

simulate the initialization process for

ADSL/ADSL2 modems, as stated before

the program simulate the downstream

ATU-C to ATU-R only.

As shown in Fig. 6, the program

consists mainly from 6 MATLAB file

codes, that simulate the ADSL/ADSL2

initialization process. Each file represent a

process that was explained in previoussubsections.

Table 2 shows the used parametersfor ADSL/ADSL2 initialization process

simulation.Table 2. ADSL/ADSL2 Test Parameters.

Parameters

Name

Value

Input Power 20 dBm

Number of

Equalizer Tap

19 tap

Margin 6 dB

Code Gain 4.2 dB for ADSL

5.5 dB for ADSL2

AWGN Power -140 dBm/Hz

NEXT Users 10

FEXT Users 10


11/15

A GUI (graphical user interface) was built

to ease the use of the initialization program,IV. Simulation Results and

Discussion

Figure 6. ADSL/ADSL2 Initialization Program Structure.


12/15

In this subsection the result for the ADSL

initialization on CSA test loop-1 is

presented. ADSL2 have similar results

except for bit loading table since ADSL2modem has higher code gain.

A. Channel Modeling ResultsADSL CSA loop-1 was simulated as shown

in figure (5.19) and filtered with high pass

filter to simulate POST splitter effect on the

channel effect. The impulse response of the

channel is much greater than 33 (v+1)

which is the maximum allowable value forADSL. For this reason, the Time Domain

Equalizer (TEQ) is required.

B.Noise Modeling ResultsUsing Eqs (1-5), the noise signals AWGN,

NEXT, FEXT were modeled. Fig. 7 shows

AWGN, NEXT and FEXT power spectrum.The Next Power is -45.593 dBm and FEXT

power is -71.475 dBm. Table 3 shows theNEXT and FEXT power for all loops.

It is noticed that NEXT power did

not change with the change of loops. This is

because NEXT is channel independent,

while FEXT is channel and couple length

dependent.

Since equalizer effect is not only on

the channel and signal but it also affect the

received noise and its residual ISI cause

noise, it called ISI noise.

C.Channel Equalization ResultsBy implementing MSSE an equalizer with

19 tap using MMSE algorithm was used to

shorten the CSA loop-1. The equalizer

shortens most of the channel impulse

response to the required length (v+1), the

remaining impulses will cause ISI as shown

in Fig 8. Table 4 shows the equalizer

efficiency for all CSA loops.

Table 3. CSA Loops NEXT and FEXT

Power.

CSA

Loop

NEXT Power

(dBm)

FEXT Power

(dBm)

10

Users

24

Users

10

Users

24

Users

Loop-1 -45.593 -43.28 -71.475 -69.143

Loop-2 -45.593 -43.28 -72.924 -70.5

Loop-3 -45.593 -43.28 -76.353 -74.157

Loop-4 -45.593 -43.28 -72.17 -69.86

Loop-5 -45.593 -43.28 -72.58 -70.373

Loop-6 -45.593 -43.28 -71.7 -69.497

Loop-7 -45.593 -43.28 -68.261 -65.981

Loop-8 -45.593 -43.28 -68.901 -66.60

Figure 7. AWGN, FEXT and NEXT Power

Spectrum.

It is noticed that the equalizers efficiency is

less than that given in section II, since the

equalizer efficiency depends on the


13/15

transmitted power, AWGN power, NEXT

users and FEXT users. Changing of these

parameters will change the equalizers

efficiency.

Figure 8. Comparison between Channel

and Equalized Channel Impulse Response.

Table 4. CSA Loops Equalizer Efficiency.

CSA Loop

Name

Equalizer Efficiency

( ) (%)Loop-1 81.2

Loop-2 89.8

Loop-3 93.1

Loop-4 80.0

Loop-5 85.1

Loop-6 87.1

Loop-7 70.8

Loop-8 85.1

D.Channel Analysis ResultsThe used transmitted signal spectrum

during the SNR calculation is shown in Fig

9. The average transmitted is 20 dBm

which is the maximum allowabletransmitted power.

The signal to noise ration SNR can

be calculated using Eq. 9 and Eq. 11

for and respectively.

Using Eq. 10, the Geometric MFB

(Match Filter Bound) SNR () forCSA loop-1 is 40.182 dB, Geometric

Equalizer SNR ( ) is 32.58 dB andthe Equalizer Efficiency is 81.2 %. Table 5shows the achieved equalizer geometric

SNR ( ) and Match Filter boundSNR ( ) for all CSA loops.

Figure 9. Transmitted Signal Power

Spectrum.

Table 5 CSA Loops Calculated and .CSA

Loop

Geometric

SNR

( )(dB)

Match

Geometric

SNR

(

)

(dB)Loop-1 32.58 40.123

Loop-2 39.57 44.1

Loop-3 36.5 39.21

Loop-4 31.36 39.2

Loop-5 34.39 40.41

Loop-6 33.31 38.24

Loop-7 26.96 38.08

Loop-8 29.8 35.01


14/15

The loops equalizer output SNR

depends on the equalizer efficiency and

more efficient equalization algorithm canachieve better SNR.

.

E.Bit Loading Table ResultsThe bit loading table is calculated by using

Eq. 13 and applied the rules of

ADSL/ADSL2. Using Eq. 15 the total bit

rate for CSA Loop-1 is 6.62 Mbps. Table 6

shows a comparison between ADSL andADSL2 Total Bit Rate.

Table 6 Comparison between ADSL and

ADSL2 Total Bit Rate over CSA Loops.

CSALoops

ADSLTotal Bit Rate

Rb (Mbps)

ADSL2Total Bit Rate

Rb (Mbps)

Loop-1 6.620 7.08

Loop-2 9.02 9.42Loop-3 7.94 8.36

Loop-4 6.2 6.68

Loop-5 7.22 7.68

Loop-6 6.86 7.32

Loop-7 4.6 5.09

Loop-8 5.66 6.07

It is noticed that ADSL2 achieves

better bit rate than ADSL in about 400Kbps which is about 1.1% because it havehigher code gain and more flexible bit

loading rules.

V. ConclusionModling standard loops has important role

in expecting the performance of the systems

before real system installation, it provide

information about potential system

performance, Time Domain Equalizer

(TEQ) can shorten the impulse response todesirable system impulse response length,

for the MMSE algorithm as TEQ algorithm,

TEQ can achieve up to 93% of zero ISI

channel. Also Time Domain Equalizer

(TEQ) increase noise level especially

higher frequencies which cause BER, usingadaptive TEQ can mitigate noise

enhancement, Noise effects on bit rate,

where higher AWGN level, more FEXTand NEXT interfered users causes higher

noise power which leads to reduce bit rate.

NEXT is the dominant channel impartment

compared to AWGN and FEXT, whereNEXT have higher power and narrow

bandwidth, Bit loading algorithm determine

the number of bits that can be carried on

each subchannel, using Default bit loading

algorithm ensure equal probability of error

on all subchannels, where it is fast andsimple and can be done offline, ADSL2

achieve higher bit rate than ADSL by more

than 400 Kbps on CSA loops also ADSL2

have higher code gain and more flexible bitloading role.

VI. References

[1] Angele A. and Lennard G. Kruger, Broadband Internet Regulation and Access:

Background and Issues, CRS Issue Brief for

Congress, USA, June, 2006.

[2] M. Rehman, "DSL vs. Cable Modem",

Research Paper, North Dakota State

University, March 2000.

[3] M. Ding, Channel Equalization To Achieve

High Bit Rates In Discrete Multitone

Systems, Ph.D. Thesis, University of Texas


15/15

at Austin, August 2000.

[4] D. Daly, "Efficient Multi-Carrier

Communication on the Digital Subscriber

Loop", Ph.D. Thesis, Dept. of Elect. AndElectronic, University College of Duplin,

Ireland, May 2003.

[5] M. Schlegal, "High Bit Rate Data

Transmission over the Telephone Loop Plant

Emphasizing on DMT Modulation Scheme",

Diploma Thesis, Dept. of Electronics,

Technological Educational Institute of

Piraeus, 2001.

[6] Z. Cai , Enhancement of ADSL designthrough concatenated coding, Elsevier

Science, Signal Processing 81 (2001) 2187

2199, December 2000.

[7] Ian C. Wong and Brian L. Evans, High-speed

Wireline Communication Systems, literature

survey Report, Embedded Signal Processing

Laboratory, University of Texas at Austin,

May 2005.

[8] . Z. H. Muhssen, "Performance Evaluation of

modulation Techniques in Asymmetric digital

subscriber line", M.Sc. Thesis, Electronic and

Communications Engineering, University of

Al-Nahrain, Iraq, September 2005.

[9] NEXTEP Broadband Networks Group, "xDSL

Modulation Techniques: Methods of

Achieving Spectrum-Efficient Modulation for

High Quality Transmissions", A Nextep

Broadband White Paper, May 2001.

[10] T. J. Willink and P. H. Wittke, Optimization

and performance evaluation of multicarrier

transmission, IEEE Trans. Inform. Theory,

vol. 43, no. 2, pp. 426440, March 1997.

[11] J. Campello, Practical bit loading for DMT,

IEEE Int. Conf. Communication, Vancouver,

June 1999, pp. 801805.

[12] D. Daly , C. Heneghan and A.D. Fagan,

Power- and bit- loading algorithms for

multitone communications, IEEE Int.,

Signal Process., Analysis, Rome, Italy,

September 2003.

[13] Dr. B. Z. Hassan, "Error Detection and

Correction via Reed-Solomon Codes",

Lecture, Al Nahrain University, Iraq, 2004.

[14] Comtech AHA corporation Primer, "Reed-

Solomon Error Correction Code (ECC)". AHA

Application Note, Moscow, 2002.

[15] J. A. C. Bingham, "ADSL, VDSL, and

Multicarrier Modulation", John Wiley andSons, Inc, 2000.

[16] D. G. Messerschmitt and A. Salvekar,

Linemod software for trans-mission line

analysis", Dept. of Electrical Eng. and

Computer Sci., University of California,

Berkeley, Internet Web Site,

http://www.stanford.edu/Cioffi/linemod/li

nemod.html.

[17] F. Van der Puteen, Alcatel Telecom and F.

Wellesplein, "Network and Customer

Installation InterfacesAsymmetric Digital

Subscriber Line (ADSL) Metallic Interface",

ANSI T1.413 Issue 2, June 1998.
http://www.stanford.edu/Cioffi/linemod/linemod.htmlhttp://www.stanford.edu/Cioffi/linemod/linemod.htmlhttp://www.stanford.edu/Cioffi/linemod/linemod.htmlhttp://www.stanford.edu/Cioffi/linemod/linemod.htmlhttp://www.stanford.edu/Cioffi/linemod/linemod.html

Documents

hothaifa paper2