During class please switch off your mobile, pager or other that may interrupt.
<number>
Today we are facing 2 problems.
(1) The voice network, specifically designed to transport voice (using circuit switching) is not ideal to transport data. Mainly because of the bursty nature of data communication and the limited capacity of the voice network (64 kbps). That is why a B-ISDN was created.
(2) slow access: the capacity of an analogue modem is limited to about 30 kbps.
The solution to this problem is ADSL, where we extend the used frequency band to about 1MHZ.
V.90 technology is able to accelerate data downstream from the Internet to your computer at speeds of up to 56Kbps.
We introduce the concept of FDM (Frequency Division Multiplexing).
Apart from the traditionally used frequencies over the UTP (300 - 3400 Hz), we start using higher frequencies for the ADSL up- and downstream channel.
As ADSL is an asymmetrical service with more capacity in the downstream direction, we need more frequencies in this direction.
It will be explained that the highest frequencies result in problems. The capacity of transporting data decreases with the increase of the used frequencies. In other words it is a pity that the frequencies reserved for POTS cannot be used for ADSL. POTS frequencies cannot be used for ADSL since the Life Line concept is still valid. The latter means that one should be able to make a phone call in case of a power failure.
VoDSL (making phone calls over the ADSL signal) does not support this Life Line concept.
ADSL speeds would increase when ACTIVE filters (splitters) were used. Because of the Life Line concept only PASSIVE are allowed since the active splitters contain OPAMPS that need to be powered.
Voice & data are transported over the same copper wire simultaneously in both directions (full duplex).
The ADSL signals travel between the Central Office (CO) and the ANT (ADSL Network Termination).
A wide range of filters/splitters is available on the market today adapted to local situations. From country to country complex impedance’s can differ as also the physical presentation of the local line.
In ADSL the voice is send in a separate part of the spectrum (FDM). The signal remains analog. The disadvantage is the need of a Pots Splitter and the fact that you have only one line.
In the case of VoDSL the telephones (up to 16 ) are connected to the VoDSL modem and the voice is multiplexed on an ATM connection over ADSL. All 16 phones can be used at the same time for different calls.
VoDSL modems these days implement 4 or 8 telephone ports.
The limitation of 16 is due to the limitation of the upstream ADSL bitrate. In the case of SHDSL the number of phones could be increased up to 32.
In the way the voice and the data traffic are combined. This is voice / data convergence in the access network only.
In most countries local regulations imply that the lifeline service must be available at all times. As our ADSL modem is locally powered a power outage will result in loss of our ADSL connection. This means that we will still need our POTS service even with VoDSL. At the moment R&D centres are looking at remote powering the ADSL modem so that ADSL would keep functioning and VoDSL is still possible. This will then make our POTS band dispensable meaning we would also be able to use those frequencies for xDSL communication.
The slow and inefficient data communication is not the only problem. Since we currently transport all data via the voice network our voice network is becoming overloaded.
(Nice to know that an average phone call lasts about 3 to 4 minutes whereas an average surf session lasts about 20 minutes)
To relieve the voice network we redirect the data in an ASAM (multiplexer) towards an ATM network.
For ADSL modem-to-modem communication we have following concepts:
ATU-C & ATU-R: ADSL Terminal Unit Central and ADSL Terminal Unit Remote
ADLT & ADNT: ADSL Line Termination & ADSL Network Termination
Both concepts apply to the ADSL modem.
This slide shows the ever growing demand for more bandwidth and the technologies to deliver this bandwidth.
FTTU is not actually considered a DSL solution because it’s an optical transmission technology. However, it’s the next (and final ?) step in bringing optical fiber to the customer premises.
VDSL is a technology that can offer both symmetrical and asymmetrical services.
Downstream bitrates up to 55 Mbps can be achieved on short loops.
VDSL is a technology to deliver high quality video streams and other bandwidth hungry services to the user.
Because of the high frequency range, the total power budget for VDSL lines is limited to avoid crosstalk towards longer ADSL lines.
These are the supported band plans for VDSL transceivers. They implement the zipper concept, which is a theoretical solution for using each other carrier for up and downstream traffic.
VDSL uses FDD, frequency division duplexing.
Current technology provides bitrates of 622 Mbps down, 155 Mpbs up.
Some scenarios deliver 622 Mbps bi-directional or even 2Gbps. This however depends on the hardware. Consult the roadmap for up-to-date information.
High quality transmission because of the all digital loop.
Low installation : installing the fiber isn’t expensive anymore nowadays. Also the operating costs are lower than for an electrical transmission because you don’t need repeaters (externally powered).
FTTU is the solution where the optical fiber runs all the way to the customer (bottom scenario in the slide above).
The figure above also depicts a PON (Passive Optical Network).
The split between the frequencies used for the upstream direction and the frequencies for the downstream direction is at 138kHz.
How to understand the difference between symbol rate and bit rate?
Assume simple Amplitude modulation => when you want to send digital information over a line you can transmit a bit over the line represented by a certain voltage level, for example +3v to represent a logical 1 and -3V to represent a logical 0.
Symbol rate = symbols per second (1/Ts) in baud
Bit rate = bits per second in bps
When representing 1bit by a certain voltage level the symbol rate = the bit rate. (Rs=R)
When adding more voltage levels you can actually specify more bits per symbol, for example +3V represents the logical bit-sequence 11, +1V represents logical 10, -1V represents logical 01 and -3V represents the logical bit-sequence 00
Here you have actually put 2 bits in a symbol and that way doubled the bitrate (R). On the other hand the symbol rate (Rs) in baud has remained the same.
The maximum attainable data speed depends on the signal to noise ratio (SNR).
The higher the (allowable) signal strength, and the lower the amount of noise on the line, the higher the capacity.
Unfortunately a lower noise level requires high quality lines which are expensive or not available.
On the other hand the signal strength is limited (governmental constraints) in order to limit the amount of crosstalk.
A decreasing signal to noise ratio will result in more bit errors (BER) on the line but with current technologies it is possible to detect AND correct these errors up to a certain level. It can be said that by introducing these error detection/correction mechanisms we can actually increase the capacity of the line for a certain SNR and BER.
The figure illustrates the distance dependency of the Shannon/Hartley capacity theorem.
Since the attenuation (signal loss) increases with distance (cable length) the maximum data speed drops with the distance.
Theoretically ADSL can reach a downstream capacity of apc. 15Mbps at 0km. Although in practice this is limited to 8,1Mbps
The signal strength is not only distance dependent.
It is also frequency dependent. This is due to the skin effect.
R = resistance ()
= resistivity (m)
d = distance, length of conductor (m)
Seff = effective cross sectional area of conductor (m2)
What means a loss of 32 dB at 150 kHz?
Attenuation (dB) = 10 x log10 (P1/P2)
P1/P2 = Inv log10 (Attenuation / 10)
P1/P2 = Inv log10 (32 / 10) = 1585
this means that our received pulse is 1585 times smaller !!
What means a loss of 55 dB at 150 kHz?
Attenuation (dB) = 10 x log10 (P1/P2)
P1/P2 = Inv log10 (Attenuation / 10)
P1/P2 = Inv log10 (55 / 10) = 316228
this means that our received pulse is 316228 times smaller !!
Higher frequencies suffer more from attenuation than lower ones, since the skin effect has more impact on the higher frequencies.
That is why the upstream signal is less attenuated than the downstream channel.
In some countries it is common practice of splicing a branching connection (bridged tap) onto a cable. Thus a bridged tap is a length of wire pair that is connected to a loop at one end and is unterminated at the other end. Approximately 80% of loops in the US have bridged taps; sometimes several bridged taps exist on a loop. One reason for a bridged tap is that it permits all the pairs in a cable to be used or reused to serve any customer along the cable route. Most countries in Europe claim to have no bridged taps, but there have been reports of exceptions.
The reflection of signals from the unterminated bridged taps results in signal loss and distortion.
When 3 is asking for phone call service a main cable is put in the ground.
Later on when 2 and 1 ask the same thing a lateral is drawn from the main cable.
Imagine 2 decides to move out (people in America frequently move) and thus cuts the lateral. Since there is no longer a telephone connected to the lateral there is no nice power dissipation anymore. This results in reflections; reflections that travel in both directions of the main cable.
A telephone cable contains up to several thousand separate wire pairs packed closely together. The electrical signals in a wire pair generates a small electromagnetic field, which surrounds the wire pair and induces an electrical signal into nearby wire pairs. The twisting of the wire pairs reduces this inductive coupling (also known as crosstalk), but some signal leakage remains.
NEXT (near end crosstalk) is a major 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 as the transmitter that is the noise source. In other words NEXT means that the receiver of one twisted pair picks up noise transmitted (by the transmitter) another one.
Transmission systems can avoid NEXT by using different frequency bands for upstream and downstream transmission. FDM (Frequency Division Multiplexed) systems avoid NEXT from like systems (also known as self NEXT). FDM systems still must cope with NEXT from other type of systems that transmit in the same frequency band, and other phenomenon known as FEXT (far end crosstalk).
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 severe than NEXT because the FEXT noise is attenuated by traversing the full length of the cable.
A major advantage of fibre optic transmission is the lack of any crosstalk.
Conclusion: NEXT is worse than FEXT for systems sharing the same frequency band in the up and downstream direction. Alcatel ADSL uses FDM and thus avoids NEXT. Here FEXT becomes the critical impairment.
When ADSL is mixed with other systems in the same cable then NEXT can occur because of overlapping frequency ranges! (see next slide)
ADSL over an ISDN line uses another spectrum than ADSL over a POTS line. When both systems are used within the same bundle of a cable Near End Crosstalk (NEXT) will occur.
Looking at an analogue signal, described by means of a sine function, modulation techniques exist by varying amplitude, phase, frequency or by a combination of these.
QAM is a modulation technique where both amplitude and phase are modified.
The amount of bits we can put on 1 symbol depends on the amount of amplitude and phase levels we distinguish. The latter is reflected in the constellation grid as shown on the slide.
Since 16 points are distinguished there are 16 different combinations of amplitude and phase.
The amplitude is the length of the vector whereas the phase is measured counter clockwise from the x-axis towards the vector.
In the example we can put up to 4 bits in 1 symbol, or in other words 4 bits are needed to construct 1 symbol. (4 bits allow you a specific bitsequence for each of the points (24)in the constellation)
Increasing the number of bits per symbol results in a higher attainable data speed.
Because the transmission lines are far from perfect noise is picked up during transmission. This results in a deformation of the analogue signal. In other words the analogue signal arriving at the other side of the transmission line might have a (slightly) different amplitude and phase.
From the constellation grid one can see that if the deformation of amplitude and/or phase is too big, mistakes might occur.
There is always 1 dot in the constellation grid that is closest to the reconstructed vector at the receiving end.
Imagine that 1001 was intended but that because of deformation the reconstructed vector is closer to 1011, the latter one is thought of to be the correct one.
Expanding the constellation grid would fulfil 2 desires:
1) Increase the data speed by putting more dots in the constellation grid. But because the grid will become more dense it is much more likely to make mistakes. Expanding the grid is a solution.
2) Imagine that I would like to put the same amount of bits per symbol under all line conditions. The more noise, the more mistakes! Expanding the grid, depending on the line conditions would be ideal!
Unfortunately the signal strength (signal power,amplitude) is limited because of cross talk limitations. This limitation corresponds with a maximum radius of the circle that describes the amplitude of vectors constructed in the constellation grid.
That is why noise and attenuation are measured first to determine how many bits one can put on the line.
If we take a closer look at the dots in the constellation grid, we see that 2 adjacent dots do not differ more then 1 bit from each other. Just to make sure that in case of an error the error is small (single bit error). Dots will differ in more bits when further apart in the constellation.
Imagine a certain frequency spectrum that is divided in 3 subchannels. To every of these 3 subchannels we can assign an appropriate QAM scheme depending on the measured SNR. The sum of these QAM signals is made and then sent to an DAC (Digital Analogue Converter). The outcome is an analogue signal that is put on the transmission line.
Dividing the frequency spectrum in subchannels can be done in different ways.
Or we can consider only a few subchannels but then we loose the flexibility of assigning different amount of data bits to relatively small fractions of the frequency spectrum; or we could consider an enormous amount of subchannels. This results in increased complexity.
255 subchannels is a nice trade off.
Since the attenuation increases with the frequency (skin effect) the SNR drops with increasing frequencies.
Thus less bits/carrier can be assigned for the upper subchannels.
This explains why it is useless to consider frequencies above 1.1 MHz.
Considering frequencies above 1.1 MHz is done in VHDSL (Very High Speed DSL). Since we have to compensate for the skin effect, in this application the distance has to be limited.
What about Discrete Multitone (DMT)?
Dividing the used frequency spectrum in subchannels results in the possibility to assign a different QAM scheme on every subchannel. So depending on the SNR of a specific carrier, more or less data bits can be transported.
We will always assign less bits to the carriers then allowed by the measured SNR. Typically we assign an average of 2 bits less.
This margin is called the Target Noise Margin and is configurable via the AWS. You specify what the average margin should be right after startup in dB. The modem measures the available SNR, then subtracts the Target Noise Margin, and then checks to what constellation it would fit. By default the TNM is 6 dB.
Why?
In case of a disturbance (e.g. RFI), we do not want our overall speed to drop. Whenever a subchannel becomes unavailable to transport data bits the spare bits in adjacent subchannels will be used. (see next slides)
It is not fully correct to speak about QAM constellations and bits here. It should be noise margins. The rule is : the ATU tries to equalise the noise margin over all tones, what could lead to bits being moved from one tone to another, and even tones to disappear or to appear. Margins are "analog" where bits are "digital". As you need 3 dB for 1 bit, a margin of 4.7 dB over all tones can not be translated into bits.
Modem does not calculate with bits. It measures the available SNR, then subtracts the Target Noise Margin, and then checks to what constellation it would fit (table slide 31). There is also the possibility to increase (or decrease) the gain of each tone (+ -2.5 dB, but sum over all tones must be 0) for those tones which have just a little bit shortage to have 1 bit more loaded. A gain change results directly in signal change, so also SNR change.
The ATU tries to equalise the noise margin over all tones.
Modems using the ANSI ADSL standard locks to a carrier in the up - and downstream direction, i.e. the ATU-R locks to the pilot (carrier 63) in the downstream part and the ATU-C locks to the pilot in the upstream part. ANSI has 2 pilot carriers, 1 in the downstream and 1 in the upstream.
Modems using the ITU-T ADSL standards only require the ATU-R to lock to the ATU-C and also use this lock for his upstream transmission.
ERROR CORRECTION/DETECTION PRINCIPLE
Imagine that the green spots on the slide represent a bitsequence that is used to transmit data. If during the transport bit errors occur (single or multiple) they will result in one of the red invalid bitsequences. At that moment the receiver can execute bit error correction by locating the nearest green spot (see example above). In this way it is possible to correct up to 2 bit errors.
If a 3 bit error occurs, it cannot be corrected because then the distance to the green spots (left and right) is equal. In this case only an error detection is possible. Even worse, when more than 3 bit errors occur, it is possible that the other side receives a “red” bitsequence that is closer to another green spot. In this case a wrong correction is executed and the error correction code fails.
The error correction possibility can be calculated as follows:
- DISTANCE between the green spots = 6 = 2 t + 1- Number of errors than can be corrected = t = 2
The distance between the valid data sequences can be increased by adding check bytes to the original data (see Reed Solomon, Trellis, …)
ERROR DETECTION PRINCIPLE
In the previous example a 3 bit error was the maximum allowed. More than 3 bit errors will result in wrong corrections. It is possible to increase the number of bit errors that can be detected in a simple way. If both sides agree not to execute error corrections, up to 5 bit errors can be detected (because the distance is 6). In other words, all red spots will result in an error detection. The disadvantage is that there is no error correction possible.
EXAMPLES
ADSL (Reed Solomon and Trellis) always works in correction mode
ATM (HEC code) uses correction mode and will temporary switch to detection mode when errors occur.
Because of the presence of impulsive noise (infrequent high amplitude bursts of noice mostly caused by central office switching transients, dial pulses, POTS ringing, neighboring railway stations, elevators,…that can corrupt the signal beyond recognition) on the twisted pair wire means must be implemented to make sure that the ADSL transceiver is sufficiently robust against this impulsive noise and to maintain an acceptable BER for good quality of services.
Since errors often come in bursts it is likely that the error correction possibility is insufficient.
Interleaving:
In stead of transmitting our RS code words directly on the line we will create a frame of the same size made up by multiple RS words by taking only a portion of each of the original RS words. This has the advantage when a burst of errors occur on the line and the original RS words are recreated on the receiving side, the errors will be spread over multiple RS words. This could mean that we are able to correct the errors within a single RS word if the number of errors are within the RS correction boundaries.
Interleaving depths:
The main disadvantage of interleaving is the high delay. Constructing the blocs that will finally be transmitted over the line take time as you have to wait for a time before you can actually start transmitting. In our example here we need 15 original RS words (1 Byte of each) before we can construct the first block that is actually transmitted over the line.
At the receiving side it will also cost extra time to reconstruct the original RS word. The first original RS word can not be reconstructed before we have received all the bytes of this first RS word.
Interleaving can be sped up by using different depths, i.e. by taking bigger chunks of the original RS words you will be able to construct your first bloc for transmission much quicker. This has the disadvantage that errors will be spread over less RS words on the receiving side with the possibility that they can not be corrected.
Alcatel introduces 3 interleaving depths: high, medium and low.
Where high uses the smallest portions to construct a frame for transmission, consequently the high interleaving depth will introduce the highest delay as this one will need more frames to construct/reconstruct before transmission/reception.
¡¡¡On the ISAM and on ASAM release 4.6 and above the Interleave time can be set in miliseconds!!!!
Trellis will also add extra information (overhead) to the transmitted data in order to perform his detection and correction.
This overhead is apc. 1/2 bit per used carrier + 4 bits.
A Viterbi coder/decoder will look at the received data and calculate what the most logical value should be for the received data similar to spoken language as illustrated on the slide.
With Trellis used; extra bytes will be needed in the DMT symbol.
For fast (no interleaving) there is exactly one RS word in one DMT symbol. Reed Solomon always performs his calculation on a word of 255 Bytes which is also the maximum length. Our user data (ATM cells) always run through this Reed Solomon coder giving you the maximum speeds as calculated on the slide.
Some payload Bytes of the RS word are used for maintenance communication. Over these Bytes, which are negotiated at start-up, following information can be transmitted
1) EOC: Embedded Operations channel
Is intended to exchange physical layer messages between the ATU–C and the ATU–R.
For example what the SNRs are, or what the output power is of the transmitter, ...
2) AOC: ADSL Overhead Control channel
The ADSL Overhead Channel is intended for bit swap operations
3) CRC: Cyclic Redundancy Check
The transmitter shall calculate a CRC over the data of last superframe, and send this CRC to the receiver. The receiver also calculates the CRC over the same data block, and compare it to the received CRC value. When those do not correspond, a CRC error event shall be generated.
4) IB: Indicator Bits
They indicate a number of states and alarms like CRC errors, loss of signal, loss of frame, etc.
5) Dummy information
The numbers given a relative and not exact. In ADSL implementations the numbers will differ a bit depending on the QAM desired and the product manufacturer.
Relative Capacity Occupation (RCO in%): proportion of the used line rate to the attainable line rate
Reed Solomon has to be supported by the ADSL modems according to the standardisation. This does not mean that Reed Solomon must be activated at all times.
Trellis on the other hand is optional!
During start-up of the ADSL modems capabilities are transferred between the modems and based on this the ATU-C (in the ASAM) decides what is used. Also during this phase it is communicated how many bytes are used in the framing overhead, this is mostly only 1Byte to have the most optimal performance.
Example for downstream: (taken from ADSL diagnostics)
ATM attainable rate:8,128 Mbps[(255-1) x 4000 x 8]
ATM used rate:8,128 Mbps
Used line rate:8,612 Mbps
Attainable line rate:10,236 Mbps
RCO:84%[(8,612 / 10,236) x 100]
Short haul local loop with Reed Solomon disabled (used ATM rate = attainable ATM rate) and Trellis activated (Used line rate >> {ATM used rate + modem-modem communication overhead}). And there’s 1 Byte of framing overhead.
During class please switch off your mobile, pager or other that may interrupt.
Course objectives:After attending this course, you will be able to:
describe the features of ADSL2, ADSL2+ and READSL2
compare the features of ADSL2, ADSL2+ and READSL2 to the ones of ADSL
tell in which cases it is interesting to offer ADSL2, ADSL2+ and READSL2 and why
Entry level requirements:
You know the basics of ADSL (IBCN0370E)
Suggested duration:
0,5 day (or 3,5 hours)
Normal class hours:08:30h => 12:00h
13:00h => 16:30h
Overview
ADSL Flavors: Next Generation ADSL Technologies.
We start with a small summary of the features and restrictions of ADSL and we give an overview of the new standards.
Then we describe the improvements of ADSL2. This is the largest part of this course.
ADSL2+ is a delta on ADSL2. It uses a broader spectrum up until 2.2 MHz and can be very interesting for a.o. deployment of remote cabinets.
READSL2 (= reach extended ADSL2) is described in Annex L of the ADSL2 standard. It uses less spectrum than ADSL, but the PSD-level (PSD = Power Spectral Density) is increased. The total power upstream remains the same; the total power downstream is even decreased with regard to the power downstream annex A. With READSL2, operators can e.g. offer (better) coverage to people living 5 to 6 km away from the DSLAM. People living further away from the DSLAM than e.g. 5.4 km wouldn’t have got ADSL in the first place, but now they can get READSL2 (this means: more customers for the operator!).
Multi xDSL cards (with or without video optimization) will be implemented: ADSL, READSL2+, ADSL2 and - in case of video optimization - also ADSL2+ will be supported. On the other hand, for VDSL, you still need another line card. In the future also multi A/V-cards will exist.For SHDSL, you also need dedicated cards.
ADSL has been implemented for quite some years now.
Field experience has brought some new ideas. The restrictions of ADSL are known and improvements were suggested. You may notice that many of the recently standardized features were already implemented by Alcatel for ADSL, but then they were proprietary solutions. Now it’s all standardized in the new standards.
Two new standards have been approved: one for ADSL2 and one for ADSL2+.READSL2 (reach extended ADSL2) is annex L of the ADSL2 standard.
ADSL reuses the existing copper wire. Operators don’t need to install e.g. fiber to the users. This represents significant savings: no installation cost (labor).
High Speed Data (Internet) Access
Maximum 8 Mbps downstream (6,144 Mbps downstream is mandatory)(up to 12 Mbps with S = 0.5. This is optional in G.dmt)
Maximum 1.5 Mbps upstream (640 kbps upstream is mandatory)
Data and voice use separate networks (no PSTN blocking), but physically on the same twisted pair (access).
Data and voice can be used simultaneously (FDM): for telephony the lower frequencies are reserved:
for POTS frequencies from 300 Hz until 3400 Hz are used. The first upstream carrier that can be used for ADSL is at 30 kHz.
For ISDN frequencies until 80 kHz are used. The first upstream carrier that can be used for ADSL is at 138 kHz.
Maximum Distance
Max. 5 km (no repeaters are possible)
Virtually any range is possible if subtending is used.A DSLAM can subtend several other DSLAMs. This means one DSLAM acts as a hub to a number of smaller subtending DSLAMs. By installing these remote subtending DSLAMs, ADSL is brought closer to the subscribers.
Discrete MultiTone (DMT) modulation: different QAM modulation schemes at different tones/carriers, depending on the measured signal to noise ratio.
Reed Solomon (mandatory) and Trellis code (optional) are used for error detection/correction.
ADSL uses a physical point-to-point connection (security) instead of a shared medium like cable networks.
Legend:
Grey zone border defined by Tier 2 reach from CO.
Red zone border defined by Tier 1 reach from CO.
The access evolution challenges are:
increase the bandwidth (new applications: e.g. more video channels)
increase the coverage (new customers)
Example: an operator offers an ATM rate of 3.5 Mbps in Tier2. If you subtract the overhead, this leaves 3 Mbps for IP traffic.
ADSL can offer a bandwidth that is broad enough to offer one video channel, if the subscriber doesn’t live too far from the DSLAM.
Marketing studies point out that in the future one video channel will not be enough anymore; three video channels would be preferred. To offer three video channels to subscribers, ADSL will not do.
Operators would like to offer xDSL to all subscribers, also the ones living further than 5 km from the DSLAM. Reach Extended ADSL 2 can be a solution for the people at a distance from 5 to 6 km from the DSLAM (this gain may not seem that impressing, but it can mean quite a lot of new potential subscribers!). But of course we couldn’t expect miracles here (Shannon’s limit).Another way to serve people at a larger distance is to bring ADSL closer to the people, by deploying remote cabinets etc.
<number>
ARPU: Average Revenue Per User
In the table you see the different bitrate corresponding to the different tiers. Next to it you see what that bandwidth can be used for: e.g. only high speed internet access in tier 1; high speed internet access plus one video channel in tier 2; etc.
Not only the downstream bitrate is important; also the upstream requirements change. Upstream requirements could vary from 0.5 Mbps to 1.5 Mbps for symmetrical high speed internet access consumer services.Business customers could be willing to pay for higher upstream bandwidth.
<number>
This slide shows that in Europe, operators already cover a high percentage of xDSL users at a distance of 2 km from the DSLAM.
In Italy, operators already reach 65% of the subscribers at a distance of only 1.5 km (within that range, you can also offer VDSL)! At 3 km 95% of the subscribers is reached. At 4 km you have full coverage. As a consequence, there won’t be any need for Reach Extended ADSL2 in Italy.
In the U.S. the situation looks quite different. At a distance of 1.5 km you only reach 15% of the subscribers; at a distance of 3 km you reach 40% of the subscribers; at 5 km you reach 80% of the subscribers.
In India you reach 40% of the subscribers at 1.5 km; 60% of the subscribers at 3 km and 70% of the subscribers at 5 km.
<number>
Deployment strategy :
Central office based : High coverage using the right xDSL technology coupled to service development ambitions
Remaining percentage of households can be flexibly covered
Via a geo-marketing approach with the right mix of remotes/DSL flavors:
Deploying remote DSLAMs
Reach Extended ADSL2
<number>
Based upon years of field experience with ADSL, some improvements to the standard were suggested and they are approved now. ADSL2 (G992.3) offers a better performance on longer loops and an improved OAM.
ADSL2+)with an extended spectrum with frequencies up until 2.2 MHz.This is a new standard: G992.5. At short distances (until 2 - 3 km) this means a boosted downstream bandwidth with bit rates up to 24 Mbps.Operators can also decide to use masks to avoid interference (so only a part of the frequency spectrum will be used). This can be interesting for remote cabinet deployment (see further).
READSL2: Reach Extended ADSL2. This is annex L of G992.3.In ADSL there are limits to the total power you are allowed to emit. If you divide the power between all the carriers, the power per carrier is smaller than when you omit certain carriers and send more power over the remaining ones. Stronger signals have a slightly longer reach. Operators can extend the reach with 600 to 900 m (up to 6 km). For some operators, this may represent a lot of new potential subscribers.
VDSL also uses DMT technology. Different band plans are possible, depending on what kind of traffic the subscriber wants to send and receive: symmetrical or asymmetrical traffic.Bandwidths can be huge at very short distances (e.g. up to 55 Mbps at 300 m from the ASAM).
ADSL technology uses existing copper telephone lines to provide better data transfer rates than dial-up. Though the standard is being widely used, it has one major drawback. It’s unable to provide consistent performance over longer distances. So, you’d get more out of an ADSL connection sitting close to an ADSL exchange than if you were sitting very far from it. To increase the bit rate (ADSL2(+))and/or the reach (READSL2) and to tackle other issues (like loop diagnostics, OAM…), a new version of the technology, called ADSL2, has been defined. ADSL2 is approved by ITU (International Telecommunication Union), and is designated as G.992.3 and G.992.4 (the latter is for ADSL2 Lite). Besides improvements in performance and data rates, the technology has features like rate adaptation, diagnostic and power management.
The improvements which are most important are related with the data rate and the power control.
With ADSL2 and its delta’s one can achieve higher data rates OR one can increase the coverage. Operators can decide (for certain subscribers) not to focus on high data rates but on longer distances. Then READSL2 is a good solution, but with this technology, one cannot reach very high data rates. The highest data rates at short distances will be achieved using ADSL2+.
Loop diagnostics: suppose the line is very bad and you don’t get a connection. The central office wouldn’t know the cause in ADSL, but with ADSL2 it is possible for an operator to find out what goes wrong.
Crosstalk issues arrising when remote cabinets are deployed can be solved using ADSL2+ and masks.
The power control handling is an important aspect for operators. First-generation ADSL transceivers operate in full-power mode day and night, even when not in use. With several millions of deployed ADSL modems, a significant amount of electricity can be saved if the modem engage in a standby/sleep mode just like computers. This would also save power for ADSL transceivers operating in small remote units and digital loop carriers (DLC) cabinets that operate under very strict heat dissipation requirements.
The G.dmt = G.992.1 standard describes the current ADSL technology.
The ADSL2 G.992.3 and G.992.4 standards, recently approved by the International Telecommunication Union, improve data rate and reach performance, dynamic rate adaptation and diagnostics, and include a power-saving standby mode as well. Another forthcoming standard, ADSL2+ (G.992.5), more than doubles the downstream data rate of ADSL to 16 Mbit/sec.
ADSL2 addresses the growing demand for bandwidth to support services such as video. Forthcoming ADSL2 and ADSL2+ gear will interoperate with existing ADSL equipment, allowing carriers to roll out new high-speed services while gradually upgrading their legacy infrastructure.
Reach Extended ADSL2 (READSL2), which is an annex of ADSL2 (G.992.3), results in an increase in reach of about 600 meters. With ADSL the reach was limited to an average of 5,4km. With READSL2 this is extended up to 6km (for 0,4mm loops).
For completeness, we also refer to the ADSL lite versions.
The lite version for the current ADSL is defined as G.lite or G992.2
For ADSL2 the lite version is defined as G.lite.bis or G.992.4
ADSL standards include annexes that specify ADSL operation for particular applications and regions around the world. Several annexes associated with the first ADSL standard will also apply to the ADSL2 standards family, including ADSL2+. Generally speaking, the annexes specify subcarriers (or tones) and their associated transmission power levels used for upstream and downstream transmission. This current slide summarizes the ADSL annexes.
ADSL Annexes A and BAnnex A specifies operation of ADSL using bandwidths higher than POTS, so as to preserve POTS operation. It is the most common application of ADSL, used throughout North America and much of Europe and Asia. Annex B works similarly to Annex A, though it is designed to work on phone lines enabled with ISDN instead of POTS, which is common in Germany and a few other regions. Both Annex A and Annex B have been extended to apply to ADSL2 and ADSL2+.
ADSL Annexes C and IAnnex C of the original ADSL standards (G.992.1, G.992.2) is for ADSL over TCM-ISDN (Japanese version of ISDN) or POTS (synchronized with the TCM-ISDN).A new annex to complement the original ADSL Annex C called “Annex I” doubles the downstream of the original Annex C, much like ADSL2+ doubles the downstream of ADSL2. This particular Annex I will not apply to ADSL2. Note that there are two “Annex I”s; one for the original ADSL, and one for ADSL2/ADSL2+. There is also a separate amendment to ADSL Annex C that improves its reach.
ADSL2 Annexes I and JAnnexes I and J for ADSL2 are for all-digital operation, meaning that traditional voice bandwidths are used by DSL to transmit digital data instead of for traditional voice. Annexes I and J are specified for all-digital operation in POTS and ISDN environments, respectively.
ADSL2 (G.992.3) Annex L : (READSL2)ADSL2 Annex L proposes new power spectral density (PSD) masks that can result in an increase in reach of about one to two kilofeet (300 to 600m) (26 AWG, 0.4mm loop). This Annex L is commonly called READSL2.
ADSL2 (G.992.3 and G.992.5) Annex M: Enhanced UpstreamAnnex M is proposed for ADSL2+ operation with an extended upstream in a POTS environment. Instead of using tones 32 through 64 for downstream as with Annex A, Annex M proposes use of tones 32 through 64 for upstream. Annex M applies to ADSL2 and ADSL2+.
<number>
Annexes I and J are for all digital operation, while Annex M Is over POTS.
Mnemonic: M = More upstream
up to tone 64
US masks similar to ADSL2 Annex J, but over POTS
Estimated US Bitrates up to 3 Mbps
For Annex L (READSL2), there’s a dedicated section (see further).
After years of field experience, many improvements to the ADSL technology were suggested. They are described in the ADSL2 G.992.3 standard.
Much has been learned over the past three years of ADSL deployments, including areas where improvements in the technology would be particularly valuable. There is a wide variety of improvements included in ADSL2, each with very different implications; some make the transceivers operate more efficiently, some make them more affordable, and some add functionality.
Carriers, service providers and subscribers have played a key role in the completion of ADSL2, having provided valuable feedback from the field that the ITU-T in turn incorporated into the standards in the form of new features and performance improvements. As a result, ADSL2 will be more user-friendly to subscribers and more profitable to carriers, and promises to continue the great success of ADSL through the rest of the decade.
You may notice that some of the ‘new’ features have been implemented for quite some time already (e.g. bit swapping was optional in G.992.1, but it was widely implemented).
Egress friendliness: you don’t emit signals on the frequencies used by radio amateurs to avoid interference.
ADSL2 has been specifically designed to improve the rate and reach of ADSL largely by achieving better performance on long(er) lines. ADSL2 accomplishes this by improving modulation efficiency, reducing framing overhead, achieving higher coding gain, improving the initialization state machine, and providing enhanced signal processing algorithms. As a result, ADSL2 mandates higher performance for all standard-compliant devices.
As a result, in ADSL2 the downstream/upstream data rate scales up to 15/1,5 Mbps with mandatory support of 8 Mbps/800 Kbps downstream/upstream.
S = number of DMT symbols per FEC codeword.The S-factor can vary.
In ADSL1 the S-factor is coupled to the DMT symbolS=1/2 in the interleaved DS path (is optional in ADSL1) (= 2 RS words in 1 DMT symbol)
In ADSL2 and ADSL2plus the S-factor can have any fractional value between 1/3 (for ADSL2+) and 64.
This slide shows the rate and reach of ADSL2 as compared to the first-generation ADSL standard. On longer phone lines, ADSL2 will provide a data rate increase of 50-80 Kbps, a significant increase for those customers who need it most. This data reach increase results in an increase in reach of about 250 feet (26 AWG).
On long loops and in an ADSL and Self crosstalk noise environment, ADSL2 will provide data rate increase of about 50-80 kbps which will result in a reach increase of about 80m .
Noise conditions for the diagram in above slide are 12 self crosstalkers (12 x ADSL) with –140dBm/Hz white noise at CO side and 6 self crosstalkers (ADSL) with also –140dBm/Hz white noise at CPE side.
ADSL2 provides better modulation and coding efficiency by:
mandatory trellis-coding (which was optional in the G.992.1).
mandatory 1-bit constellations which provide higher data rates on long lines where the signal-to-noise ratio (SNR) is low. In the current ADSL, when the SNR was low, the carrier was not used. In the case of ADSL2, now on such carriers are placed 1-bit of data. This gives per carrier a gain of (4000 symbols per second X 1bit) 4 Kbps. For example a low SNR related with 10 carriers results in a gain of 4Kbps X 10 carriers = 40 Kbps.
In addition, receiver determined tone reordering enables the receiver to spread out the non-stationary noise due to AM radio interference to get better coding gain from the Viterbi decoder.
Improved performance is also obtained by allowing data modulation on the pilot tone.
ADSL2 systems reduce framing overhead by providing a frame with a programmable number of overhead bits. Therefore, unlike the first-generation ADSL standards where the overhead bits per frame are fixed and consume 32 Kbps of actual payload data (1 byte overhead X 8bits x 4000 symbols/sec =32Kbps). In the ADSL2 standard the overhead bits per frame can be programmed from 4 to 32Kbps. In first-generation ADSL systems, on long lines where the data rate is low (e.g. 128 Kbps) a fixed 32 Kbps (or 25% of the total data rate) is allocated to overhead information. In ADSL2 systems, the overhead data rate can be reduced to 4 Kbps, which provides an additional 28 Kbps for payload data.
On long lines where data rates are lower, ADSL2 achieves higher coding gain from the Reed-Solomon (RS) code. This is due to improvements in the ADSL2 framers that improve flexibility and programmability in the construction of the RS codewords.
Determination of the pilot tone location by the receiver in order to avoid channel nulls from bridged taps or narrow band interference from AM radio.
Determination of the carriers used for initialization messages by the receiver in order to avoid channel nulls from bridged taps or narrow band interference from AM radio. In ADSL when carriers for the initialization are used which are disturbed by RF interferences, the initialization process of the modems can take a lot more time than in the normal case. With this new concept (receiver feedback concept), the initialization prccess will not be delayed anymore by RFIs.
The receiver can request the transmitter to disable some tones (subcarrier “blackout”) in order to enable Radio Frequency Interference (RFI) cancellation/measurement receiver algorithms during initialization and ShowTime.
Another major difference lies in the improved rate adaptivity concept. In the current G.992.1 initialization the selected downstream data rate is not the optimal one because the ATU-R receiver can only select the data rate out of 4 options given by the ATU-C. Due to the fact that the CO does not know all the details about the implementation of the CPE receiver, the 4 options are mostly below the optimum data rate. In ADSL2 the ATU-C only sends required constraints on the data stream characteristics (min data rate, max data rate, max latency, max BER) to the ATU-R, without specifying the exact data rate. The ATU-R receiver will select the optimal data rate/configuration taken into account the constraints given by the CO
Power cutback capabilities at both ends of the line to reduce near-end echo and the overall crosstalk levels in the binder.
As a summary we can say that the key message is: “The receiver gives feedback to the CO”, which results in an optimal initialization sequence and an improved performance.
First-generation ADSL transceivers operate in full-power mode day and night, even when not in use. With several millions of deployed ADSL modems, a significant amount of electricity can be saved if the modems engage in a standby/sleep mode just like computers. This would also save power for ADSL transceivers operating in small remote units and digital loop carrier (DLC) cabinets that operate under very strict heat dissipation requirements.
To address these concerns, the ADSL2 standard brings in two power management modes that help reduce overall power consumption while maintaining ADSL’s “always-on” functionality for the user. These modes include:
L2 low-power mode (keep alive mode)
Enables statistical power savings at the ADSL transceiver unit in the central office (ATU-C) by rapidly entering and exiting low power mode based on Internet traffic running over the ADSL connection.
L3 low-power mode (sleep mode)
This mode enables overall power savings at both the ATU-C and the remote ADSL transceiver unit (ATU-R) by entering into sleep mode when the connection is not being used for extended periods of time. When the user returns on-line, ADSL transceivers use a fast start-up (duration 3 seconds) to reinitialize and enter into show-time.
L0 : Full power modeL2 : Low power mode (keep alive mode)L3: Sleep mode.
This slide illustrates the Power Mangement concept.
The L2 low power mode is one of the most important innovations of the ADSL2 standard. When large files are being downloaded, ADSL2 operates in full power mode (called L0 power mode) in order to maximize the download speed. When internet traffic decreases, such as when a user is reading a long text page, ADSL2 systems can transition into L2 low power mode, in which the data rate is significantly decreased and overall power consumption is reduced.
While in L2, the ADSL2 system can instantly re-enter L0 and increase to the maximum data rate as soon as the user initiates a file download. The L2 entry/exit mechanisms and resulting data rate adaptations are accomplished without any service interruption or even a single bit error, and as such, are not noticed by the user.
The L3 power mode is a sleep mode where no traffic can be communicated over the ADSL connection when the user is off-line. When the user goes on-line again, the ADSL transceivers require approximately 3 seconds to re-initialize and enter into steady state communication mode. (Why is it possible to start up so quickly? The modem will make assumptions. In sleep-mode the modem will not perform all the necessary measurements and in 3 seconds the modem cannot make enough measurements! The modems will start at a certain rate (based on assumptions), continue measuring and adapt the rate accordingly (seamless rate adaptation - see further).)
Determining the cause of problems in consumer ADSL service has at times been a challenging obstacle in ADSL deployments. To tackle the problem, ADSL2 transceivers have been enhanced with extensive diagnostic capabilities. These diagnostic capabilities provide tools for trouble resolution during and after installation, performance monitoring while in service, and upgrade qualification.
In order to diagnose and fix problems, ADSL2 transceivers provide for measurements of line noise, loop attenuation, and signal-to-noise ratio (SNR) at both ends of the line. These measurements can be collected using a special diagnostic testing mode even when line quality is too poor to actually complete the ADSL connection.
Additionally, ADSL2 includes real-time performance monitoring capabilities that provide information on line quality and noise conditions at both ends of the line. This information is interpreted by software and then used by the service provider to monitor the quality of the ADSL connection and prevent future service failures. It can also be used to determine if a customer can be offered higher data rate services.
Bit Swapping (BS):
Is a mandatory receiver initiated feature to maintain the operating conditions of the modem (such as margin and BER) during changing environment conditions (changing narrowband interferers, temperature changes, water in the binder, changing crosstalk environment…). It reallocates the data bits among the allowed carriers without modification of the higher layer control parameters in the ATU (framing layer and above) . After a bit swapping reconfiguration the total data rate and the data rate on each latency path is unchanged. Bit swapping can only be used when the line conditions vary slowly and as long as the target noise margin is sufficient.
Dynamic Rate Repartitioning (DRR)
Is optional and used primarily to reconfigure the data rate allocation between multiple latency paths by modifying the framing multiplexer parameters. DRR does not modify the total data rate but does modify the individual latency path data rates. Because DRR is used in response to higher layer commands, DRR is an application option. The ability to support DRR is identified during the initialization procedure. DRR can be used for instance for channelized Voice over DSL (CVoDSL). When you don’t have a CVoDSL-call, you only have a latency path for data. As soon as a call is to be set up, a latency path for voice is needed as well and the total data rate must be repartitioned.
Seamless Rate Adaptation (SRA)
Is optional and used to reconfigure the total data rate by modifying the framing multiplex control parameters and modifications to the bits parameters. Since the total data rate is modified, at least one latency path (or more) will have a new data rate after the SRA. Because SRA is used in response to higher layer commands, SRA is an application option. The ability to support SRA is identified during the initialization procedure. SRA can be used for instance when the ADSL2 detects changes in the channel conditions — for example, a local AM radio station turning off its transmitter for the evening — and adapts the data rate to the new channel condition transparently to the user.
Both DRR and SRA can be transmitter or receiver initiated.
After start-up we will use a lower QAM then possible on most of the carriers We will assign less bits to the carriers than allowed by the measured SNR. Typically we assign an average of 2 bits less (as if the SNR were e.g. 6 dB less).
Example : QAM-4096 corresponding with 12 bits per symbol used QAM on that carrier : QAM-1024 (10 bits per symbol). This results in extra bits that could be allocated on that carrier
This noise margin is called the Target Noise Margin and is configurable via the AWS. You specify what the average margin should be right after start-up in dB. The modem measures the available SNR, then subtracts the Target Noise Margin, and then checks to what constellation it would fit. By default the TNM is 6 dB.
Why do we allocate less bits than allowed to the carriers?In case of a limited disturbance (e.g. RFI), we do not want our overall speed to drop. Whenever a carrier becomes unavailable to transport the foreseen number of data bits, the spare bits in adjacent carriers will be used. (see next slide)However, if the disturbance is very drastic and the noise margin is insufficient, bit swapping cannot solve the problem.
During showtime (modem operation), the SNR is measured on all carriers at regular intervals (default 1 sec)
if the SNR on a certain carrier degrades resulting in a lower QAM which can be used on that carrier, some bits of that carrier will be reallocated to other carriers where the maximum QAM is higher than the actual used QAM.
the modems will try to spread out the reallocated bits over numerous carriers.
It is not fully correct to speak about QAM constellations and bits here. It should be noise margins. The rule is : the ATU tries to equalize the noise margin over all tones, what could lead to bits being moved from one tone to another, and even tones to disappear or to appear. Margins are "analog" where bits are "digital". As you need 3 dB for 1 bit, a margin of 4.7 dB over all tones can not be easily translated into bits.
A modem does not calculate with bits. It measures the available SNR, then subtracts the Target Noise Margin, and then checks to what constellation it would fit.
Telephone wires are bundled together in multi-pair binders containing 25 or more twisted wire pairs. As a result, electrical signals from one pair can electro-magnetically couple onto adjacent pairs in the binder. This phenomenon is known as “crosstalk” and can impede ADSL data rate performance. As a result, changes in the crosstalk levels in the binder can cause an ADSL system to drop the connection. Crosstalk is just one reason that ADSL lines drop connections. Others include AM radio disturbers, temperature changes, and water in the binder.
ADSL2 addresses these problems by seamlessly adapting the data rate in real-time. This new innovation, called seamless rate adaptation (SRA), enables the ADSL2 system to change the data rate of the connection while in operation without any service interruption or bit errors. ADSL2 simply detects changes in the channel conditions and adapts the data rate to the new channel condition transparently to the user.
SRA is based on the decoupling of the modulation layer and the framing layer in ADSL2 systems. This decoupling enables the modulation layer to change the transmission data rate parameters without modifying parameters in the framing layer which would cause the modems to lose frame synchronization resulting in uncorrectable bit errors or system restart. The protocol used for SRA works as follows:
1. The receiver monitors the SNR of the channel and determines that a data rate change is necessary to compensate for changes in channel conditions.
2. The receiver sends a message to the transmitter to initiate a change in data rate. This message contains all necessary transmission parameters for transmitting at the new data rate. These parameters include the number of bits modulated and transmit power on each subchannel in ADSL multi carrier system.
3. The transmitter sends a “Sync Flag” signal which is used as a marker to designate the exact time at which the new data rate and transmission parameters are to be used.
4. The Sync Flag signal is detected by the receiver and both transmitter and receiver seamlessly and transparently transition to the data rate.
ADSL2 provides an optional short initialization sequence to allow the ATU’s to quickly enter Showtime from a L3 power management state or as a fast recovery procedure from error conditions during ShowTime. This will reduce the initialization time from more than 10 seconds (as is required for G.992.1 ADSL) to 3 seconds in these cases.
Any ATU that implements this optional short initialization procedure should implement also “Seamless Rate Adaptation”(SRA), which is one of the three possible “On-line Reconfiguration” procedures (explained before). This SRA can be used to further increase the data rate and optimize the ATU settings following a Fast Start-up , because the fast estimates made during this short training phase will most of the times not be optimal.
G.992.1 ADSL was designed to carry data in overlay with analog Plain Old Telephone Service (POTS) or ISDN signals over one single twisted pair. This is achieved by using separate frequency bands for POTS and ADSL data. A splitter based on Low Pass (LP) and High Pass (HP) filters separates both frequency bands.
An all-digital Loop system does not reserve the lower part of the spectrum for POTS or ISDN: the upstream Power Spectral Density (PSD) is downward extended. The extra tones will increase the capacity on all loops, since the lowest tones have very little loop attenuation. ADSL2 defines two “All Digital Mode ADSL” optional modes that allow for transmission of ADSL data in the POTS or ISDN frequency bandwidth, adding around 100/750 kbps of extra upstream data rate.
ADSL2 Annex I: “All Digital Mode ADSL with improved Spectral Compatibility with ADSL over POTS” is identical to an ADSL system but with a downward extended upstream frequency bandwidth using 31 tones in Upstream (tone 1-31 in stead of 6-31 for ADSL).The lowest frequencies suffer less from attenuation. Under optimal conditions, the SNR is so good that 15 bits per carrier can be sent for those 5 extra upstream carriers: 5 carriers * 15 bits per carrier per symbol * 4000 symbols per second = 300 kbps.
ADSL2 Annex J: “All Digital Mode ADSL with improved Spectral Compatibility with ADSL over ISDN” is identical to an ADSL over ISDN system but with a downward extended upstream frequency bandwidth using maximum 63 tones in upstream (tone 1-63 in stead of 28-63 for ADSL over ISDN upstream) described in a family of allowed upstream PSD masks.
The All Digital Mode could be interesting for more symmetrical services e.g. companies which have ADSL but also separate lines for voice communication (e.g. PRA), so they don’t use the POTS or ISDN band on the twister pairs used for ADSL. Why not use all the frequencies in that case?
ADSL2 provides the ability to split the bandwidth into different channels with different link characteristics for different applications. For example, ADSL2 enables simultaneous support of a voice application, which might have low latency but a higher error rate requirement, and a data application, which might have high latency but lower error rate requirement.
In ADSL2 up to 4 frame bearers and 4 latency paths are supported.
In ADSL you had to choose: either fast or interleaved (with different interleaving depths)
In ADSL2 four different latency paths can be multiplexed
ADSL2’s channelization capability also provides support for Channelized Voice over DSL (CVoDSL), a method to transport derived lines of TDM voice traffic transparently over DSL bandwidths.
There are no carriers reserved for CVoDSL. A certain carrier may contain CVoDSL traffic but also data (multiplexing).
CVoDSL is unique among voice over DSL solutions in that it transports voice within the physical layer. In ADSL systems, this allows the transport of derived voice channels while maintaining both POTS and standard-compliant full-rate DSL data access. CVoDSL uses physical layer "channels" of DSL bandwidth over the local loop to deliver PCM DS0s – 64 Kbps from the DSL CPE to the next-generation access equipment. The access equipment then transmits the voice DS0s – 64 Kbps directly to the circuit switch via PCM. This approach eliminates the need for packetizing the voice traffic over the copper loop into upper layer protocols such as ATM and IP. The access equipment could alternatively packetize the DS0s into ATM or IP for transport to a media gateway or packet switch. The result is a simple, flexible, cost-effective method to enable next-generation equipment with derived voice functionality.
If no voice traffic is to be sent using CVoDSL, the carriers involved can be used for data traffic (dynamic rate repartition).
The elimination of cell overhead leaves more bandwidth for payload (data, voice…). This is particularly important for standard-compliant ADSL which is by far the most common DSL technology used for residential applications -- where bandwidth allocated for upstream data flow represents about 10% of the total bandwidth. Upstream bandwidth for ADSL systems varies between service providers, but typically falls between 90 kbps and 768 kbps. Since upstream bandwidth is a constraint to symmetrical services such as voice, every effort should be made to preserve as much upstream bandwidth as possible.
This slide illustrates the bandwidth efficiency of various voice-over-DSL techniques. CVoDSL has a clear advantage. Note that the efficiency of VoP is largely dependent on payload sizes. Bandwidth efficiency will increase with payload size, but so will latency.
<number>
Different frame bearers can be used for different information streams: for voice, data, video etc.
1 to 4 latency paths that can accept input from 0, 1 or more of the frame bearers
but each frame bearer shall be transported on one and only one latency path.
A common requirement among carriers is the ability to provide different service level agreements (SLA’s) to different customers. Data rates to homes and businesses can be significantly increased by bonding multiple phone lines together. To enable bonding, the ADSL2 standards support the ATM Forum’s inverse multiplexing for ATM (IMA) standard developed for traditional ATM architectures. Through IMA, ADSL2 chipsets can bind two or more copper pairs in an ADSL link. The result is a far greater flexibility with higher data rates (see further).
The IMA standard specifies a new sublayer that resides between the ADSL physical layer (PHY) and the ATM layer. At the transmitter side, this sublayer, called the IMA sublayer, takes in a single ATM stream from the ATM layer and distributes this stream to multiple ADSL physical interfaces. At the receiver side, the IMA sublayer takes in ATM cells from multiple ADSL physical interfaces and reconstructs the original ATM stream.
The IMA sublayer specifies IMA framing, protocols and management functions that are used to perform these operations when the physical interfaces are lossy (bit errors), asynchronous, and have different delays. In order to work under these conditions, the IMA standard also requires modifications to some of the standard ADSL physical interface functions such as the discarding of idle cells and corrupted cells at the receiver. ADSL2 includes only an IMA operation mode to provide the necessary physical layer modifications for IMA to work in combination with ADSL. Remark: the IMA sublayer is not part of the ADSL2 standard!
With IMA. several smaller lines are treated as one single logical connection. The fact that a number of smaller lines are used should be transparent to the application and the rest of the network, cell order and format is retained. The slide illustrates the operation of an IMA system. Cells arriving from the sending ATM layer originate at the left hand side. The IMA-transmitter multiplexes these cells in a Round-Robin manner onto the three physical links in the IMA-group, interleaving IMA Control Protocol (ICP) cells for control and synchronization purposes. IMA cells (ICP or filler) are specially coded OAM cells that IMA does not convey to the ATM layer. At the receiver side, the cells on each of the physical links may arrive after different delays on each of the physical connections, for example due to different lengths. On the right hand side, the IMA receiver employs the ICP cells to realign the user cells before multiplexing them back into accurately reproduced version of the original ATM cell stream, which it then delivers to the destination ATM layer.
Also a typical use of IMA, is for the connection between a Hub-ASAM and a Subtending-ASAM. See ADSL course for further details.
In this graphic the result of two bonded ADSL2 lines is a doubling of the downstream data rate.
In ADSL2+ the downstream spectrum is extended to 2.2 MHz (512 carriers).
Either all carriers are used. At short distances this will result in higher bit rates.
Or only the highest frequencies are used. ADSL2+ can then be used to improve spectral compatibility between CO and remote cabinet.
While the current ADSL standard (G.992.1) specifies a downstream frequency band up to 1.1 Mhz, ADSL2+ specifies a downstream frequency up to 2.2 Mhz. The result is a significant increase in data rates on shorter phone lines.
ADSL2+ from the CO: frequencies used from 0.138 MHz until 2.2 MHz.
Remote ADSL2+ from the remote terminal: frequencies used from 0,5 until 2.2 MHz (in this example).
A number of applications (such as some video streams or combinations of video and data streams) can benefit from higher downstream rates than are currently possible with ADSL2 G.dmt.bis (effectively limited to <15 Mbps). The VDSL standard supports these higher rates, but by design only on loops shorter than approximately 1.5 km on 0.4 mm/26 AWG twisted pair due to the standardized VDSL frequency plans, upstream transmit power and synchronization limitations. By increasing the ADSL downstream bandwidth, higher bit rates can be provided on loops up to 2.5 km or even 3 km 0.4 mm/26AWG.
However, ADSL2+ does not bring a performance increase on longer lines, because of the skin effect, which gives an additional attenuation for the higher frequency bands.
ADSL2+ is a simple extension to the ADSL standard that should add less than 5% to the chip set cost.
ADSL2+ can also be used to improve spectral compatibility between a central office and a remote cabinet, by using only tones between 1.1 and 2.2 MHz from that remote cabinet and masking the downstream frequencies below 1.1 MHz. See further.
By increasing in ADSL2+ the downstream bandwidth, higher downstream bitrates can be provided on loops up to 2.5 km on 0,4mm/26 AWG. There is no significant increase of the downstream bitrate, on lines longer than that.
Evidently, nothing changes to the upstream bandwidth.
For ADSL2+ here the assumption is made that it is not a remote deployment and therefore the full spectrum is used.
ADSL2+ can also be used to improve spectral compatibility in a mixed central office - remote cabinet deployment. In this example ADSL2+ provides the capability to use only tones between 1.1 MHz and 2.2 MHz by masking the downstream frequencies below 1.1 MHz. This can be particularly useful when ADSL services from both the central office (CO) and a remote terminal (RT) are present in the same binder as they approach customers’ homes. The crosstalk from the ADSL services from the remote terminal onto the lines from the CO can significantly impair data rates on the line from the CO. ADSL2+ can correct this problem by using frequencies below 1.1MHz from the central office to the remote unit, and frequencies between 1.1 MHz and 2.2 MHz from the remote terminal to the customer premise. This will eliminate most of the crosstalk between the services and preserve data rates on the line from the central office.This downstream spectrum is under operator control through the CO-MIB, via a PSD (power spectral density) mask which can be programmed in terms of breakpoints (within certain constraints).
In general only tones between a certain fx and 2.2 MHz can be used. In the example on the slide fx = 1.1 MHz, but fx can go from tone 100 to tone 280 in steps of 10 tones. This means fx could have been e.g. 500 kHz.
A second use of the ‘MIB Programmable Downstream PSD Mask’, is the definition of RFI bands (e.g. HAM band at 1.8-2 MHz). In order to avoid egress interference to the sensitive amateur radio receivers, the PSD in this band should be less then -80 dBm/Hz. The notching for RFI bands is controlled through MIB Programmable PSD Mask.
Operators want to extend their coverage. Extending the reach with 600 m can be worthwhile: maybe a lot of new potential subscribers can be reached.
How can the reach be extended?By using the same total power in a smaller frequency band (PSD masks).
We can’t expect miracles here: Shannon’s limit will not allow it.
One of the challenges the DSL service providers are facing today with CO-based ADSL deployment is the reach limitation of ADSL. Today ADSL is typically offered on line lengths up to approximately 4.5km 26AWG. Therefore, a certain number of remote customers cannot be served by ADSL from the CO.
In October 2002, a Long Reach DSL project was started in ITU-T with the goal to extend the reach of DSL. This should result in a larger service coverage and thus increased operator revenues with a minimum of extra investment. At the January 2003 ITU meeting the READSL2 text was "determined", i.e. technically frozen and agreed to be a separate annex L of G.992.3.
The upstream PSD masks are characterized by the use of a higher PSD level in a smaller frequency band but maintaining the same total power as ADSL(2). In upstream the 2 mandatory PSD masks with increased PSD level are very similar : same total power, only different PSD level and different number of used tones. The 2 downstream masks however have a fundamental difference. The Alcatel preferred mask is the mandatory non-overlapped downstream mask, which is similar to the existing ADSL downstream mask but with applying a boosted PSD over half the bandwidth. This mask gives the best performance in most noise conditions. The other optional mask is a 2 band downstream PSD that, compared with the first proposal, contains an additional low frequency downstream PSD in overlap with the ADSL upstream frequencies.
Operator control is provided to enable/disable certain downstream or upstream READSL2 masks via the MIB (Management Information Base). Within the enabled masks, the ATU-C selects the downstream and upstream mask. This selection is communicated towards ATU-R via G.994.1 code points. The maximum power and PSD level can be controlled by the operator through the MIB parameters.
G.992.3 Annex L contains one mandatory downstream power spectral density (PSD) mask, one optional downstream PSD mask (not on the slide), and two mandatory upstream PSD masks number 1 (on the slide) and number 2.
Increased loop reach is achieved by transmitting at higher PSD levels while using less frequency bandwidth than ADSL2.
ADSL2READSL2
peak in-band US PSD mask-34.5 dBm/Hz-32.9 dBm/Hz (mask #1)
peak in-band DS PSD mask-36.5 dBm/Hz-33.5 dBm/Hz
US bandwidth112 kHz78 kHz (mask #1)
max frequency of DS PSD mask1.1 MHz552 kHz
This trade-off between bandwidth and PSD levels is an important aspect of the READSL2 system in that it allows reach-extended systems to be deployed in the same environment as legacy ADSL systems without causing them serious harm.
READSL2 systems provide increased performance on long lines under various crosstalk conditions. The increase in transmit PSD levels result in increased data rates, which are achieved even though the overall transmit bandwidth is less than ADSL systems. This is primarily because the higher frequency spectrum is not usable on long lines due to the large channel attenuation.
Selecting the PSD masks
The READSL2 PSD masks were specifically designed to provide improvement loop reach under a large variance of deployment guidelines, crosstalk conditions, and loop topologies. The resulting data shows that different combinations of the upstream and downstream READSL2 PSD masks will provide the best loop reach performance. It is expected that selection of the best combination of PSDs by the operator will be possible, based on knowledge of the loop conditions in the network. Additionally, the standard will provide a mode of operation where the ADSL transceivers are allowed to autonomously select the best combination of PSD masks during initialization.
Another important aspect of the READSL2 system is that it is designed so that the RE mode of operation can be easily implemented on an ADSL2 modem (CO or CPE). This is accomplished by utilizing RE PSD masks that are achievable by simply turning off carriers and/or shaping the PSD on each carrier of the multi-carrier modulation system.
This graphic shows that READSL2 is really optimized for long loops.
Noise conditions for the diagram in above slide are 12 self crosstalkers with –140dBm/Hz white noise at CO side and 6 self crosstalkers with also –140dBm/Hz white noise at CPE side.
On short lines ADSL2 has a better upstream performance than READSL.
As you can see, the reach extension depends on the service that has to be offered.