Description

Method and device for data processing in a digital

subscriber line environment

The invention relates to a method and to a device for data processing in a digital subscriber line environment. The invention also suggests a communication system comprising at least one such device.

DSL or xDSL is a family of technologies that provide digital data transmission over the wires of a local telephone network. High speed Internet access is gaining importance and is often realized via xDSL services using existing copper lines. Also, other applications emerge that require broadband transmission services, e.g., triple play offers comprising subscriber access to Internet, TV and voice data transmission. A bandwidth consuming application is the transmission of TV data via xDSL, wherein one HDTV channel requires a data rate amounting to 12 Mbit/s.

Therefore, requirements for high speed Internet access are increasing. Operators are optimizing services that are offered to their customers . This becomes a difficult task as an increasing amount of users as well as high data rates leads to higher crosstalk between subscriber lines in a cable binder. In most cases, crosstalk noise limits the performance. However, the actual problem is that crosstalk noise varies over time: There may be low crosstalk noise when a significant amount of customers switched off their equipment and there may be a considerable amount of crosstalk noise during business hours when the majority of customers use their devices. The fluctuation of crosstalk is getting more dynamic in case low power modes are used (see L2 mode in ADSL2 G.992.3 and ADSL2plus G.992.5) . In this case, the applied transmit power varies with the actual data throughput.

It is in particular problematic if a modem has been initialized during a low crosstalk noise period and the crosstalk noise increases during operation of the modem, e.g., by additional modems (customer premises equipments, CPEs) being activated. As a consequence, the modem that has been initialized during a time of low crosstalk experiences transmission errors and connection outages during periods of high crosstalk. Hence, the modem will have to conduct a restart and synchronization, which leads to a significant service interruption (e.g., up to 75 seconds), which is inacceptable for the user especially when watching

television via the broadband access.

Fig.l shows a schematic diagram comprising a power of a noise 101, a margin 102 and user data 103 over time. An impulse noise during a time period 104 affects the user data thereby leading to CRC errors, which may be visible to a user's IPTV application. Additional noise from the DSL during a time period 105 results in a retraining of the modem adjusting its user data/margin differently starting at a time 106. Unfortunately, the retraining leads to an outage of the user data 103 and thus of the IPTV service, which will continue at said time 106 at a reduced data rate .

Fig.2 shows a schematic diagram visualizing several portions of a reception power in a logarithmic scale over a frequency for illustration purposes comprising a flat noise margin 203, a floor 201 of a receiver and a crosstalk noise 202 on top of this noise floor 201. The flat noise margin 203 is applied on top of both, the noise floor 201 and the crosstalk noise 202. An area beyond the noise margin 203 corresponds to the portion of a received signal power to be available for data transmission 204, i.e. is proportional to an attainable data rate. In case the crosstalk noise 202 increases beyond the noise margin, a retraining of the modem will become necessary and the noise margin will be adjusted at the cost of a reduced attainable data rate.

Hence, Fig.2 shows an operational case of signals received at an xDSL modem in case a connection between a DSLAM and a CPE has been established. xDSL with DMT modulation can be used according to, e.g., ADSL G.992.1, ADSL2 G992.3,

ADSL2plus G992.5, VDSL2 G.993.2. Even if the transmit power is constant over a certain frequency range, the received signal power 204 (available for data transmission) declines at higher frequencies due to an attenuation of the channel. The amount of data that can be transported at a certain subcarrier frequency is roughly proportional to a signal- to-noise ratio (SNR) . In case only thermal noise and implementation effects contribute to the noise, the achievable data rate may be substantially proportional to the area between the receiver noise floor 201 and the received signal power at the receiver. Crosstalk noise 202 from other lines and services limit the achievable data throughput. A moderate increase of the crosstalk noise 202 can be compensated as the signal-to-noise ratio is not fully exploited for data transmission. Instead, the noise margin 203 is provided for safety reasons . The SNR margin may represent an acceptable increase of noise received (in dB) such that the system still meets a target bit error rate (BER) amounting to 10 ^~7 .

A large target noise margin can be utilized by the modem during initialization. The high noise margin protects the system against an increasing noise level, but allows only a reduced suboptimal data rate. Such high noise margin stabilizes the system in case the noise increases . If however the increased noise due to additional operating DSL services is dependent on the frequency, the maximum noise level will be different for different frequencies .

Fig.3 shows a schematic diagram visualizing a huge noise margin 303. A power (e.g., in a logarithmic scale) over a frequency is used to illustrate a noise floor 301 of a receiver and a crosstalk noise 302 on top of this noise floor 301. The flat noise margin 303 is applied on top of both, the noise floor 201 and the crosstalk noise 202 according to Fig.2 and shows how the high crosstalk noise 302 can be tolerated by a large flat margin 303. An area beyond the noise margin 303 corresponds to a received signal power to be available for data transmission 304, i.e. is proportional to an attainable data rate.

According to Fig.3, the data rate corresponding to the received signal power is significantly reduced in case the modem is initialized using this high noise margin 303 in a situation where the external noise from other lines is already on a high level. This high noise level leads to a reduced data rate and the large noise margin further reduces the data rate based on the fact that the target noise margin is applied on top of the measured external noise independent of its absolute level. Such kind of noise margin is however not required in case the DSL is

initialized during a situation in which a further increase of noise level is not expected.

A so-called Virtual Noise was introduced by xDSL

recommendations (ITU G.993.2, G.992.3, G.992.5) : An absolute Virtual Noise level is set and the maximum of external and Virtual Noise level is used by the modem during initialization instead of the measured external noise. An operator usually sets the Virtual Noise to a level of expected noise when all modems are active.

Therefore, the actual noise situation does not matter when the modem is being initialized. The modem during

initialization considers the maximum expected noise level, which is set by the Virtual Noise. Hence, an additional target noise margin can be set to a rather small amount since the maximum noise level is already taken into account . It is an advantage of Virtual Noise that a set of

breakpoints can be used to set frequency-dependent noise levels . This is an improvement over the flat noise margin and allows for higher data rates. This beneficial effect is also referred to as shaping gain.

However, not all xDSL variants and/or CPEs support Virtual Noise. For example, ADSL (G.992.1) does not provide any Virtual Noise, ADSL2 (G.992.3) and ADSL2plus (G.992.5) were augmented recently with Amendment 5, but corresponding CPEs are and will be built not considering this latest

recommendation .

An approach to solve this issue is by using an Artificial Noise (see, e.g., EP 1 641 173 Al ) . Artificial Noise works similar to Virtual Noise, but injects real noise into the loop, whereas Virtual Noise is a numerical correction term in the SNR computation. The level of the Artificial Noise as well as the level of the Virtual Noise are set to the expected maximum noise level.

Fig . 4 shows a degradation of the SNR based on Artificial Noise. A curve 402 shows a SNR at a receiver comprising external noise and Artificial Noise. A curve 403 depicts a SNR at a receiver based on the Virtual Noise approach explained above. A curve 401 shows a SNR at a receiver without external noise (with only Artificial Noise being present ) .

Basically, the following disadvantages exist: The external noise (crosstalk and RFI) and Artificial Noise sum up to an increased noise level, which leads to a decreased signal- to-noise ratio (SNR) .

The level of SNR degradation depends on the noise level of the line and the level of the Artificial Noise. The situation is worst in case the Artificial Noise and the external noise reach the same level. This is indicated in Fig.4 as a SNR degradation amounting to 3 dB, which corresponds to a significant loss with regard to an achievable data rate.

An ADSL2plus system may use up to 506 tones in downstream direction while an SNR degradation of 3 dB corresponds to a reduction of about 1 bit per tone. This may sum up to a downstream rate loss of up to 2 Mbit/s.

The problem to be solved is to overcome the disadvantages described above and in particular to prevent or reduce a SNR reduction in case Artificial Noise is used on digital subscriber lines .

This problem is solved according to the features of the independent claims . Further embodiments result from the depending claims .

In order to overcome this problem, a method for data processing in a digital subscriber line environment is provided,

— wherein a noise level of a line is determined, — wherein an Artificial Noise is adapted based on the noise level determined. Artificial Noise can be applied like Virtual Noise. Hence, Artificial Noise can be set to a predefined level, e.g., a maximum expected loop noise level, to avoid undesired effects .

This also means that there may be no need for applying Artificial Noise in case the external noise at the CPE has already reached or exceeded the Artificial Noise level. Since loop noise and Artificial Noise add up, it may be beneficial to ensure that the sum of both does not fall below the predefined level of the Artificial Noise.

It is suggested herein to determine the loop noise of a line and to adapt an Artificial Noise to that level. Hence, the Artificial Noise can be switched off in case the external noise is above the predefined level of Artificial Noise. In case the external noise is below the predefined level of Artificial Noise, it will be set in a way that the sum of both results in the desired level of noise.

Advantageously, this approach avoids or reduces SNR degradation and loss in data rate. It is noted that loop noise is in particular referred to as noise that could be determined (in particular measured) on the line, comprising in particular artificial noise.

External noise, however, could be regarded as noise without any artificial noise, e.g., thermal noise, crosstalk, RFI, etc.

In an embodiment, the Artificial Noise is adapted to the noise level of the line. In another embodiment, the Artificial Noise is switched off in case the noise level of the line reaches or exceeds a predefined threshold, in particular a predefined level of Artificial Noise .

It is noted that such switching off is preferably done for single carriers (also referred to as subcarriers) . It is also noted that the Artificial Noise is received with a damped due to attenuating effects of the line or along the line. Hence, the predefined threshold may be used to compare the damped Artificial Noise obtained at the receiver.

In a further embodiment, in case the noise level does not reach or does not exceed the predefined threshold, the Artificial Noise will be set such that the sum of the noise level and the Artificial Noise reaches or equals the predefined threshold.

In a next embodiment, the noise level is determined during initialization, training or showtime state of a terminal .

The terminal can be regarded as a DSL terminal device, e.g., a CPE. The terminal can be connected to a CO or DSLAM. For example, ADSL2 G.992.3 or ADSL2plus G.992.5 each provides measurement of the loop noise during

initialization.

It is also an embodiment that the noise level is derived or estimated based on a signal-to-noise ratio. Hence, the SNR can be used to determine or derive the actual noise level. Updated measurements of the SNR(i) per subcarrier can be requested from ADSL2 G.992.3 or ADSL2plus G.992.5 during showtime, where (i) is the subcarrier index and is equivalent to a point in frequency f (i) . Pursuant to another embodiment, the noise level is based on an iterative mechanism utilizing the signal-to-noise ratio, in particular according to

AN _new (i) = VN(i) + AN _M (i) - ^P(l)

SNR(i) wherein

SNR(i) is a signal-to-noise ratio per

subcarrier i;

AN _r is a new value for the Artificial

Noise ;

AN is an old value of the Artificial

Noise ;

P(i) denotes a frequency dependent transmit power spectral density;

(i) corresponds to the subcarrier index and is equivalent to a point in frequency f (i) ;

VN (i) is the desired level of an equivalent

Virtual Noise .

This scheme has the advantage that it does not require measurements of the channel or the loop noise. Furthermore, due to its iterative nature, any bias error in the

application of Artificial Noise or measurement of the channel transfer function can be reduced or eliminated. The value for the new Artificial Noise AN _new (i) may become negative in case the loop noise of the line exceeds the desired Virtual Noise level. In such case, the

corresponding values of the new Artificial Noise AN _new (i) are set to zero.

According to an embodiment, the signal-to-noise ratio is determined according to

SNR(i) = (2 ^b{i) - 1)- SNRGAP ^■ SNRM wherein

b(i) is the number of bits that can be

transported via a subcarrier i;

SNRGAP is an additional factor that is used to achieve the desired level of BER for a given modulation scheme and

implementation ;

SNRM is the signal-to-noise-ratio margin; SNR(i) is a signal-to-noise ratio per

subcarrier i .

According to another embodiment, the Artificial Noise is used as a replacement for transmitter-referred Virtual Noise in downstream direction.

In yet another embodiment, the Artificial Noise to be injected into a line is determined by AN(i)=VN(i)- ^

H(i) wherein

H(i) is the magnitude of a channel transfer function ;

LN(i) is the loop noise;

(i) corresponds to the subcarrier index and is equivalent to a point in frequency f ( i ) ;

VN(i) is the desired level of an equivalent

Virtual Noise;

AN ( i ) is the Artificial Noise.

One way of retrieving the loop noise LN(i) is to set the system into a loop diagnostic mode prior to an

initialization of the line. The test parameters Quiet line noise PSD per subcarrier (QLNps) provides the data required. The loop diagnostic mode also provides

information about the channel transfer function (Hlog and Hlin) . According to a next embodiment, an adaptation of the noise level of a line is conducted.

Pursuant to yet an embodiment, the Artificial Noise is initially set to a predefined value and adapted later, in particular during showtime of the digital subscriber line.

For example, during start-up, a rather low SNR noise margin is used; at a later stage during adaptation of the

Artificial Noise, the noise margin can (e.g., gradually) be set to its normal range. As an alternative, seamless rate adaptation (SRA) can be used after the start-up.

The problem stated above is also solved by a device for data processing in a digital subscriber line environment, comprising or being associated with a processing unit that is arranged

- for determining a noise level of a line,

- for adapting an Artificial Noise based on the noise level determined.

The device may be associated with a CPE or a DSLAM/CO.

It is noted that the steps of the method stated herein may be executable on this processing unit as well.

It is further noted that said processing unit can comprise at least one, in particular several means that are arranged to execute the steps of the method described herein. The means may be logically or physically separated; in

particular several logically separate means could be combined in at least one physical unit. Said processing unit may comprise at least one of the following: a processor, a microcontroller, a hard-wired circuit, an ASIC, an FPGA, a logic device.

The solution provided herein further comprises a computer program product directly loadable into a memory of a digital computer, comprising software code portions for performing the steps of the method as described herein.

In addition, the problem stated above is solved by a computer-readable medium, e.g., storage of any kind, having computer-executable instructions adapted to cause a computer system to perform the method as described herein.

Furthermore, the problem stated above is solved by a communication system comprising at least one device as described herein. Embodiments of the invention are shown and illustrated in the following figure :

Fig.5 shows a schematic flow chart comprising steps of how to use Artificial Noise in an efficient manner.

Artificial Noise can be applied like Virtual Noise. Hence, Artificial Noise can be set to a predefined level, e.g., a maximum expected loop noise level, to avoid the undesired effects as explained in the introductory portion, e.g., retraining, loss of connection, etc.

This also means that there is no need for applying

Artificial Noise in case the external noise at the CPE has already reached or exceeded the Artificial Noise level. Since external noise and Artificial Noise add up, it may be beneficial to ensure that the sum of both does not fall below the predefined level of the Artificial Noise.

It is suggested herein to determine the external noise of a line and to adapt an Artificial Noise to that level. Hence, the Artificial Noise can be switched off in case the external noise is above the predefined level of Artificial Noise. In case the external noise is below the predefined level of Artificial Noise, it will be set in a way that the sum of both results in the desired level of noise.

Advantageously, this approach avoids or reduces SNR degradation and loss in data rate. Implementation and Further Advantages

In order to adapt the level of the Artificial Noise to the level of the external noise, the actual noise level can be determined at the time of or during initialization and showtime at the CPE. For example, ADSL2 G.992.3 or

ADSL2plus G.992.5 each provides measurement of the loop noise during initialization.

As an alternative, the SNR can be used to determine or derive the actual noise level. Updated measurements of the SNR(i) per subcarrier can be requested from ADSL2 G.992.3 or ADSL2plus G.992.5 during showtime, where (i) is the subcarrier index and is equivalent to a point in frequency f (i) ·

If no such measurement is available, the SNR(i) can be estimated using the calculations shown below.

Although the SNR and other related values are often expressed in a logarithmic scale, the following

calculations are based on a linear scale for easier transformation purposes. It is assumed that an Artificial Noise AN(i) is used as a replacement for a transmitter referred Virtual Noise VN(i) in downstream direction. The desired noise level at the CPE is therefore given by: s _{m ΓΛ} P(i) - H(i) P(i)

SNR _{T arei} ,, (i) = = , ( 1 \

^{T <sget} AN(i) - H(i) VN{i) ^{( i )} wherein

P(i) denotes a frequency dependent transmit power spectral density (PSD) and may include fine gains;

H(i) is the magnitude of a channel transfer

function ;

LN(i) is the loop noise;

(i) corresponds to the subcarrier index and is equivalent to a point in frequency f(i);

VN(i) is the desired level of an equivalent

Virtual Noise;

AN ( i ) is the Artificial Noise.

Information about the transmit PSD can be obtained from management information base (MIB) . Since the loop noise LN(i) and attenuated Artificial Noise sums up the actual SNR is given by:

This can be rewritten as :

1 _ LN(i) AN(i)

-+ · (3)

SNR(i) P(i) ^■ H(i) P(i)

The desired noise level amounts to:

SNR(i)=SNR _Taget (i) ₍₄₎ Using the equations (1) to (3), the Artificial Noise that needs to be injected in the line to reach the target SNR can be determined as follows :

AN(i) = VN(i) - ^^- _{( 5 )}

HO) ^{13 ;}

As indicated above, information about the actual loop noise LN(i) and the transfer function H of the loop are required to solve this equation (5) .

One way of retrieving the loop noise LN(i) is to set the system into a loop diagnostic mode prior to an

initialization of the line. The test parameters Quiet line noise PSD per subcarrier (QLNps) provides the data

required. The loop diagnostic mode also provides

information about the channel transfer function (Hlog and Hlin) .

Later (e.g., during showtime, i.e. when the loop is active) the loop noise may vary. This may requires an adaptation of the loop noise during showtime. If measurements of the loop noise LN(i) cannot be done during showtime, an indirect approach using measurements of the SNR(i) per subcarrier is conducted. For example, an iterative scheme employing only measured SNR(i) values can be used. It is assumed that SNR(i) was measured during a time when an Artificial Noise AN _old (i) was applied. The new value for the Artificial Noise AN _new (i) can thus be determined as follows:

This scheme has the advantage that it does not require measurements of the channel or the loop noise. Furthermore, due to its iterative nature, any bias error in the application of Artificial Noise or measurement of the channel transfer function can be reduced or eliminated. The value for the new Artificial Noise AN _new (i) may become negative in case the loop noise of the line exceeds the desired Virtual Noise level. In such case, the

corresponding values of the new Artificial Noise AN _ne „(i) are set to zero.

Updated SNR(i) values per subcarrier can be retrieved during showtime from ADSL2 G.992.3 or ADSL2plus G.992.5. It is not necessary to request an update of the test parameter frequently. For example, bitswap activities and/or changes in SNR margin (SNRM) may trigger an update of the test parameter. The SNR margin is the maximum increase (scalar gain, in dB) of the reference noise PSD (at all relevant frequencies), such that the BER of each bearer channel does not exceed 10 ^~7 .

If it is not possible to use loop diagnostic before initialization of a line, a starting condition may set the Artificial Noise level to the Virtual Noise level (i.e., AN(i) = VN(i)) and provide adaptation at a later stage. In such exemplary scenario, the Artificial Noise is higher during initialization which results in a lower data rate . This can be compensated or avoided according to the following approaches :

(1) During start-up, a rather low SNR noise margin is

used; at a later stage during adaptation of the

Artificial Noise, the noise margin can (e.g.,

gradually) be set to its normal range.

(2) Seamless rate adaptation (SRA) can be used after the start-up .

The approach presented can also be used in an ADSL

(G.992.1) environment or in case loop diagnostic mode and measurements of the SNR are not supported. A small noise margin can be used during initialization and an adaptation can be conducted later using the following indirect approach to estimate the SNR: A number of bits b(i) that can be transported on a subcarrier can be estimated, e.g.: where

Γ = SNRGAP ^■ SNRM (8) with

b(i) is the number of bits that can be

transported via a subcarrier i;

Id corresponds to "log2" ( " logarithmus

dualis " ) ;

SNRGAP is an additional factor (e.g. 4.79) that is used to achieve the desired level of BER for a given modulation scheme and implementation ;

SNRM is the signal-to-noise-ratio margin;

S(i) is a received signal power on the subcarrier i (content information);

N(i) is a received noise power on the subcarrier i (unwanted signal) .

It is noted that S(i)/N(i) may be replaced by the signal to-noise ratio SNR.

Equations (7) and (8) can be re-formulated to obtain an estimate of the SNR:

SNR(i) = (2 ^b(i) - 1)- SNRGAP ^■ SNRM (9) By using this approach, the solution presented herein is also applicable, e.g., for ADSL (G.992.1) .

It is advantageous to adapt the Artificial Noise in the described way in order not to be affected by SNR

reductions . Up to 3 dB SNR reduction will result in a decreased data rate and possibly in a higher transmission error rate. Furthermore, using this approach, Artificial Noise will be a true equivalent to Virtual Noise due to independence of the actual noise situation on the line.

Fig.5 shows a schematic flow chart comprising steps of how to use Artificial Noise in an efficient manner. In a step

501 an actual external noise level (which could also be perceived as loop noise level) is determined and in a step

502 the Artificial Noise is set or adapted based on such external noise level determined.

The step 501 may consider at least one of steps 503 to 505: In the step 503, the noise level of the line is determined during initialization, training or showtime of a terminal. In the step 504, external noise level is determined or estimated based on a SNR. Pursuant to the step 505, the estimated external noise level can be adapted, e.g., during showtime.

The step 502 may comprise a step 506 and/or a step 507. In the step 506, Artificial Noise may be switched off in case the external noise level of the line reaches or exceeds a predefined threshold (e.g., a predefined level of

Artificial Noise) . According to the step 507, in case the external noise level does not reach or does not exceed the predefined threshold, the Artificial Noise can be set such that the sum of the external noise level and the Artificial Noise reaches or equals the predefined threshold. List of Abbreviations :

ADSL Asymmetric Digital Subscriber Line

BER Bit Error Rate

CO Central Office

CPE Customer Premises Equipment

CRC Cyclic Redundancy Check

DELT Dual Ended Line Test

DMT Discrete Multi Tone

DSL Digital Subscriber Line

DSLAM Digital Subscriber Line Access Multiplexer

FEXT Far End Crosstalk

HDTV High Definition Television

IPTV Internet Protocol Television

MIB Management Information Base

PSD Power Spectral Density

QLN Quiet Line Noise

RFI Radio Frequency Interference

SNR Signal-to-Noise Ratio

SRA Seamless Rate Adaptation

VDSL Very High Speed Digital Subscriber Line xDSL Any of the various types of Digital Subscriber

Lines (DSL)