Transmission of data in communication systems such as DSL systems, Ethernet systems or other data communication systems is typically influenced by noise. Noise influencing the data transmission can be classified into different noise types. Near-end noise is generated at the near-end of a transmitter. Examples of near-end noise include echo noise and NEXT (Near-end crosstalk) noise. Echo noise originates in a transceiver when a part of the signal transmitted via a transmitter over a link couples into a receive path of the same transceiver thereby disturbing the receiving of data via the receiver of that link. NEXT noise occurs when a plurality of transceivers are arranged at one side of the transmission system and signals transmitted by one of the transceivers couple into the receive paths of another transceiver.
Contrary to the near-end noise, far-end noise is noise which is introduced at the far-end side of a transmitter. FEXT (far-end crosstalk) occurs typically when a plurality of links of the transmission system such as a plurality of wires, cables or lines are assimilated in a same bundle. During the transmission, the signals transmitted on the one link partially couples into other links. Thereby, noise is introduced at the receivers of the far-end side originating from the signals transmitted on the other links.
While for echo, NEXT and FEXT noise the noise source is the transmission of signals in the system itself, in another noise type referred to as alien noise the noise is introduced into the transmission system from outside of the transmission system.
While alien noise is hard to address, echo, NEXT and FEXT noise can be compensated by using adaptive filters. In order to compensate the noise, a replica of the respective transmit signals are provided from a respective transmit path to an adaptive filter. By properly setting the filter coefficients of the adaptive filter, a replica of the noise is generated at the output of the adaptive filter. Noise-compensated receive signals are then generated by subtracting the noise replica from the received signals.
a and 6b show further diagrams according to an embodiment; and
The following detailed description explains exemplary embodiments of the present invention. The description is not to be taken in a limiting sense, but is made only for the purpose of illustrating the general principles of embodiments of the invention while the scope of protection is only determined by the appended claims.
In the exemplary embodiments shown in the drawings and described below, any direct connection or coupling between functional blocks, devices, components or other physical or functional units shown in the drawings or described herein can also be implemented by an indirect connection or coupling. Functional blocks may be implemented in hardware, firmware, software, or a combination thereof.
Further, it is to be understood that the features of the various exemplary embodiments described herein may be combined with each other, unless specifically noted otherwise.
In the various figures, identical or similar entities, modules, devices etc. may have assigned the same reference number.
Referring now to
Further referring to
In order to compensate the near-end noise, i.e. echo or NEXT noise, an adaptive filter 202 is provided as shown in
To determine and set the filter coefficient of the adaptive filter 204, a training of the filter coefficients of the adaptive filter 202 prior to activating the link is performed. The training of the adaptive filter prior to the learning of the parameters of the link may also be referred herein as prelearning or simply as filter learning. In the prelearning the adaptive filter compensating the near-end noise is therefore trained prior to determining or training the link itself, i.e. prior to determining the parameters for communicating data over this link such as a SNR determination, equalizer training etc. In the prelearning, the filter is therefore trained right from the start, i.e. from a previously untrained state where the starting of a session for that link has just been indicated and no previously training of the filter has been performed for this session. The previously untrained adaptive filter is then trained such that no significant residual near-end noise remains after the compensation node 206 after the training (learning) of the near-end noise compensation filters. Subsequent to this prelearning of the compensation filters the learning of the link is performed in a virtually near-end noise free environment.
a shows an operation sequence 300 according to an embodiment for illustrating the above. The operation sequence starts with a silent phase 302 in which no signals are transmitted on the link. The silent phase may for example be obtained when a transceiver is not connected to the link or when a transceiver or a modem containing the transceiver is not powered. In embodiments, a link may also be forced to enter a silent phase when a restart is performed. After the silent phase 302, a handshake phase 304 occurs. In the handshake phase 304 both transceivers send handshake signals. The handshake signals in the handshake phase 304 may include handshake signals according to the ITU G.hn standard but are not limited to this specific type of handshake signals. After the handshake phase 304, the above described prelearning phase 306 is entered in which the NEXT and/or echo compensation is trained. It is to be noted that in one embodiment the prelearning may be performed during the handshake phase 304, i.e. the handshake signals of handshake phase 304 are transmitted simultaneously with the training of the NEXT and/or echo compensation. After successfully completing the prelearning phase 306, the learning of the link is performed in phase 308. Having the learning of the link completed, the systems are ready to enter the data mode phase 310 in which the transmission of (user) data between the two modems starts. The data transmission phase 310 is sometimes also referred to as showtime. For updating the coefficients of the adaptive filter during training in the prelearning phase, several techniques may be used. In one embodiment, the training may include a LMS (least mean square) algorithm which will be described later in more detail.
In the following, several embodiments will be described which provide a successful training of the adaptive filter in the prelearning phase prior to the link activation when the transceiver (link partner) or a plurality of transceivers at the other side (remote side) of the transmission system is not silent, i.e. signals are transmitted to the side of the trans-mission system performing the training. As will be described below in more detail, the signals transmitted from the other side during the filter training may in one embodiment include handshake signals. In one embodiment, the signal transmitted during the training may be signals limited to a predetermined frequency band. In other embodiments, the signals may be BPSK (Binary Phase Shift Keying) modulated signals. However, this list of signal types should not be understood as a limitation of the invention. Since the embodiments described herein do not setup any communication channel to the remote side it is possible that the link partner may or may not send any of the mentioned signals at any time with any duration. In other words, the embodiments described herein allow to use the concept of training echo and/or NEXT during a transmission of signals from the remote side but the transceivers are also capable without any change or reconfiguration to train echo and/or NEXT compensation when the link is silent during the prelearning phase, i.e. when the transceiver at the remote side is programmed or configured to be silent during the prelearning phase. The embodiments described herein therefore provide a great flexibility in that no reconfiguration, exchange of components or switching is necessary to provide training for different type of modems connected to the link, i.e. to train transceivers which transmit signals such as handshake signals during the prelearning and transceivers which do not transmit signals during the prelearning.
Generally any type of band-limited signals transmitted during the prelearning by the remote side is suitable for proper operation. This includes most of the signals which are used for startup indications and configuration exchange in modern communication systems. An example is the ITU-G.handshake (G.hs) signal for DSL systems.
The training of the adaptive filter in the presence of signals transmitted from the remote side to the side performing the filter training is achieved according to embodiments by utilizing a feedback path for updating the adaptive filter and eliminating a part of the feedback signal before the update signal is utilized in the adaptive filter. In other words, a part of an error signal determined during the training is eliminated before updating the adaptive filter based on the determined error signal. According to one embodiment shown in
The training signals provide a near-end noise for the receive path 106. During the training, the task is to determine the amount of this noise and to successively update the filter coefficients in order to approach filter coefficients which provide at least acceptable near-end noise compensation. In the receive path, the signals downstream of the node 206 represent the compensated signals which have been subtracted by the output of the adaptive filter. The signal downstream of node 206 for example the signal at node 214 would represent the momentary error of the cancellation (compensation) provided by the adaptive filter when no signals are received at the receive path 106 from the other side (remote side). However, in the presence of signals from the remote side received by the receive path 106, this is no longer true. In other words, the signals received from the remote side provide a noise source for the error signal to be determined during the filter learning. This noise source can in practice dominate the receive signal power such that conventional filter training implementations do not converge to satisfactory filter coefficients.
The update filter 212 however removes this noise source prior to utilizing the feedback signal for updating the filter during filter learning. Therefore, even during the receiving of the signals from the remote side, learning of the filter coefficients is possible. The filter 212 may in one embodiment be a simple notch filter. However, any other filter tailored to the type of remote signal may suit as well. For very low frequency signals a high-pass filter may be used in one embodiment.
It is to be noted that by utilizing the update filter for filtering the error signal, the error signal fed back to the adaptive filter does no longer include the disturbing signals received from the link partner. However, not only the received signal from the remote side is eliminated by the update filter at all notch frequencies of the update filter 212 but also all components of the error signal which is required for training the coefficients are filtered out at the notch frequencies of the filter. In other words, the adjustment of the adaptive filter parameters is not influenced by the near-end noise at the frequencies eliminated by the update filter 212.
Therefore, the filter coefficients determined by utilizing the update filter may differ from filter parameters when using existing filter training with a silent link partner since the learning of the adaptive filter 202 will be provided without any information of the near-end noise at the notch frequencies. However, when using certain signals, the elimination of these signals at their respective transmit frequencies by the update filter provides only a small or negligible effect on the learned filter coefficients. Such signals include but are not limited to signals which have a small frequency bandwidth compared to the overall frequency bandwidth of the transmission system and/or signals which are located at low frequencies. Since the near-end noise follows a high-pass characteristic, the influence to the near-end noise gets stronger at higher frequencies. Or in other words the near-end noise is small or negligible at such low frequencies. Therefore, according to embodiments, signals transmitted during the training which have one or both of the above described criteria causes an influence of the filtering by the update filter which is small or negligible.
For example, handshake signals according to G.handshake (G.hs ITU G.994.1) are transmitted within a narrow frequency band at only low carrier frequencies. By setting the update filter 212 such that the G.handshake signals are eliminated, the influence of the elimination for the error update is rather small. For example, for SHDSL systems the carrier frequencies of the G.handshake are located at 12 kHz and 20 kHz and BPSK modulation is used for transmitting the signals. Broadband systems such as DSL-systems typically employ a bandwidth of 500 kHz and more. Therefore, the bandwidth of the G.handshake signals is less than 8% of the total bandwidth. Thus, according to one embodiment, the signals which are transmitted from the link partner at the remote side to the link partner training for near-end noise compensation are G.handshake signals. Since handshake signals are required to be transmitted by some technical standards, the above described embodiment allows providing a near-end noise training in compliance with these technical standards. It is to be understood that the above are only examples of signals which can be transmitted by the link partner during the near-end noise training.
In general, during the prelearning at the near-end side, the other side of the transmission system, i.e. the link partner, has no knowledge of the prelearning. Therefore, the link partner will generally continue transmitting signals to the other side. As outlined above, the embodiments described can address such situations in that it provides a concept for prelearning in the presence of a continuously transmitting link partner.
In embodiments, the update filter 212 is placed in the update feedback path of node 214. This avoids a notch in the receive signal which would impact data-transmission negatively. Placing the update filter upstream of node 206 would change the near-end noise transfer function and would not allow a proper operation since the filter has to be bypassed after prelearning. Bypassing the filter at the position upstream of node 206 after learning would then cause a phase change for signals provided to node 206. Since the phase of the signals provided during the prelearning to node 206 is different than the phase after the prelearning, the learning would be false and a proper operation after the prelearning would not be possible. Having the update filter 212 placed in the update feedback path of node 214 allows to remove (bypass) this filter when switching from prelearning into modem training or datamode which is not the case in other arrangements for the adaptive filter.
The update filter 212 is in one embodiment a band stop filter. In one embodiment wherein handshake signals are transmitted by the remote side, a notch of the band stop filter is set to the handshake carrier frequency of the respective link partner. In other embodiments, the update filter 212 may be a filter with a high-pass characteristic. Such filters may be useful for example when a high bandwidth has to be filtered out by update filter 212.
A diagram 500 illustrating an embodiment for learning of the adaptive filter shown in
At 504, the previously untrained adaptive filter is trained during the receiving of the first signals at the first side. In embodiments, for training the adaptive filters training signals are transmitted by the transmit path 114 and the training signals are at least partially transmitted concurrent with the receiving of the first signals by the receive path. Although 502 and 504 are show in separate blocks, it is to be understood that the operation described in 502 and 504 may be simultaneously and therefore may be provided also in a single block. Since the first signal provide a noise source distorting the training, the first signals may also be referenced in the following as distorting signals.
A flow diagram 600 explaining in more detail a procedure for learning of the adaptive filter in the prelearning phase according to a further embodiment is shown in
For updating the filter coefficient an update algorithm is provided which may be for example a LMS (least mean square) algorithm. In the embodiment shown in
In the embodiment having the update filter arranged in the feedback loop between node 206 and update input 210, the switching from an error-based update to a decision-based update can be achieved at any time, e.g. after successful modem training, without providing a distortion to the operation such as an interrupting of the filter operation or a changing of the CTC impulse response. An embodiment having the update filter upstream of the node 206 provides distortion to the operation and requires some additional measures to address these distortions.
LMS techniques which may be used in embodiments for updating the filter coefficients will now be described in the following. A LMS algorithm according one embodiment may use the following algorithm:
c(n+1)=c(n)+μ·e(n)·x(n).
In the above algorithm, c(n) may represent the coefficient presently used, c(n+1) may represent the calculated new coefficient, e(n) may represent the slicer error, x(n) may represent the filter input and μ may represent a weighting factor. In one embodiment, e(n) may represent a sign of the slicer error rather than the value of the slicer error. In one embodiment, x(n) may represent the sign of the filter input. In a further embodiment e(n) may represent the sign of the slicer error and x(n) may represent the sign of the filter input. In one embodiment, the weighting factor μ may be a variable which may be adjusted during the filter training.
The described concept of training near-end noise may be provided when a single link of a plurality of link is to be activated from a previous deactivated state. Other embodiments include a situation when a plurality of links is going to be activated such as for example when the whole transmission system is starting from a previous idle state.
The above described concept can be implemented such that multiple adaptive filters are trained in parallel.
a shows an embodiment of a prelearning for multiple adaptive filters in parallel. In this embodiment the multiple adaptive filters for cancelling NEXT noise introduced by the transmit paths of multiple transceivers (disturbers) into one receive path (victim) of a transceiver (transceiver 5) are trained in parallel. In
b shows a further embodiment of a prelearning for multiple adaptive filters in parallel. In this embodiment, only one disturber, i.e. the transmit path of transceiver 1 transmits training signals. The adaptive filters which are respectively coupled between this disturber transmit path and the respective victim receive paths of the other near-end transceivers (in
c shows a further embodiment wherein each of the multiple transceivers transmits training signals. In this embodiment, the NEXT cancellation filters between each receive path and each transmit paths of the multiple transceivers are trained in parallel. For simplicity,
As shown in
An embodiment for implementing a parallel training of echo noise and NEXT noise in the presence of a received signal in the victim receive path is shown in
In the above description, embodiments have been shown and described herein enabling those skilled in the art in sufficient detail to practice the teachings disclosed herein. Other embodiments may be utilized and derived there from, such that structural and logical substitutions and changes may be made without departing from the scope of this disclosure.
This Detailed Description, therefore, is not to be taken in a limiting sense, and the scope of various embodiments is defined only by the appended claims, along with the full range of equivalents to which such claims are entitled.
Such embodiments of the inventive subject matter may be referred to herein, individually and/or collectively, by the term “invention” merely for convenience and without intending to voluntarily limit the scope of this application to any single invention or inventive concept if more than one is in fact disclosed. Thus, although specific embodiments have been illustrated and described herein, it should be appreciated that any arrangement calculated to achieve the same purpose may be substituted for the specific embodiments shown. This disclosure is intended to cover any and all adaptations or variations of various embodiments. Combinations of the above embodiments, and other embodiments not specifically described herein, will be apparent to those of skill in the art upon reviewing the above description.
It is further to be noted that specific terms used in the description and claims may be interpreted in a very broad sense. For example, the terms “circuit” or “circuitry” used herein are to be interpreted in a sense not only including hardware but also software, firmware or any combinations thereof. The term “data” may be interpreted to include any form of representation such as an analog signal representation, a digital signal representation, a modulation onto carrier signals etc. The term “information” may in addition to any form of digital information also include other forms of representing information. The term “entity” may in embodiments include any device, apparatus circuits, hardware, software, firmware, chips or other semiconductors as well as logical units or physical implementations of protocol layers etc. Furthermore the terms “coupled” or “connected” may be interpreted in a broad sense not only covering direct but also indirect coupling.
It is further to be noted that embodiments described in combination with specific entities may in addition to an implementation in these entity also include one or more implementations in one or more sub-entities or sub-divisions of said described entity. For example, specific embodiments described herein described herein to be implemented in a transmitter, receiver or transceiver may be implemented in sub-entities such as a chip or a circuit provided in such an entity.
The accompanying drawings that form a part hereof show by way of illustration, and not of limitation, specific embodiments in which the subject matter may be practiced.
In the foregoing Detailed Description, it can be seen that various features are grouped together in a single embodiment for the purpose of streamlining the disclosure. This method of disclosure is not to be interpreted as reflecting an intention that the claimed embodiments require more features than are expressly recited in each claim. Rather, as the following claims reflect, inventive subject matter lies in less than all features of a single disclosed embodiment. Thus the following claims are hereby incorporated into the Detailed Description, where each claim may stand on its own as a separate embodiment. While each claim may stand on its own as a separate embodiment, it is to be noted that—although a dependent claim may refer in the claims to a specific combination with one or more other claims—other embodiments may also include a combination of the dependent claim with the subject matter of each other dependent claim.
It is further to be noted that methods disclosed in the specification or in the claims may be implemented by a device having means for performing each of the respective steps of these methods.
This application claims the benefit of the priority date of U.S. provisional application 61/142,910 filed on Jan. 7, 2009, the content of which are herein incorporated by reference.
Number | Date | Country | |
---|---|---|---|
61142910 | Jan 2009 | US |