Equalization on high-speed data channels with sparsity impulse response

Abstract

A physical layer transceiver for connecting a host device to a wired channel medium divided into a total number of link segments includes a host interface for coupling to the host device, a line interface for coupling to the wired channel medium, and a feed-forward equalization circuit (FFE) operably coupled to the line interface to add back time-scattered components into a signal. Each individual filter segment is selectably configurable to correspond to each individual link segment by adjusting a respective delay line. The FFE circuit also includes control circuitry configured to detect a signal energy peak of at least one specific link segment of the total number of link segments, and, upon detecting the signal energy peak of the specific link segment, configure each one of the respective individual filter segments to correspond to each specific link segment by adjusting a respective delay line.

Description

Technical Field

[0001] [Cross - Reference to Related Applications] This disclosure claims the benefit of U.S. Provisional Patent Application No. 63 / 105,127, filed on October 23, 2020, and commonly assigned, the entire disclosure of which is incorporated herein by reference.

[0002] This disclosure relates to the environmental setting of filters for equalizing high - speed data channels. More specifically, this disclosure relates to the distribution of filter taps for high - speed data channels having a sparse impulse response.

Background Art

[0003] The background description provided herein is for the purpose of generally providing the context of the present disclosure. The research of the inventors herein, as long as that research is described in this background section, is not admitted as prior art for the subject matter of the present disclosure, either explicitly or implicitly, in the same way as aspects of the description that may not be admitted as prior art at the time of filing.

[0004] Physical layer (PHY) devices for high-speed wired serial communications, such as 1000 Base-T or 10G Base-T lines, typically employ adaptive filters for equalization. A typical adaptive filter may include a feedforward equalizer (FFE) circuit. The adaptive filter can have multiple taps, each tap covering a segment of the equalizer circuit, corresponding to each segment of the wired link. The coefficients for each segment are adapted to an algorithm that maximizes the effectiveness of the equalizer circuit, such as least mean squares. In some relatively long-distance equalization scenarios, i.e., when link partners are not located on a single chip or circuit board, there may be a relatively long cable where only a relatively small number of points along it require equalization, and the rest of the cable is relatively quiet. The impulse response of such a link can be described as "sparse." This can specifically be the case for a channel with localized tensile attenuation, where the link only needs equalization at the vicinity of the attenuation. Sparsity impulse responses can lead to large-scale intersymbol interference (ISI), which limits the signal-to-interference noise ratio (SNIR) and degrades the channel's capabilities.

[0005] Equalization filters are relatively expensive and consume a considerable amount of power, making it impractical and undesirable to provide enough filter taps to cover the entire length of a long link. However, there can be negative consequences for not filtering specific link segments. For example, if secondary signal peaks, far from the main signal peak, arise from reflections of parts of the signal, it may be advantageous to include these secondary signal peaks as part of the overall signal power, thereby increasing the SNIR. However, without a feedforward equalization filter, these secondary signal peaks may be dismissed as noise and discarded. [Overview of the project]

[0006] According to an implementation of the subject matter of this disclosure, a physical layer transceiver for connecting a host device to a wired channel medium divided into a total number of link segments comprises a host interface for coupling to the host device, a line interface for coupling the wired channel medium, and a feedforward equalization circuit operably coupled to the line interface to add back time-scattered components into the signal. The feedforward equalization circuit comprises a plurality of filter segments. Each individual filter segment of the plurality of filter segments can be selectively configured to correspond to each individual link segment of the total number of link segments by adjusting its respective delay line. The feedforward equalization circuit also comprises a control circuit configured to detect the signal energy peak of at least one particular link segment of the total number of link segments, and, upon detecting each individual signal energy peak of at least one particular link segment of the total number of link segments, to configure each individual filter segment of the plurality of filter segments to correspond to each of the at least one particular link segment of the total number of link segments by adjusting its respective delay line.

[0007] In this first implementation of a physical layer transceiver, the number of filter segments may be less than the total number of link segments in the wired channel medium.

[0008] In a first implementation of such a physical layer transceiver, the filter segments may include a plurality of rover filter segments, each of which can be selectively configured to filter each of the total number of selectable link segments using its respective delay line, and the control circuit is configured to detect the signal energy peak of at least one particular link segment of the total number of link segments by selectively advancing the unused rover filter segments of the plurality of rover filter segments through the link segments by adjusting their respective delay lines, and to measure the transmit state of each of the total number of link segments as the unused rover filter segments pass through the link segments.

[0009] A third implementation of such a physical layer transceiver may include a correlation circuit configured to calculate the correlation between the received data and one or more errors in the received data. The control circuit may be configured to determine from the correlation the signal energy peak of at least one particular link segment of the total number of link segments.

[0010] In a fourth implementation of such a physical layer transceiver, the control circuit may be configured to adjust the gain of each of the individual filter segments of a plurality of filter segments corresponding to each of at least one particular link segment of the total number of link segments when it detects one signal energy peak in each of at least one particular link segment of the total number of link segments.

[0011] According to a fourth embodiment of the implementation, the control circuit may be configured to adjust the gain of each individual filter segment of a plurality of filter segments corresponding to each of at least one specific link segment of the total number of link segments by configuring each individual filter segment of a plurality of filter segments corresponding to each of at least one specific link segment of the total number of link segments.

[0012] In a fifth implementation of such a physical layer transceiver, the control circuit may be further configured to power down those of multiple filter segments that are not configured to correspond to at least one particular link segment of the total number of link segments when a signal energy peak segment is detected in fewer than the number of filter segments.

[0013] A method, as implemented in the subject matter of the present disclosure, filters a wired channel medium divided into a total number of link segments, wherein the wired channel medium comprises a filter circuit having a plurality of filter segments, and each individual filter segment of the plurality of filter segments can be configured to correspond to each individual link segment of the total number of link segments. The method includes detecting a signal energy peak in at least one specific link segment of the total number of link segments, and, upon detecting each individual signal energy peak in each of the at least one specific link segment of the total number of link segments, configuring each individual filter segment of the plurality of filter segments to correspond to each of the at least one specific link segment of the total number of link segments.

[0014] In a first implementation of such a method, a plurality of filter segments comprise a plurality of rover filter segments, each of which can be selectively configured to filter each link segment of the total number of link segments by adjusting its respective delay line, and detecting the signal energy peak of at least one particular link segment of the total number of link segments may include selectively advancing unused rover filter segments of the plurality of rover filter segments through the link segments by adjusting their respective delay lines, and measuring the transmit state of each link segment of the total number of link segments as the unused rover filter segments pass through the link segments.

[0015] In a second implementation of such a method, detecting the signal energy peak of at least one specific link segment of the total number of link segments may include calculating the correlation between the received data and one or more errors in the received data, and determining from the correlation the signal energy peak of at least one specific link segment of the total number of link segments.

[0016] A third implementation of such a method may further include detecting one signal energy peak in each of at least one particular link segment of the total number of link segments, and then adjusting the gain of each of the individual filter segments of a plurality of filter segments corresponding to each of at least one particular link segment of the total number of link segments.

[0017] According to the first aspect of the third implementation, adjusting the gain of each individual filter segment of a plurality of filter segments corresponding to each of at least one specific link segment of the total number of link segments may include constituting at least one additional gain of each individual filter segment of a plurality of filter segments corresponding to each of at least one specific link segment of the total number of link segments.

[0018] According to the second aspect of the third implementation, when the number of signal energy peak segments detected is less than the number of filter segments, it may further include powering down those of the multiple filter segments that are not configured to correspond to at least one particular link segment of the total number of link segments.

[0019] A wired communication system in an implementation of the subject matter of this disclosure includes a wired channel medium divided into a total number of link segments, and a plurality of physical layer transceivers coupled to the wired channel medium. Each of the plurality of physical layer transceivers includes a host interface for coupling to a host device, a line interface for coupling to the wired channel medium, and a feedforward equalizer circuit operably coupled to the line interface to add components scattered over time back into the signal. Each feedforward equalizer circuit includes a plurality of filter segments, each of which is selectably configurable by adjusting its delay line so that each individual filter segment of the plurality of filter segments corresponds to each individual link segment of the total number of link segments. Each feedforward equalizer circuit also includes a control circuit configured to detect a signal energy peak of at least one particular link segment of the total number of link segments, and, upon detecting one signal energy peak of each of the at least one particular link segment of the total number of link segments, to configure each individual filter segment of the plurality of filter segments to correspond to each of the at least one particular link segment of the total number of link segments by adjusting its delay line.

[0020] In this first implementation of a wired communication system, the number of filter segments in each physical layer transceiver may be less than the total number of link segments in the wired channel medium.

[0021] In a second implementation of such a wired communication system, each physical layer transceiver may include a plurality of rover filter segments, each of which can be selectively configured to filter each link segment of the total number of link segments by adjusting its respective delay line, and the control circuit may be configured to detect the signal energy peak of at least one particular link segment of the total number of link segments by selectively advancing the unused rover filter segments of the plurality of rover filter segments through the link segments by adjusting their respective delay lines, and to measure the transmit state of each link segment of the total number of link segments as the unused rover filter segments pass through the link segments.

[0022] In a third implementation of such a wired communication system, each physical layer transceiver may include a correlation circuit configured to calculate the correlation between one or more errors in the transmitted and received data. The control circuit may be configured to determine from the correlation the signal energy peak of at least one particular link segment of the total number of link segments.

[0023] In a fourth implementation of such a wired communication system, each physical layer transceiver may be configured such that, upon detecting a signal energy peak in each of at least one specific link segment of the total number of link segments, adjusts the gain of each of the individual filter segments of a plurality of filter segments corresponding to each of at least one specific link segment of the total number of link segments.

[0024] In a fifth implementation form of such a wired communication system, in each respective physical layer transceiver, a control circuit adjusts each respective delay line to configure at least one additional one of each individual filter segment corresponding to each one of at least one specific link segment of the total number of link segments, thereby adjusting the gain of each one of the individual filter segments of a plurality of filter segments corresponding to each one of at least one specific link segment of the total number of link segments, and can be configured to do so.

[0025] In a sixth implementation form of such a wired communication system, in each respective physical layer transceiver, when the number of segments of signal energy peaks is detected to be less than the number of a plurality of filter segments, the control circuit can be further configured to power down those of the plurality of filter segments that are not configured to correspond to each one of at least one specific link segment of the total number of link segments.

Brief Description of the Drawings

[0026] Further features, the nature, and various advantages of the present disclosure will become apparent upon consideration of the following detailed description, which is to be construed in conjunction with the accompanying drawings, in which like reference numerals refer to like parts throughout.

[0027] [Figure 1] It represents an Ethernet physical link that can be an in-vehicle Ethernet (registered trademark) physical link where embodiments of the subject matter of the present disclosure can be used.

[0028] [Figure 2] It represents an Ethernet physical link that can be an enterprise Ethernet physical link where embodiments of the subject matter of the present disclosure can be used.

[0029] [Figure 3]This is a module diagram of a physical layer transceiver in an implementation form of the subject matter of this disclosure.

[0030] [Figure 4] This document shows the correspondence between each filter segment and channel segments, based on the implementation of the subject matter of this disclosure.

[0031] [Figure 5] This disclosure shows a channel having a mobile auxiliary rover filter segment, as an implementation of the subject matter of this disclosure.

[0032] [Figure 6] This document demonstrates how to replace one filter segment with another, as demonstrated by an implementation of the subject matter of this disclosure.

[0033] [Figure 7] Figure 6 shows the resulting channel after substitution.

[0034] [Figure 8] This is a flowchart illustrating the operation method based on the implementation of the subject matter of this disclosure. [Modes for carrying out the invention]

[0035] As described above, physical layer devices for high-speed wired serial communication, such as 1000 Base-T or 10G Base-T lines, typically use adaptive filters for equalization. A typical adaptive filter can have multiple taps, each tap covering a segment of the equalization circuit, corresponding to each segment of the wired link. The coefficients for each segment are adapted to an algorithm that maximizes the effectiveness of the equalization circuit, such as least mean squares.

[0036] Equalization filters are relatively expensive and consume a considerable amount of power, making it impractical and undesirable to provide enough filter taps to cover the entire length of a long link. However, there can be negative consequences for not filtering specific link segments. For example, if secondary signal peaks, far from the main signal peak, arise from reflections of parts of the signal, it may be advantageous to include these secondary signal peaks as part of the overall signal power, thereby increasing the SNIR. However, without a feedforward equalization filter, these secondary signal peaks may be dismissed as noise and discarded.

[0037] Accordingly, depending on the implementation of the subject matter of this disclosure, a feedforward equalizer over long-distance wired links—i.e., wired links beyond a single chip or circuit board component—may be provided in a “rover” filter segment, which may be configured to cover different portions of the link. The rover filter segments are not physically movable, but rather each rover filter segment is configured to use a programmable delay line, with each rover filter segment altering each portion of the link it corresponds to. The rover filter segments may be configured during the training interval prior to execution time—i.e., prior to the enterdata mode—to filter link segments that require equalization.

[0038] Filtering using feedforward equalization avoids the error propagation from one symbol to another that can occur with any form of feedback equalization (e.g., decision feedback equalization, i.e., DFE), with or without rover filter segments, because feedback equalization is performed only after the feedback decision has been made (e.g., by a data slicer) based on data that has not yet been equalized. Feedforward equalization, on the other hand, is performed before the decision is made and uses data that has been "feedforwarded" (using an appropriate delay in the current data), and therefore the current decision is not affected by erroneous feedback from decisions on unequalized data. Also, using feedforward equalization increases the power of the signal because it adds and returns components that may have been scattered over time into the signal. Therefore, feedforward equalization is particularly advantageous for longer cables, where variability in the channel response may be higher, including higher variability in the delay of reflections.

[0039] The wired connection in question could be, for example, a fixed cable in a data center under typically stable conditions. Under these conditions, most of the cable may be free of reflection and interference, but connectors or bends in the cable can cause reflection and additional interference in neighboring sections of the cable. Such links can be considered to have a sparsity impulse response as defined above, but equalization may be necessary only for link segments adjacent to cable connectors or bends in the cable where signal artifacts may be present above the intended signal.

[0040] Since such cables do not typically move, it can be expected that the segments requiring filtering will always be the same. Nevertheless, according to the implementation of the subject matter of this disclosure, when a new link is established on such a cable (e.g., when power is turned on), the link is measured (as described in more detail below), and the FFE rover filter tap is moved to the segment of the link where a signal energy peak is detected (as noted above, this does not have to be an actual physical one).

[0041] In other situations, the wired connection in question could be, for example, an in-vehicle or other mobile Ethernet cable. Such cables may also be susceptible to interference only at neighboring connectors or bends in the cable. However, if the cable is not properly or adequately secured to the vehicle, the cable may move (including bending or twisting, depending on the situation) as the vehicle accelerates or road conditions change. The cable may also be exposed to sufficiently large temperature fluctuations—for example, resulting from changes in weather conditions and solar loads—which alter the physical environmental conditions of the cable, along with the rest of the vehicle. Thus, segments of the cable requiring equalization may be more likely to change from the establishment of one link to the establishment of the next. In either case, as in the case of more fixed cables, according to the implementation of the subject matter of this disclosure, when a new link is established on such a cable (e.g., upon power-up), the link is measured (as described in more detail below), and the FFE rover filter tap is moved to the segment of the link where the signal energy peak is detected.

[0042] Whether in a fixed or portable application, one function of the FFE rover segment is to ensure that interference peaks are discarded while, conversely, the power of any peaks derived from the intended signal (e.g., due to reflection) is added back to the primary signal path.

[0043] As used herein, “rover filter configuration” includes the number of active rover filter segments and a programmed delay for each rover filter segment (determining its position relative to the link segments). The rover filter configuration may also include the output gain and / or coefficient values ​​for each rover filter segment. The rover filter configuration can be changed by changing the number of active rover filter segments (either increasing or decreasing that number), changing the position of the active rover filter segments (e.g., by adjusting the delay), or adjusting the gain of a particular rover filter segment (which may be achieved by arranging multiple rover filter segments parallel to each other at a particular position).

[0044] As described above, before the rover filter configuration can be established, the state of the channel should be detected to locate the channel along which it has a signal energy peak. In some implementations, such detection can be performed by a mobile auxiliary rover or by calculating the correlation between the receive and transmit paths in the physical layer transceiver, both of which are described below.

[0045] In particular, in the implementation of auxiliary rover filters, an otherwise unused rover filter segment, which may be referred to as an auxiliary rover filter segment, can "move" the link by continuously adjusting the programmable delay line of the auxiliary rover filter segment. At each link segment location, the auxiliary rover filter segment becomes adaptable, and then quality metrics (such as signal-to-noise ratio (SNR), signal-to-interference-to-noise ratio (SNIR), bit error rate, or decoder workload) with the present auxiliary rover filter segment are measured and compared to the same quality metrics measured before the auxiliary rover filter segment was added. If the quality metrics improve with the presence of the auxiliary rover filter segment, the rover filter segment is assigned to that link segment location.

[0046] In the implementation of correlation, the correlation calculation may be performed between the uncorrected signal in the physical layer transceiver's receiving path and the slicer error in the physical layer transceiver's receiving path, and it may act as a proxy for the link partner's transmitted signal (which is not directly available at the receiver). The correlation indicates the magnitude of the signal energy peak without delay and filtering. A rover filter segment may be added at a position where the unfiltered signal energy peak is above a predetermined threshold, as indicated by the correlation output. For confirmation, the SNR or SNIR can be measured after adding the new rover filter segment at a position where the unfiltered signal energy peak is above a predetermined threshold. The new rover filter segment is maintained at a position where the unfiltered signal energy peak is above a predetermined threshold only if the SNR or SNIR improves by a predetermined amount.

[0047] The subject matter of this disclosure can be better understood by referring to Figures 1 through 8.

[0048] Implementations of the subject matter of this disclosure can be found in physical layer transceivers (PHYs) of fixed Ethernet links (e.g., “Enterprise”, “Data Center”, “Cloud”, “Carrier”, or “Metro”, links), or in automotive or other mobile Ethernet links.

[0049] An in-vehicle Ethernet physical link 100, in which an implementation of the subject matter of this disclosure may be used to connect two physical layer transceivers 103, is shown in Figure 1, one of which may be located in the vehicle's electronic control unit (ECU) and the other in the vehicle's functional module. Each PHY 103 is connected to a channel medium 101 via its respective connector 102, which in its implementation may be a single shielded or unshielded twisted copper wire pair 111, or a cable including a coaxial cable or optical fiber.

[0050] A single-cable physical link 100 may also be used in enterprise implementations, but an enterprise Ethernet physical link 200 in which the implementation of the subject matter of this disclosure may be used is shown in Figure 2, connecting two physical layer transceivers 203 corresponding to each link partner, each of which may be located in its respective data processing or storage device. Each PHY 203 is connected to a channel medium 201 via its respective connector 202, which in its implementation may include four shielded or unshielded twisted copper wire pairs 211, 221, 231, 241, or four coaxial cables or optical fibers, or a combination thereof.

[0051] From the perspective of this disclosure, PHY103 and PHY203 are identical in relevant respects. The implementation of PHY300 shown in Figure 3 can be used as either PHY103 or PHY203, for example.

[0052] The PHY300 connects a host device, such as an automotive module or a functional module 301 which may be a data processing or storage module for an enterprise system, to a wired channel medium (cable) 101 / 201 using an encoding and decoding circuit 302. One or more adaptive filters, which may be a feedforward equalizer 303, filter out the effects of interference and redirect reflected data signals back to the main signal path.

[0053] In some implementations according to the subject matter of this disclosure, the PHY 300 transmits data from the functional module 301 to the wired channel medium (cable) 101 / 201 via the hybrid circuit 310 and line interface 323 through the host interface 322 and digital-to-analog converter 304, and receives a remote (target) signal from the wired channel medium (cable) 101 / 201 via the line interface 323 and hybrid circuit 310, and via the analog-to-digital converter 305. The received remote (target) signal is processed by an adaptive filter circuit which may include an adaptive feedforward equalizer 303. The transmitted signal is similarly filtered at another PHY 300 at the other end of the wired channel medium (cable) 101 / 201, where it is received as a remote (target) signal. The adaptive feedforward equalizer 303 can be used to remove interference, but also to direct reflected peaks of the received data back to the main data path.

[0054] While the implementations of the subject matter of this disclosure are described herein in the context of a transceiver, the subject matter of this disclosure can also be used in a receiver, in which case the transmission path including the digital-to-analog converter 304 may be omitted along with the hybrid circuit 310.

[0055] The adaptive feedforward equalizer 303 may be implemented as a rover filter. The programmable delay of each segment of the rover filter 303 may be adjusted to "advance" or "move" each rover filter segment (also referred to as a "link segment") corresponding to each channel segment. As noted above, the rover filter segments are not physically movable, but rather each rover filter segment is configured to use a programmable delay line that changes each portion of the link to which each rover filter segment corresponds. The controller 320 may include a detection circuit 321 that can adjust the filter configuration, including both the position and filter parameters of each segment, based on the detected channel transmission state, as will be described in more detail below.

[0056] Figure 4 shows the correspondence of each filter segment to the segments of channel 400 in an implementation of the subject matter of this disclosure. The signal energy in channel 400 may include the main signal peak 401, additional peaks 402 that may result from signal reflection, and other peaks 403 that may result from various other forms of interference. The main channel 400 can be expected to be relatively quiet, with no signal reflection or interference present.

[0057] For illustrative purposes, the origin of the reflected peak 402 can be understood by referring to Figure 1, where the transmitted signal is represented by arrow 120. If an additional connector 112 is present on channel 101, a portion 121 of the signal 120 is reflected back at 122, then reflected and transmitted at 123, with each reflection potentially producing a peak 402.

[0058] Slots 404 represent positions to which filter segments can be assigned. These positions correspond to physical segments of the channel medium. As noted above, each rover filter segment can "move forward" or "move" to various slots 404 by adjusting its programmable delay line (again, as noted above, this does not necessarily have to be an actual physical movement). Therefore, filling all slots 404 with filter segments can lead to an over-engineered implementation with excessive power consumption, rather than allowing rover filter segments to be selectively assigned to filter slots 404 that correspond to channel segments requiring filtering, even if other slots 404 remain without filter segments.

[0059] For example, the arrangement of active rover filter segments 405 can be configured to handle a channel having the impulse profile shown in Figure 4. In arrangement 405, most of the filter segments 415 are concentrated to correspond to the ends of the channel, in which case, in addition to the main signal peak 401, there is a peak 402 representing a reflection from the connector. Additional active rover filter segments 425, 435 are configured to correspond to an additional reflection peak 403 located downstream of peak 402.

[0060] Arrangement 405 can represent the rover filter configuration established when the channel link is established. That is, if slot 404, shown by the dashed line, indicates the potential location of the rover filter, then segment 415, shown by the solid line, represents the location where the active rover filter segment is deployed.

[0061] As described above, the detection circuit 321 of the controller 320 can measure the transmission status of channel 400 using one or more mobile rover filter segments or correlations for the purpose of positioning rover filter segments.

[0062] In some implementations, the detection circuit 321 of the controller 320 measures the transmit state of a channel using one or more mobile rover filter segments. "Additional," "supplementary," or "auxiliary" rover filter segments, i.e., rover filter segments that have not yet been deployed at a particular channel segment location, can be temporarily deployed (e.g., sequentially) at each channel segment location as mobile rover filter segments. Alternatively, auxiliary rover filter segments can be temporarily deployed even for channel segments that have already been filtered and whose existing filter segments are being verified as correct.

[0063] An "additional" "mobile" rover filter segment is shown in Figure 5, which is similar to Figure 4 in that it shows the already deployed rover filter segments 415, 425, and 435. In addition, a mobile rover filter segment 501 that moves to all positions to detect new or increased interference requiring the deployment of an additional rover filter segment is represented by multiple phantom filter segment positions, and arrow 502 indicates that, under the control of the controller 320, the mobile rover filter segment 501 is "moved" from one phantom filter segment position to another phantom filter segment position. If signal energy that may be a reflection of the desired signal is detected at an unfiltered channel segment position by the mobile rover filter segment 501, a dedicated rover filter segment can be deployed at that channel segment position. If the channel impulse response shown in Figures 4 and 5 is modified (not shown), and as a result the signal energy is detected at the channel segment location filtered by the mobile rover filter segment 501 whose function is detection, then a new dedicated rover filter segment can be deployed at that channel segment location, or an existing rover filter segment can be modified. Using feedforward equalization, once deployed, the dedicated rover filter segment can detect reflected signal energy that may be added to the peak 401 of the main signal, while interference can be removed by other techniques such as echo cancellation.

[0064] In some other implementations, the detection circuit 321 of the controller 320 measures the transmit state of the channel using correlation. The correlator 307 of the physical layer transceiver 300 may be used to perform correlation between the received symbol 311 and the slicer error 312 derived by the slicer 316 from the output 313 of the adaptive feedforward equalizer 303. In this case, the slicer error 312 may act as a proxy for the signal transmitted by the link partner (which is not directly available at the receiver). Alternatively (not shown), the correlator 307 can perform correlation between the slicer error 312 output by the slicer 316 and the output 313 of the adaptive feedforward equalizer 303, which is the input to the slicer 316 and represents a noisy version of the signal transmitted by the link partner.

[0065] As a further alternative (also not illustrated), the second input to the correlator 307 may be a replica of a pseudo-random bit sequence (PRBS) transmitted by the link partner during link training, in addition to the slicer error 312 output by the slicer 316. The seed for the PRBS generator can be estimated at the receiver and used to regenerate the entire training sequence of the remote transmitter. This sequence can then be used by the correlator 307 to measure the power peak of the remote signal impulse response as an alternative to the detected symbol.

[0066] The correlation output indicates the delay and the magnitude of the reflected signal energy or interference. Dedicated rover filter segments may be deployed at channel segment locations where the reflected signal energy or interference exceeds a certain threshold, as indicated by the correlation output.

[0067] According to an implementation of the subject matter of this disclosure, the presence of signal energy in a channel segment is detected by a mobile rover filter segment 501, or by a correlator 307, or by any other technique, after a dedicated rover filter segment has been deployed in that channel segment. Then, a quality metric (e.g., the SNIR described above) in that channel segment with the deployed dedicated filter can be compared to the quality metric in that channel segment before the dedicated filter was deployed. If the presence of the dedicated filter does not improve the quality metric, then the dedicated filter can be removed, and the system can return to the previous state.

[0068] In one scenario, the link segment from which signal energy is detected may be a link segment that is not currently equalized. In that scenario, by adjusting the programmable delay line, the rover filter segment can be "moved" or advanced to correspond to the link segment that needs equalization, thereby adding the energy of that segment to the main signal.

[0069] In another scenario, a link segment in which a change in transmission state is detected may be a link segment that is currently being equalized and still requires equalization, but also requires adjustment of filter parameters (e.g., filter gain and / or coefficients). In some implementations, the filter segment in operation may be adjusted. In other implementations, instead of attempting to adjust the filter segment in operation, a new filter segment may be advanced so that the existing filter segment corresponds to the link segment that requires adjustment, or the existing filter segment may be advanced away from the link segment. In some such implementations, to prevent a discontinuity in transmission, the gain of the new filter segment is gradually increased while the gain of the existing filter segment is gradually decreased, until the gain of the existing filter segment reaches zero, allowing the existing filter segment to advance away from the link segment being equalized.

[0070] An implementation in which an existing filter segment is adjusted by moving a new filter segment to its position is illustrated in Figure 6, similar to Figure 4. Arrow 601 indicates a position where there is signal energy in the channel and a filter is present, but for some reason, an increase in the gain of rover filter segment 602 or other parameters is required. This increase is achieved in this implementation by moving a new rover filter segment 612 to its position. The higher order of the filtering parameters of rover filter segment 612 relative to rover filter segment 602 is graphically shown by the relative sizes of rover filter segment 602 and rover filter segment 612, as depicted. As mentioned above, to prevent discontinuity, when rover filter segment 612 is moved to its position, rover filter segment 602 can be reduced until it is completely inoperable, while rover filter segment 612 can be adapted by increasing until it reaches the required parameters. The final result is shown in Figure 7, in which case the filter configuration 405 is replaced by the filter configuration 705 in which the rover filter segment 612 replaces the rover filter segment 602, and the filtering parameters are represented in the drawing not by relative size, but rather by differences in hatching direction. Alternatively, the filtering characteristics of the rover filter segment can be increased simply by moving the mobile rover filter segment 612 to a position parallel to the rover filter segment 602, in which case Figure 6 represents the final state.

[0071] An implementation of Method 800 according to the present disclosure is illustrated in Figure 8, starting from 801, and involves detecting a signal energy peak in at least one specific link segment of a total number of link segments of a wired channel medium having a filter circuit comprising a plurality of filter segments, wherein each individual filter segment of the plurality of filter segments can be configured to correspond to each individual link segment of the total number of link segments. According to Alternative Embodiment 811, detection 801 is performed by selectively advancing an unused rover filter segment from a plurality of rover filter segments through a link segment and measuring the transmit state of each link segment of the total number of link segments as the unused rover filter segment passes through the link segment. According to Alternative Embodiment 821, detection 801 is performed by calculating a correlation between one or more errors in transmitted data and received data and determining a signal energy peak in at least one specific link segment of the total number of link segments from the correlation.

[0072] Next, when 802 detects one signal energy peak for each of at least one specific link segment of the total number of link segments, each of the individual filter segments of the multiple filter segments is configured to correspond to each of at least one specific link segment of the total number of link segments, and method 800 ends.

[0073] Therefore, it can be seen that the feedforward equalizer for long-distance wired links is provided with a rover filter segment that can be configured to cover various parts of the link using a programmable delay line.

[0074] As used herein and in the following claims, “one of A and B” shall be interpreted as “A or B.”

[0075] The above is merely an example of the principle of the invention, and the invention can be implemented in embodiments other than those described, which are presented for illustrative purposes only, not limitation. It should be noted that the present invention is limited only by the subsequent claims.

Claims

1. A physical layer transceiver for connecting a host device to a link of a wired channel medium including a high-speed data channel having a sparse impulse response, wherein the link of the wired channel medium is divided into link segments, each representing a portion of the entire link, and the physical layer transceiver A host interface for connecting to a host device; A line interface for connecting to the aforementioned wired channel medium; and A feedforward equalization circuit operably connected to the line interface to add components scattered over time back into the signal, Multiple filter segments, wherein each individual filter segment of the multiple filter segments is configured to be selectable to correspond to each individual link segment among all link segments by filtering the link segments that require equalization and adjusting their respective delay lines, and To detect the signal energy peak of at least one specific link segment among all link segments; and When detecting one of the signal energy peaks in each of the at least one specific link segment among all the link segments, the delay lines are adjusted to configure each of the individual filter segments of the plurality of filter segments to correspond to each of the at least one specific link segment among all the link segments. A control circuit configured to perform the following actions. Feedforward equalization circuit having Equipped with, The plurality of filter segments comprises a plurality of rover filter segments, each rover filter segment is configured to use a programmable delay line to modify a portion of the link to which the rover filter segment corresponds, and to filter selectable link segments among all link segments. The control circuit is configured to detect a signal energy peak in at least one specific link segment among all the link segments, based on the measured transmission state of each link segment. A physical layer transceiver.

2. The physical layer transceiver according to claim 1, wherein the number of filter segments is less than the total number of link segments of the wired channel medium.

3. The physical layer transceiver according to claim 1 or 2, wherein the control circuit is configured to detect the signal energy peak of at least one specific link segment among all link segments by adjusting each delay line, thereby changing the link segment to which the unused rover filter segment of the rover filter segment corresponds.

4. A correlation circuit configured to calculate the correlation between data received from the wired channel medium and one or more errors in the received data, The control circuit is configured to determine the signal energy peak of at least one specific link segment among all link segments from the correlation. A physical layer transceiver according to any one of claims 1 to 3.

5. The physical layer transceiver according to any one of claims 1 to 4, wherein the control circuit is configured to adjust the gain of the individual filter segments of the plurality of filter segments corresponding to each of the at least one specific link segments among all link segments when it detects one of the signal energy peaks of each of the at least one specific link segments among all link segments.

6. The physical layer transceiver according to claim 5, wherein the control circuit is configured to adjust the gain of each of the individual filter segments of the plurality of filter segments corresponding to each of the at least one specific link segment among all the link segments by configuring each of the individual filter segments of the plurality of filter segments corresponding to each of the at least one specific link segment among all the link segments.

7. The physical layer transceiver according to any one of claims 1 to 6, wherein the control circuit is further configured to power down the filter segments of at least one particular link segment among all link segments that are not configured to correspond to each of the filter segments when the number of signal energy peak segments detected is less than the number of filter segments.

8. A method for filtering links in a wired channel medium including a high-speed data channel having a sparse impulse response, wherein the links in the wired channel medium are divided into link segments, each representing a portion of the entire link, and a physical layer transceiver for connecting a host device to the wired channel medium comprises a filter circuit having a plurality of filter segments, each individual filter segment of the plurality of filter segments being configurable to filter the link segments requiring equalization and to correspond to each individual link segment of all link segments, To detect the signal energy peak of at least one specific link segment among all link segments; and When one of the signal energy peaks is detected in each of the at least one specific link segment among all link segments, each of the individual filter segments of the plurality of filter segments corresponding to each of the at least one specific link segment among all link segments is configured. Equipped with, The plurality of filter segments comprises a plurality of rover filter segments, each rover filter segment is configured to use a programmable delay line to modify a portion of the link to which the rover filter segment corresponds, and to filter selectable link segments among all link segments. Detecting the signal energy peak includes detecting the signal energy peak in at least one specific link segment among all link segments, based on the measured transmission state of each link segment. A method for filtering wired channel media.

9. A method for filtering a wired channel medium according to claim 8, wherein detecting the signal energy peak of at least one specific link segment among all link segments is further comprising adjusting the respective delay lines such that an unused rover filter segment modifies the corresponding link segment of the rover filter segment.

10. Detecting the signal energy peak of at least one specific link segment among all link segments to calculate the correlation between the data received from the wired channel medium and one or more errors in the received data; and The correlation includes determining the signal energy peak of at least one specific link segment among all link segments, A method for filtering a wired channel medium according to claim 8 or 9.

11. A method for filtering a wired channel medium according to any one of claims 8 to 10, further comprising detecting one of the signal energy peaks of each of the at least one specific link segments among all link segments, and adjusting the gain of the individual filter segments of the plurality of filter segments corresponding to each of the at least one specific link segments among all link segments.

12. A method for filtering a wired channel medium according to claim 11, wherein adjusting the gain of the individual filter segments of the plurality of filter segments corresponding to each of the at least one specific link segment among all link segments constitutes at least one additional one of the individual filter segments of the plurality of filter segments corresponding to each of the at least one specific link segment among all link segments.

13. A method for filtering a wired channel medium according to claim 11 or 12, further comprising powering down the filter segments of the plurality of filter segments that are not configured to correspond to each of the at least one particular link segment among all link segments when the number of signal energy peak segments detected is less than the number of the plurality of filter segments.

14. A link of a wired channel medium including a high-speed data channel having a sparse impulse response, wherein the link of the wired channel medium is divided into link segments, each representing a portion of the entire link; and A plurality of physical layer transceivers connected to the wired channel medium, wherein each of the plurality of physical layer transceivers filters the link segment that requires equalization, and each of the plurality of physical layer transceivers performs Each host interface for connecting to each host device; Each line interface for connecting to the aforementioned wired channel medium; and Each feedforward equalization circuit is operably connected to each of the line interfaces to add the time-scattered components back into the signal, By adjusting each delay line, each individual filter segment of the multiple filter segments can be selectively configured to correspond to each individual link segment among all link segments, and To detect the signal energy peak of at least one specific link segment among all link segments; and When detecting one of the signal energy peaks in each of the at least one specific link segment among all link segments, the delay lines are adjusted to configure each of the individual filter segments of the plurality of filter segments to correspond to each of the at least one specific link segment among all link segments. A control circuit configured to perform Each of the feedforward equalization circuits, including Multiple physical layer transceivers having Equipped with, The plurality of filter segments comprises a plurality of rover filter segments, each rover filter segment is configured to use a programmable delay line to modify a portion of the link to which the rover filter segment corresponds, and to filter selectable link segments among all link segments. The control circuit is configured to detect a signal energy peak in at least one specific link segment among all the link segments, based on the measured transmission state of each link segment. Wired communication system.

15. The wired communication system according to claim 14, wherein in each physical layer transceiver, the number of filter segments is less than the total number of link segments of the wired channel medium.

16. In each physical layer transceiver; The control circuit is configured to detect the signal energy peak of at least one specific link segment among all link segments by adjusting each delay line, thereby changing the link segment to which the unused rover filter segment of the rover filter segment corresponds. The wired communication system according to claim 14 or 15.

17. Each physical layer transceiver is: The system includes a correlation circuit configured to calculate the correlation between one or more errors in the transmitted data and the data received from the wired channel medium, The control circuit is configured to determine the signal energy peak of at least one specific link segment among all link segments from the correlation. A wired communication system according to any one of claims 14 to 16.

18. The wired communication system according to any one of claims 14 to 17, wherein in each physical layer transceiver, the control circuit is configured to adjust the gain of the individual filter segments of the plurality of filter segments corresponding to each of the at least one specific link segments among all link segments when it detects one of the signal energy peaks of each of the at least one specific link segments among all link segments.

19. The wired communication system according to claim 18, wherein in each physical layer transceiver, the control circuit is configured to adjust the gain of each of the individual filter segments of the plurality of filter segments corresponding to each of the at least one specific link segment in all the link segments by adjusting each delay line, thereby constituting at least one additional one of each of the individual filter segments of the plurality of filter segments corresponding to each of the at least one specific link segment in all the link segments.

20. The wired communication system according to any one of claims 14 to 19, wherein in each physical layer transceiver, the control circuit is further configured to power down the filter segments of the plurality of filter segments that are not configured to correspond to each of the at least one particular link segment among all link segments when the number of signal energy peak segments detected is less than the number of the plurality of filter segments.

Images