CN112166558B

CN112166558B - Dynamic constellation adaptation of the limiter

Info

Publication number: CN112166558B
Application number: CN201980033587.7A
Authority: CN
Inventors: Y·阿藏科
Original assignee: MACOM Technology Solutions Holdings Inc
Current assignee: MACOM Technology Solutions Holdings Inc
Priority date: 2018-05-18
Filing date: 2019-05-15
Publication date: 2024-11-12
Anticipated expiration: 2039-05-15
Also published as: CN112166558A; EP3794729A1; WO2019222373A1; EP3794729A4

Abstract

Systems and methods for adaptive thresholding for constellation selection based on statistical distribution of received data symbols. The expected ratio of received symbols to values within a particular range is preset based on an expected statistical distribution of data symbols across multiple constellations. An adapted threshold is then obtained based on the desired ratio. Further, the constellation value may be based on a statistical distribution of received data symbols. The count ratio of received symbols falling within the first range to all symbols in the setting is compared to the expected ratio. The first value is repeatedly adjusted to adjust the first range until the count rate equals the expected rate. The final first value is used to determine the optimally adapted constellation.

Description

Dynamic constellation adaptation of limiter

Technical Field

The present disclosure relates generally to the field of signal processing in communication systems, and more particularly to the field of demodulation mechanisms in receivers.

Background

Multi-level pulse amplitude modulation (Multi-level Pulse Amplitude Modulation, PAM) has become the modulation mechanism of choice in signal transmission, both between chips on a printed circuit board (Printed Circuit Board, PCB) and from one end of a long-haul optical fiber to the other. On the transmitter side, the amplitude of the carrier pulse varies according to the sampled value of the message signal and based on the constellation (level) level of a particular PAM-N scheme. Accordingly, on the receiver side, demodulation is performed based on constellation levels by detecting the amplitude level of the carrier at each period.

As a signal is transmitted from a transmitter to a receiver, various factors in the communication channel may cause variations in the shape and amplitude of the signal, such as non-linear variations due to noise and phase interference. When characterized using eye diagrams, non-linear or amplitude compression can change the eye height of different transition eyes, resulting in errors due to lower Signal-Noise-Ratio (SNR).

At the receiver side, when a signal affected by noise is converted into a digital signal and demodulated, it is likely to map to a constellation point that is not exactly the same as the signal constellation level. A slicer (sheer) is used to determine which signal constellation level is closest to the received symbol based on a set of thresholds, typically defined as uniformly spaced. Unfortunately, nonlinearities may cause received symbols to be closer to another constellation than transmitted symbols. Since the nominal threshold used in the limiter does not take into account the nonlinearity, incorrect modulation tends to occur. In the prior art, various computationally intensive calculations have been developed to identify the actual closest signal constellation level. These calculations consume a significant amount of valuable computing resources in the receiver.

Disclosure of Invention

Embodiments of the present disclosure provide a cost-effective mechanism to dynamically adapt the constellation selection threshold and constellation level to signal nonlinear distortion, thereby facilitating accurate data recovery during demodulation at the receiver.

Embodiments of the present disclosure use threshold adaptation logic to dynamically adapt the threshold of the slicer to non-linearities or other distortions in the received signal based on the statistical distribution of the data symbols. In some embodiments, the slicer is configured to map the received symbols to an N-level pulse amplitude modulation (PAM-N) scheme of N constellations using N-1 thresholds. In general, the modulated and transmitted data symbols are distributed over N scaled constellations substantially at some known percentage or ratio. Typically, the plurality of symbols are substantially evenly distributed over the N scaled constellations when they are transmitted. Thus, at the limiter of the receiver, inputs within a particular range defined by a best-fit threshold (e.g., below a particular threshold) are expected to constitute a corresponding particular ratio (desired ratio) of the total input, regardless of signal nonlinearities. In order to find a best-fit threshold that deviates from the calibrated threshold, threshold adaptation logic (1) identifies a first value that causes the slicer input falling within a first range to form a first ratio with the total slicer input, wherein the first ratio is the desired ratio minus the error ratio, the ratio being substantially less than the desired ratio; and (2) identifying a second value that causes the limiter inputs within the second range to constitute a second ratio of the total limiter input, wherein the second ratio is the desired ratio plus the same error ratio. An adapted threshold value is then derived based on the first and second values. In some embodiments, the adaptation logic of the threshold value uses a comparator to compare the slicer input to the first or second value and thus uses a counter to maintain a count of the slicer input below the first or second value (first or second range). Thus, the ratio of the input to the total input in the first or second range may be calculated based on the count. The first or second value is adjusted until the first or second ratio is reached. The final first or second value is then used to calculate the best-fit threshold by simple arithmetic operations.

According to embodiments of the present disclosure, a set of thresholds for a slicer may be dynamically determined and adapted to statistically signal non-linearities and amplitude compression. As a result, data demodulation and data recovery at the receiver can be advantageously performed with a significantly reduced error rate. The threshold adaptation process does not involve complex processing or computationally intensive computations and can be implemented using simple circuitry, including comparators and counters, for example. Thus, the present disclosure provides reduced design and development costs and operating power consumption compared to conventional approaches.

In accordance with another aspect of the disclosure, constellation adaptation logic is used to dynamically adapt the values of the constellation to signal nonlinearities and other distortions based on the statistical distribution of the data symbols. In some embodiments, to find an optimally adapted constellation that deviates from the nominal constellation, the constellation adaptation logic compares the first value to a set of slicer inputs and maintains a count of slicer inputs (e.g., below the first value) that fall within a particular range defined by the first value based on the comparison. A count ratio of the number of inputs to the number of input sets within a range (e.g., below a first value) is also derived. The first value is adjusted until the count ratio reaches a count ratio associated with the scaled constellation. The first value is then designated as the adapted constellation to be used by the slicer.

According to embodiments of the present disclosure, slicer constellation levels may be statistically determined dynamically and adapted for signal nonlinearity and amplitude compression. As a result, data demodulation and data recovery can be advantageously performed at the receiver with a further reduced error rate. Similarly, constellation adaptation does not involve computationally intensive computations or complex circuit designs. Therefore, design and development costs and operating power consumption can be advantageously further reduced as compared with the conventional method.

The foregoing is a summary and thus contains, by necessity, simplifications, generalizations, and omissions of detail; accordingly, those skilled in the art will appreciate that this summary is illustrative only and is not intended to be in any way limiting. Other aspects, inventive features, and advantages of the present invention, as defined solely by the claims, will become apparent in the non-limiting detailed description set forth below.

Drawings

Embodiments of the present invention will be better understood from a reading of the following detailed description taken in conjunction with the drawings in which like reference designators refer to like elements.

Fig. 1 illustrates an exemplary data range for determining a best-fit threshold for constellation selection based on a desired statistical distribution of data symbols across multiple constellations in accordance with an embodiment of the present disclosure.

Fig. 2 illustrates a configuration of an exemplary threshold adaptation unit capable of dynamically adapting a constellation selection threshold used by a slicer according to an embodiment of the present disclosure.

Fig. 3A is a flow chart depicting an exemplary process of determining an adaptation threshold for constellation selection in accordance with an embodiment of the present disclosure.

Fig. 3B is a flowchart depicting an exemplary process of determining a first value or a second value associated with a threshold value in accordance with an embodiment of the present disclosure.

Fig. 4 illustrates an exemplary data range for determining a best-fit constellation based on a desired statistical distribution of data symbols across multiple constellations in accordance with an embodiment of the present disclosure.

Fig. 5 illustrates a configuration of an exemplary constellation adaptation unit capable of dynamically determining a best-fit constellation according to an embodiment of the present disclosure.

Fig. 6 is a flow chart depicting an exemplary process of determining a best-fit constellation for demodulation in accordance with an embodiment of the present disclosure.

Fig. 7 illustrates a network signal transmission system including constellation adaptation logic and constellation selection threshold adaptation logic at a receiver in accordance with an embodiment of the present disclosure.

Fig. 8 illustrates a set of simulation results comparing a Symbol Error Rate (SER) generated using a default slicer threshold to a SER generated according to obtaining an optimal disclosure threshold using an embodiment of the present disclosure.

Fig. 9 shows another set of simulation results comparing the default slicer threshold generated SER with the best-fit threshold generated SER obtained in accordance with an embodiment of the present disclosure.

Detailed Description

Reference will now be made in detail to the preferred embodiments of the present invention, examples of which are illustrated in the accompanying drawings. While the invention will be described in conjunction with the preferred embodiments, it will be understood that they are not intended to limit the invention to these embodiments. On the contrary, the invention is intended to cover alternatives, modifications and equivalents, which may be included within the spirit and scope of the invention as defined by the appended claims. Furthermore, in the following detailed description of embodiments of the invention, numerous specific details are set forth in order to provide a thorough understanding of the invention. However, it will be recognized by one of ordinary skill in the art that the present invention may be practiced without these specific details. In other instances, well-known methods, procedures, components, and circuits have not been described in detail as not to unnecessarily obscure aspects of the embodiments of the present invention. Although a method may be described as a series of numbered steps for clarity, the numbers do not necessarily indicate the order of the steps. It should be understood that some steps may be skipped, performed in parallel, or performed without the requirement of a strict order of sequence. The drawings showing embodiments of the invention are semi-schematic and not to scale, and in particular, some dimensions are for clarity of presentation and are shown exaggerated in the drawings. Similarly, although the views in the drawings for ease of description generally show similar orientations, this depiction in the figs. In general, the invention can be operated in any direction.

Dynamic constellation adaptation of limiter

Embodiments of the present disclosure provide a threshold adaptation mechanism that determines an optimally adapted threshold for slicer selection at a receiver for constellation selection based on a statistical distribution of data symbols across multiple constellations in a modulation scheme. More specifically, to determine the best-fit threshold, threshold-fit logic coupled to or included in the slicer is configured to have a desired ratio of received symbols having values within a range (e.g., below the best threshold that is offset from the nominal threshold due to non-linearities). The optimal threshold will be found to be the adaptive threshold. The desired ratio may be determined based on a desired statistical distribution of data symbols across multiple constellations. The first ratio and the second ratio are defined based on the desired ratio, the first ratio being equal to the desired ratio minus the error ratio, and the second ratio being equal to the desired ratio plus the error ratio. The threshold adaptation logic then determines a first value that can bring the received symbol within a first range (e.g., below the first value) to form a first ratio of a set of slicer inputs and a second value that can bring the received symbol within a second range (e.g., below the second value) to form a second ratio of a set of slicer outputs. An adapted threshold value is then obtained based on the first and second values.

Embodiments of the present disclosure also provide a constellation adaptation mechanism that determines an adapted constellation of slicers at a receiver based on a statistical distribution of data symbols across multiple constellations. More specifically, to determine an optimal constellation that deviates from a nominal constellation, constellation adaptation logic is provided with a desired ratio of received symbols that is within a certain range (e.g., below an optimal constellation value that has been shifted from the nominal value due to non-linearities). The optimal constellation will be found to be the adapted constellation. The desired ratio may be determined based on a desired statistical distribution of slicer inputs across multiple constellations. The constellation adaptation logic compares the first value to a set of received symbols and maintains a count of symbols within a particular range defined by the first value (e.g., below the first value) based on the comparison. A count ratio of the number of inputs in the first range to the total number of inputs of the group is also derived. The first value is adjusted until the ratio reaches the desired ratio. The first value is then designated as the best-fit constellation to be used by the slicer.

Embodiments herein are described in detail with reference to a 4-stage PAM limiter. However, it will be appreciated that the present disclosure is not limited to any particular modulation scheme or any particular number of constellations in a modulation scheme. The constellation adaptation and threshold adaptation mechanisms provided herein may be used in any suitable modulation device other than a slicer.

Statistically, the data symbols modulated and transmitted at the transmitter are distributed substantially in a number of constellations at some known percentage or ratio. For example, due to modulation, a large number of symbols are substantially uniformly distributed over the N scaled constellations. During signal transmission, the received symbols will change in an unpredictable manner, as non-linearities and amplitude compression may result in different noise levels and different constellation offsets for the constellation. But despite the impairment of the signal during transmission, the distribution of the aggregate symbols relative to multiple constellations is still expected to occur at the receiver side. In particular, for a large number of received symbols at the receiver, symbols falling within a particular data range defined by the actual constellation or actual threshold are expected to constitute a known ratio of total received symbols, irrespective of signal nonlinearity or amplitude compression.

Fig. 1 illustrates an exemplary data range for determining a best-fit threshold for constellation selection based on a desired statistical distribution of data symbols across multiple constellations in accordance with an embodiment of the present disclosure. Constellation selection may be performed at a slicer. In this example, the data symbols are modulated and demodulated according to the PAM-4 scheme, with nominal constellations C (1) -C (4) of [ -3, -1, +1, +3] respectively. Thus, the slicer uses 3 thresholds TH (1) to TH (3) to decide which constellation to assign to the received symbols, the nominal thresholds being [ -2,0, +2], respectively.

As specified by the modulation process at the transmitter side, the data symbols are uniformly distributed over 4 constellations. Thus, data symbols below the best-fit TH (1) should all be associated with constellation C (l) and expected to constitute 25% of the total received symbols. Data symbols falling between the best fit TH (1) and the best fit TH (2) should be associated with constellation C (2), and thus data symbols falling below the best fit TH (2) are expected to constitute 50% of the total number of received symbols. Likewise, data symbols falling below the best fit TH (3) should constitute 75% of the total received symbols.

To find the best TH (i) (i=l, 2 and 3), two values TH (i) 1 and TH (i) 2 are first determined according to the following definition: (1) Symbols equal to or less than TH (i) l occupy all the received symbols(Or expressed as) ; (2) Symbols having a TH (i) 2 or less are occupied in all received symbols(Or expressed as). In this case, N is equal to 4, andIs substantially smaller thanFor example, p=5 or less.

As shown, with respect to TH (l), TH (l) 1 is defined as a value below which the sign of (25-P)% is below, and TH (l) 2 is defined as a value below which the sign of (25+P)%. For TH (2), TH (2) 1 is defined as a value below which the sign of (50-P)% is below, and TH (2) 2 is defined as a value below which the sign of (50+P)% is below. For TH (3), TH (3) 1 is defined as a value below which the sign of (75-P)% is below, and TH (3) 2 is defined as a value below which the sign of (75+P)%. TH (i) can then be derived as TH (i) =average (TH (i) 1, TH (i) 2). That is, TH (1) =average value (TH (1) 1, TH (1) 2); TH (2) =average (TH (2) 1, TH (2) 2); and TH (3) =average value (TH (3) 1, TH (3) 2).

It will be appreciated that the present disclosure is not limited to any particular data range and corresponding desired ratio selected for determining the adaptation threshold. For example, in some embodiments, the best-fit threshold TH (i) may be determined based on a desired ratio of symbols above the first and second values TH (i) 1 and TH (i) 2. In some other embodiments, more than two data ranges may be defined and used to generate the best-fit threshold. In still other embodiments, a single data range may be used to find the best-fit threshold. For example, 50% of the symbols are expected to be below the best-fit threshold TH (2). If the difference between the two thresholds is too large, the error rate may be set to a lower value. In some embodiments, different P values may be used to find different thresholds.

Fig. 2 shows a configuration of an exemplary threshold adaptation unit 200 according to an embodiment of the present disclosure, which threshold adaptation unit 200 is capable of dynamically adapting a constellation selection threshold used in a slicer 210. Slicer 210 includes decision circuitry configured to map each input to one of a plurality of constellations by comparing the input to a constellation threshold. For example, each input is sampled digital data output from the equalizer. The limiter 210 may be implemented in any manner known in the art. The slicer inputs are also fed to a threshold adaptation unit 200, which threshold adaptation unit 200 uses a comparator 201 to compare each input with a "TH" value stored for example in a register. In the example shown in fig. 1, the TH value is TH (i) 1 or TH (i) 2. The threshold adaptation unit 200 comprises a counter 1205 for keeping a count of all inputs to be compared with the TH value. A counter 2204 is coupled to the output of the comparator 201 and counts inputs falling within a specific value range defined by TH based on the comparison result.

Using the example shown in fig. 1 to find the best-fit threshold TH (1), the TH register 202 is first set to the initial TH (1) 1 value. The total number of slicer inputs is programmed to be 2 ^M, which can be tracked by the counter 1205. Each time the comparator outputs a result indicating that the current input r is less than TH, the counter 2204 is incremented by 1. For the 2 ^M input, the TH adjustment logic 203 is calculatedA ratio of inputs below TH value is determined. If the ratio is not equal to the desired ratio (25-P)%, the TH adjustment logic 203 adjusts the TH value accordingly. Another 2 ^M input window is then compared to the new TH value byAn input ratio below the new TH is obtained. This process is repeated for a number of 2 ^M input windows until counter 1, finding TH values such thatThis TH value is then designated TH (l) l. In the same way, by varying the TH value to obtainTH (l) 2 can be determined. The TH adjustment logic 203 may then generate the best adapted TH (l) by averaging the determined TH (l) 1 and TH (l) 2.

The threshold adaptation unit 200 may be used to sequentially determine TH (l) 1, TH (l) 2, TH (2) 1, TH (2) 2, TH (3) 1, TH (3) 2, and the like. In some other embodiments, the threshold adaptation unit 200 comprises a repeating circuit as shown in fig. 2, which may generate different TH (i) X (x=1 or 2) values in parallel. The resulting adapted threshold is provided to a slicer to make constellation selection decisions for subsequent inputs accordingly. The threshold adaptation process may be performed periodically and on an as needed basis.

The TH adjustment logic 203 may be implemented in software, firmware, hardware, or a combination thereof. Counters 204 and 205 may have a resolution of, for example, 32 bits. For an error rate of 10 ^-6, the number of sample inputs for one symbol generated under the threshold is 4×10 ⁶. For 100 samples below the threshold, the total number of samples is 100×4×10 ⁶, corresponding to 29 bits.

According to embodiments of the present disclosure, a set of best-fit thresholds for constellation selection may be dynamically determined and adapted to statistically signal non-linearities and amplitude compression. As a result, data demodulation and data recovery can be advantageously performed at the receiver with reduced error rates. Furthermore, threshold adaptation does not involve computationally intensive computations and can be implemented using simple circuitry, including comparators and counters, for example. Therefore, design and development costs and operation power consumption can be advantageously reduced as compared with the conventional method.

Fig. 3A is a flow chart depicting an exemplary process 300 of determining an adapted threshold for constellation selection in accordance with an embodiment of the present disclosure. For example, process 300 may be performed by a threshold adaptation unit coupled to or included in a slicer as shown in fig. 2. In 301, an index of a threshold i is initialized, where i < N, and N represents the number of constellations defined in the modulation scheme. In particular, for PAM-4, N is 4. In 302, i is increased to 1. In 303, a first value TH (i) 1 associated with TH (i) is determined. TH (i) 1 defines a particular data range (e.g., < TH (i) l), and is associated with a first predetermined count ratio (e.g.,) And (5) associating. The optimal TH (i) 1 is defined such that a value of which the input falls within the data range satisfies a first preset count ratio. For example, for i=3, the data range associated with the first value TH (3) 1 includes any value less than TH (3) 1, and the preset count ratio of the range is (75-P)%.

In 304, a second value TH (i) 2 associated with TH (i) is determined. TH (i) 2 defines another data range (e.g., < TH (i) 2) and is proportional to another preset count (e.g.) And (5) associating. The optimal TH (i) 2 is defined as a value at which the input falls within the data range satisfies the second preset count ratio. For example, for i=3, the data range associated with the first value TH (3) 2 includes any value of TH (3) 2 or less, and the preset count ratio within this range is (75+p)%. In 305, the average value of TH (i) 1 and TH (i) 2 is assigned as the best adapted TH (i), e.g., TH (i) =average value (TH (i) 1, TH (i) 2). The processes 302-305 are repeated to determine each optimal threshold.

Fig. 3B is a flowchart depicting an exemplary process of determining a first value TH (i) X (x=1 or 2) associated with a threshold TH (i), according to an embodiment of the present disclosure. At 351, multiple inputs of the slicer are accessed. For example, the total number of inputs used per TH (i) X change may be set to a window of 2 ^M inputs. At 352, each of the plurality of inputs is compared to a current TH (i) X value (set in TH register 202 in FIG. 2). In 353, a count ratio of the number of inputs (counter 2)) lower than TH (i) X to the total number of inputs (counter 1)) to be compared is determined. At 354, a determination is madeWhether or not to equal a preset count ratio associated with TH (i) X. If not, then the TH (i) X value is adjusted in 355, and processes 351-354 are repeated for the adjusted TH (i) X value. Processes 351-355 are repeated untilEqual to a preset count ratio. At 356, the final TH (i) X is assigned as the optimal TH (i) X for deriving TH (i).

It will be appreciated that the present disclosure is applicable to a variety of other suitable modulation schemes. For example, for Quadrature Amplitude Modulation (QAM) with 2N constellations, the same threshold adaptation process shown in fig. 3A and 3B may also be performed on a single constellation dimension to find the best adapted threshold and adapted constellation. The data ranges and associated preset count ratios shown in fig. 1 may also be applied to a single constellation dimension in a QAM-2N scheme.

According to another aspect of the disclosure, the constellation level may be dynamically adjusted based on a statistical distribution of the received symbols relative to the plurality of constellations. In general, inputs that are within a particular data range and mapped to a particular constellation are expected to constitute a particular ratio (desired ratio) of the total inputs indicated by the statistical distribution. Fig. 4 illustrates an exemplary data range for determining a best-fit constellation based on a desired statistical distribution of data symbols across multiple constellations in accordance with an embodiment of the present disclosure. In this example, the data symbols are modulated and demodulated according to the PAM-4 scheme, with nominal constellations C (1) -C (4) of [ -3, -1, +1, +3] respectively. Therefore, the slicer uses 3 thresholds TH (1) to TH (3) to decide which constellation is allocated to the received symbol, the nominal thresholds are [ -2,0, +2], respectively.

As specified by the modulation process at the transmitter side, the data symbols are uniformly distributed over 4 constellations. Thus, the input of the limiter drops below the optimally adapted C (1) and is expected to constitute 12.5% (=0+12.5%) of the total input. Inputs of C (2) below the optimal adaptation are expected to constitute 37.5% of the total input (=25++12.5%). Likewise, inputs falling below C (3) of the best fit should constitute 62.5% (=50% +12.5%) of the total inputs, and data symbols falling below C (4) of the best fit should constitute 87.5% (=75% +12.5%) of the total inputs. This is equivalent to the median of the points belonging to one constellation. Where no assumption of gaussian noise exists.

It will be appreciated that the present disclosure is not limited to any particular data range and corresponding desired ratio selected for determining the best fit constellation. For example, in some embodiments, the best-fit constellation may be determined based on a desired ratio of symbols above a constellation level. In some other embodiments, more than one data range may be defined and used to generate a best-fit constellation in a similar manner to the threshold adaptation process described above.

Fig. 5 illustrates a configuration of an exemplary constellation adaptation unit 500 capable of dynamically determining a best-fit constellation according to an embodiment of the present disclosure. Slicer 510 includes decision circuitry configured to map each input to one of a plurality of constellations by comparing the input to a constellation threshold. The limiter 510 may be implemented in any manner known in the art. The slicer inputs are also fed to a constellation adaptation unit 500, which constellation adaptation unit 500 uses a comparator 501 to compare each input with a "C (j)" value stored for example in a register 502. The constellation adaptation unit 500 comprises a counter 1505 for counting all inputs used in the constellation adaptation process. A counter 2504 is coupled to the output of the comparator 501 and counts the inputs falling within a certain value range defined by C (j) according to the comparison result.

Using the example shown in fig. 4 to find the best adapted threshold C (l), the C (j) register is first set to an initial value. The total number of slicer inputs is programmed to 2 ^M, which can be tracked by counter 1505. Each time the comparator outputs a result that the input r is less than C (j), the counter 2204 is incremented by 1. For each window of 2 ^M inputs, the C (j) adjust logic 503 calculatesThe ratio of inputs below C (j) is determined. If the ratio is not equal to the desired ratio of 12.5%, C (j) adjustment logic 503 adjusts the C (j) value accordingly. Then another window of 2 ^M inputs will be compared with the new C (j) value to base onThe ratio of inputs below the value of C (j) in register 502 is obtained. This process is repeated for multiple windows of 2 ^M inputs until a single window is generatedC (j) of (2). This C (j) value is then designated as the best fit C (l). Also, by changing the value of C (j)C (2) can be determined. The same procedure was performed to generate adapted C (3) and C (4) by using preset count ratios of 62.5% and 87.5%, respectively.

The constellation adaptation unit 500 may be used to sequentially determine C (1) to C (4). The constellation adaptation unit 500 may also be used to perform a threshold adaptation procedure as described with reference to fig. 1-3B. In some other embodiments, constellation adaptation unit 500 includes a repeating circuit as shown in fig. 5 that may generate different levels of adapted constellations in parallel. The adjusted constellation is provided to a slicer to make constellation selection decisions for subsequent inputs. The constellation adaptation process may be performed periodically and as needed.

Constellation adjustment logic 503 may be implemented in firmware, software, hardware, or a combination thereof. The counter may have a resolution of 32 bits, for example. For bit error rates of 10- ⁶, the number of sample inputs for one symbol generated under the threshold is 4×10 ⁶. For 100 samples below the threshold, the total number of samples is 100×4×10 ⁶, corresponding to 29 bits.

In some embodiments, to prevent interaction with the equalization gain at the receiver, the gain of the 4 best constellations may be kept at a fixed value. For example, the number of the cells to be processed,Wherein constant=sum (abs [ -3, -1, +1, +3 ]).

Fig. 6 is a flow chart depicting an exemplary process 600 of determining a best-fit constellation for demodulation in accordance with an embodiment of the present disclosure. For example, process 600 may be performed by a constellation adaptation unit coupled to a slicer as shown in fig. 5. In 601, an index of a constellation j is initialized, where j+.N, and N represents the number of constellations defined in the modulation scheme. In 602, j is incremented by 1. In 603, multiple inputs of the slicer are accessed. For example, the total number of inputs used per C (j) change may be preset to a window of 2 ^M inputs.

At 604, each of the plurality of inputs is compared to a current C (j) value. In 605, the total number of inputs to be compared is determined (counter 1). At 606, the number of inputs (counter 2) below the current C (j) value is determined. At 606, a determination is madeWhether or not the preset count ratio related to C (j) is equal to the preset count ratio ofIf not, the C (j) value is adjusted at 608, and the foregoing 602-607 is repeated for the adjusted C (j) value. Repeating the foregoing 602-608 untilEqual to a preset count ratioIn 609, the final C (j) value is assigned as the best-fit constellation C (j), which is provided for use by the notch.

Fig. 7 shows a network signal transmission system 700 according to an embodiment of the present disclosure, the network signal transmission system 700 comprising constellation adaptation logic 739 and constellation selection threshold adaptation logic 738 at a receiver 730. In simplified form, system 700 includes an optical transmitter 710, an optical cable 740, an optical receiver 720, and an electrical receiver 730. The optical transmitter 710 has a driver 711 and a Mach-Zehnder interferometer (MZI) 712 and operates to receive data modulated according to PAM-4. The modulated data is transmitted for transmission over the fiber optic cable 740. The optical receiver 720 has a Photodetector (PD) 721 and a transimpedance amplifier (TIA) 722, and is configured to receive data from the fiber optic cable 740. The electrical receiver 730 receives signals from the optical receiver 720 and performs data and clock recovery. The electrical receiver 730 includes an analog-to-digital converter (ADC) 731, an equalizer 734, a slicer 735, a Timing Recovery (TR) system 737, and a phase interpolator 736. Slicer 735 is coupled to or integrated with threshold adaptation logic 738 and constellation adaptation logic 739, which threshold adaptation logic 738 and constellation adaptation logic 739 may generate optimally adapted thresholds and optimally adapted constellations that vary over time due to non-linearities or other signal distortions. The threshold adaptation logic 738 and constellation adaptation logic 739 may be used in any other suitable signal transmission and processing system. In some embodiments, a single circuit as shown in fig. 2 or 5 may be implemented to perform both threshold adaptation and constellation adaptation using different configurations of data ranges, TH/C (j) values, and desired count ratios.

Fig. 8 illustrates a set of simulation results comparing Symbol Error Rates (SER) generated using default slicer thresholds and best-fit threshold generated SER obtained according to an embodiment of the present disclosure. The data presented in diagrams 810, 820, and 830 are obtained using different input signal-to-noise ratios (SNRs) for the constellation in PAM-4. In each figure, the SNR is the same for all 4 constellations, 16dB in 810, 18dB in 820, and 20dB in 830. Each data plot shows SER variation as a function of constellation shift number 2 by using default slicer thresholds (811, 821 and 831) and using best-fit thresholds (812, 822 and 832). Comparison of the two curves in each figure shows that SER can be significantly reduced using the optimally adapted threshold and that the amount of SER reduction increases with the offset of the constellation.

Fig. 9 shows another set of simulation results comparing a Symbol Error Rate (SER) generated by a default absolute threshold to a SER generated by an optimal adaptive threshold according to an embodiment of the present disclosure. The data shown in graphs 910 and 920 are obtained using different sets of input SNR for the constellation in PAM-4. For graph 910, the input SNR for the four constellations is [18, 18, 18, 14] db. For graph 920, the input SNR for these four constellations is [20, 16] dB. Each data plot shows SER variation as a function of constellation offset number 2 using default slicer thresholds (911 and 921) and using best-fit thresholds (912 and 922). Comparing the two curves in each graph, it can be demonstrated that SER can be significantly reduced using the optimal adaptation threshold, and that the SER reduction effect increases with increasing constellation offset.

Claims

1. A method of dynamically determining a threshold of a slicer, the method comprising:

accessing a first plurality of inputs provided to the slicer;

comparing each of the first plurality of inputs to a first value;

Determining a first count ratio of a number of first inputs included in the first plurality of inputs to a total number of the first plurality of inputs based on the comparing with the first values, wherein each of the first inputs has a value included within a first range defined by the first values; and

A threshold value of the slicer is determined based on the first value and the first count ratio.

2. The method of claim 1, further comprising:

accessing a second plurality of inputs provided to the slicer;

comparing each of the second plurality of inputs to a second value; and

Determining a second count ratio of a number of second inputs to a total number of the second plurality of inputs based on the comparing with the second value, wherein each of the second inputs has a value included in a second range defined by the second value, and

Wherein said determining said threshold value of said slicer is further based on said second value and said second count ratio.

3. The method of claim 2, wherein the determining the threshold comprises averaging the first value and the second value.

4. The method of claim 2, further comprising: determining the first value and the second value, wherein the determining the first value comprises: adjusting the first value until the first count rate is equal to a calibration rate minus a preset value, and wherein the determining the second value comprises: and adjusting the second value until the second count rate is equal to the calibration rate plus the preset value.

5. The method of claim 4, wherein the limiter is configured to use Pulse Amplitude Modulation (PAM) comprising N constellation levels, where N is an integer greater than 1, wherein the scaling ratio is equal toWherein i is an integer greater than 0 and less than N, and wherein the preset value is equal to or less than 5%.

6. The method of claim 4, wherein the slicer is configured to use Quadrature Amplitude Modulation (QAM) including 2x N constellation levels, where N is an integer greater than 1, wherein the scaling ratio is equal toWherein i is an integer greater than 0 and less than N, and wherein the preset value is equal to or less than 5%.

7. The method of claim 2, wherein the first plurality of inputs and the second plurality of inputs are outputs from an equalizer in a receiver.

8. The method of claim 2, wherein the first range includes any value below the first value, and wherein the second range includes any value below the second value.

9. A receiver, comprising:

a limiter configured to:

generating a first output value in response to an input greater than a threshold; and

Generating a second output value in response to an input less than the threshold; and

A threshold adaptation unit coupled to the slicer, the threshold adaptation unit comprising:

a comparator configured to compare each of a first plurality of inputs to a first value, wherein the first plurality of inputs are provided to the limiter; and

A first counter configured to generate a number of first inputs included in the first plurality of inputs in response to a comparison decision output from the comparator, and each of the first inputs having a value contained within a first range, wherein the first range is defined by the first value;

Wherein the threshold adaptation unit is configured to:

Determining a first count ratio of the number of first inputs to a total number of the first plurality of inputs; and

An adapted threshold value of the slicer is determined based on the first value and the first count ratio.

10. The receiver of claim 9, wherein the threshold adaptation unit is further configured to: accessing a second plurality of inputs provided to the slicer; the comparator compares each of the second plurality of inputs to a second value; and wherein the threshold adaptation unit further comprises a second counter configured to: determining a second count ratio of a number of second inputs included in the second plurality of inputs to a total number of the second plurality of inputs based on the comparison with the second values, wherein each of the second inputs has a value included within a second range defined by the second values,

The adapted threshold value is also determined based on the second value and the second count ratio.

11. The receiver of claim 10, wherein the threshold adaptation unit is further configured to: the first value and the second value are averaged to generate the adapted threshold value for the limiter.

12. The receiver of claim 10, wherein the threshold adaptation unit is further configured to:

Determining the first value by changing the first value until the first count rate equals a calibration rate minus a preset value; and

Determining the second value by changing the second value until the second count rate is equal to the calibration rate plus the preset value.

13. The receiver of claim 12, wherein the limiter is further configured to: using Pulse Amplitude Modulation (PAM) comprising N constellation levels, where N is an integer greater than 1, wherein the first output value and the second output value are selected from the N constellation levels, and wherein the scaling ratio is equal toWherein i is an integer greater than 0 and less than N, and wherein the preset value is equal to or less than 5%.

14. The receiver of claim 12, wherein the limiter is further configured to: using Quadrature Amplitude Modulation (QAM) comprising 2 XN constellation levels, where N is an integer greater than 1, wherein the scaling ratio is equal toWherein i is an integer greater than 0 and less than N, and wherein the preset value is equal to or less than 5%.

15. The receiver of claim 10, wherein the first range includes any value below the first value, and wherein the second range includes any value below the second value.

16. A receiver, comprising:

An analog-to-digital converter ADC configured to convert a received analog signal into a digital signal;

An equalizer coupled to the ADC and configured to: generating an equalized signal in response to the digital signal; and sending the equalized signal as an input to a limiter;

the slicer is coupled to the equalizer and configured to:

A threshold adaptation unit coupled to the slicer and configured to:

accessing a first plurality of inputs provided to the slicer;

comparing each of the first plurality of inputs to a first value;

Determining a first count ratio of a number of first inputs included in the first plurality of inputs to a total number of the first plurality of inputs based on the comparison with the first values, wherein each of the first inputs has a value that is included within a first range defined by the first values;

accessing a second plurality of inputs provided to the slicer,

Comparing each of the second plurality of inputs to a second value;

determining a second count ratio comprising a number of second inputs to a total number of the second plurality of inputs based on the comparison with the second values, wherein each of the second inputs has a value that is included within a second range defined by the second values;

Determining an adapted threshold based on the first value, the second value, the first count rate, and the second count rate; and

And sending the adapted threshold value to the limiter.

17. The receiver of claim 16, wherein the threshold adaptation unit is further configured to determine the adapted threshold by averaging the first value and the second value.

18. The receiver of claim 16, wherein the threshold adaptation unit is further configured to:

determining the first value by adjusting the first value until the first count rate equals the calibration rate minus a preset value, and

Determining the second value by adjusting the second value until the second count rate is equal to the calibration rate plus the preset value.

19. The receiver of claim 18, wherein the limiter is configured to: using one of Pulse Amplitude Modulation (PAM) comprising N constellation levels and Quadrature Amplitude Modulation (QAM) comprising 2 x N constellation levels, where N is an integer greater than 1, wherein the scaling ratio is equal toWherein i is an integer greater than 0 and less than N, and wherein the preset value is equal to or less than 5%.

20. The receiver of claim 16, wherein the first range includes any value lower than the first value, and wherein the second range includes any value lower than the second value.