Test and measurement device and input signal analysis method

Abstract

To more accurately and more quickly identify and measure jitter from a signal under test.SOLUTION: A test measurement system 300 analyzes signals using machine learning. The test measurement system 300 obtains a recovered clock signal on the basis of a digital signal, sets window positions for a fast Fourier transform (FFT) of the digital signal, windows the digital signal into a series of windowed waveform data on the basis of the window positions, transforms each of the windowed waveform data into frequency-domain windowed waveform data using FFT, and obtains high-order spectrum data of each of frequency-domain windowed waveform data. The test measurement system 300 includes a neural network 318 configured to receive the high-order spectrum data of the frequency-domain windowed transform waveform data and classify each windowed waveform data on the basis of the high-order spectrum data.SELECTED DRAWING: Figure 3

Description

[Technical field]

[0001] FIELD OF THE DISCLOSURE The present disclosure relates to systems and methods relating to test and measurement systems, and more particularly to test and measurement systems for signal analysis. [Background technology]

[0002] Jitter is a well-known technical term used to define the deviation of events in a signal from their ideal timing. Jitter results in important edges in a sequence of data bits being misplaced from their ideal positions. This jitter can result from errors in the edge positions of the recovered clock, from reflections in the system or from inter-symbol interference in the system, or as a result of the transfer function of the system. Other effects such as distortion can also occur. In addition, there are other important waveform measurements that help characterize the accuracy and performance of the data recovery of a channel. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] JP 2018-59912 A [Patent Document 2] JP 2018-159702 A Summary of the Invention [Problem to be solved by the invention]

[0004] In modern serial data communication systems, the serial data clock is usually not transmitted separately from the data, so jitter can cause data errors at the receiver, making it very important to characterize and classify the various measurements of waveform parameters and system characteristics.

[0005] There is a continuing need for improvements in test and measurement equipment to more accurately and quickly identify and measure jitter in signals under test.

[0006] Embodiments of the disclosed technology address these and other deficiencies of the prior art. [Means for solving the problem]

[0007] Disclosed herein is an example test and measurement system that can analyze an input signal, such as a pseudorandom binary sequence (PRBS) signal, by applying bicoherence and associated clock recovery based on calculations such as short-term fast Fourier transform (FFT) calculations to analyze the signal. In an example of the disclosed technology, bicoherence is utilized to identify jitter and separate the jitter from the signal including other distortions such as inter-symbol interference (ISI), clock edge jitter, and other distortions. The resulting vector space of bicoherence may be fed into a neural network, which may be trained to recognize and classify various characteristics of the signal.

[0008] Aspects, features and advantages of embodiments of the present invention will become apparent from the following description of the embodiments, taken in conjunction with the accompanying drawings. [Brief description of the drawings]

[0009] [Figure 1] FIG. 1 is an example of a graph of absolute phase reference positions for bispectrum calculations. [Diagram 2] Figure 2 is a bandwidth diagram showing a time vs. frequency plot of an ideal pseudorandom binary sequence signal. [Diagram 3] FIG. 3 is a block diagram of an exemplary test and measurement system in accordance with aspects of the disclosed techniques. [Figure 4] FIG. 4 is a block diagram of another exemplary test and measurement system in accordance with aspects of the disclosed techniques. [Diagram 5]FIG. 5 is a block diagram of another exemplary test and measurement system in accordance with aspects of the disclosed techniques. [Figure 6] FIG. 6 is a block diagram of another exemplary test and measurement system in accordance with aspects of the disclosed techniques. [Figure 7] Figure 7 is a timing diagram showing the recovered clock signal and the window function of the input signal. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0010] To understand bicoherence, we provide the definitions of first and second order cumulants. The first order cumulant of a stationary process is the average. The first order cumulant can be expressed as Equation 1, where E is the expectation, which is the average value of the signal.

[0011] C 1x =E{x(t)} (1)

[0012] For higher order cumulants, this average value is usually subtracted.

[0013] The second order cumulant in the time domain is the autocorrelation function and can be written as shown in Equation 2, where the overline represents the complex conjugate.

[0014]

number

[0015] Auto-correlation is similar to filter convolution as shown in Equation 2. In filter convolution, the time order of the filter coefficients is reversed before performing the multiplication, accumulation, and shifting. In auto-correlation, the order of x(n) and x(n+k) with a shift of k is not reversed. When the second order cumulant is transformed into the frequency domain, it is called the auto power spectrum and is shown in Equation 3. Multiplying x by its complex conjugate results in zero phase. The square of the x(n) term represents the power. Thus, the power spectrum represents the second order cumulant in the frequency domain and can be written as shown in Equation 3.

[0016]

number

[0017] An alternative representation of Equation 3 is shown in Equation 4, where X is the spectrum of x.

[0018]

number

[0019] The third-order cumulant transformed into three-dimensional space is called the bispectrum. The third-order cumulant in the time domain is shown in Equation 5.

[0020]

number

[0021] The bispectrum is defined as the Fourier transform of the third order cumulant, as shown in Equation 6.

[0022]

number

[0023] An alternative notation for the bispectrum in the frequency domain is shown in Equation 7 as a triple product.

[0024]

number

[0025] The variable B is the bispectrum at frequencies f1 and f2, and X(f) is the Fourier transform of the signal x(t). The overline indicates complex conjugation. Using the rules of complex multiplication, the magnitude of the bispectrum is equal to the product of the magnitudes of the function X at each frequency. The phase of the bispectrum is the sum of the phases of X at each frequency. The bispectrum depends on the magnitude of X at each frequency.

[0026] If X is a continuous, time-invariant signal, and if X(f1), X(f2) and X(f1+f2) are perfectly phase-locked, and if the short-term FFTs are computed at several random time points in x(t), the bispectrum will always be the same. If summed, the bispectrums of each point sum without cancellation. The phase of each short-term FFT is adjusted to an absolute reference time on the waveform.

[0027] However, if X is a time-varying signal, i.e. if the phase of each frequency is random in time and the magnitude is constant, then when adding up the bispectrums, there will be a tendency for cancellation. The result will tend towards zero depending on the phase and magnitude relationship of the time-varying signal. Therefore, for time-varying signals such as pseudorandom binary sequence signals, clock recovery is performed so that the phase of the signal does not average to zero.

[0028] Bicoherence normalizes the bispectrum to remove the dependence on magnitude, as shown in Equation 8.

[0029]

number

[0030] where n is the index of each FFT frame and varies from 1 to the number of FFT frames in the waveform record. The numerator of Equation 8 contains the magnitude of the spectrum. If phase coupling is high, the magnitude value will be large, but if the coupling is low it will approach zero. In the denominator of Equation 8, we are only summing the magnitudes, essentially setting all the phases to zero, and then normalizing the bispectrum to get the bicoherence, b, which has a maximum value of 1 and a minimum value of zero.

[0031] Adding two sine waves of different frequencies and passing them through a nonlinear element creates a multiplication process. The resulting output consists of sine waves of two frequencies, which are the sum and the difference of the two frequencies, as shown in Equation 9.

[0032] sin(a)·sin(b)={cos(ab)-cos(a+b)} / 2 (9)

[0033] Therefore, in the bispectrum equation shown in Equation 7, X(f1) and X(f2) may be two sinusoids in phase, and X(f1+f2) may be the distortion component after passing through a nonlinear device. This bispectrum is commonly used to analyze nonlinear distortion components.

[0034] If X is a time-invariant signal composed of coherent frequencies, the bispectrum may be calculated by taking multiple short-term FFTs along the time record of x(t) and then adding the bispectrum of each FFT. The zero-phase reference point of the FFT is usually defined as the time position of the first sample in the time window of the FFT. As the FFT window moves along the time record, the phase of X(f1) and X(f2) rotates as a function of time position. Thus, the bispectrum from an FFT based on that phase reference will cause the sum of the bispectrum from all FFT positions to approach zero.

[0035] Therefore, an absolute phase reference position is needed on the x(t) waveform. Then, the phase of each FFT result can be adjusted to the absolute time position as shown in FIG. 1. In FIG. 1, the phase of the sine wave 100 is 90 degrees relative to the absolute zero phase reference point 102. However, the phase calculated from FFT1 is 0.0 degrees and the phase calculated from FFT2 is 180 degrees. So, the phase value of FFT1 is corrected by offset 1 so that FFT1 has a zero phase reference 104. The phase value of FFT2 is corrected by offset 2 so that FFT2 has a zero phase reference 106. This allows FFT1 and FFT2 to return the 90 degree phase of the signal relative to the absolute zero phase time position 102. This allows the bispectrum value to be calculated more accurately by averaging multiple time windows.

[0036] A pseudorandom binary sequence (PRBS) signal has a pseudorandom bit pattern whose length is determined by the degree of the polynomial used to generate the PRBS signal. Thus, the signal is not time invariant and its bispectrum tends towards zero.

[0037] Consider an ideal PRBS signal consisting of a series of ideal step signals spaced at multiples of the unit interval (UI). A time vs. frequency plot 200 for the derivative of x(t) with respect to time t is shown in Figure 2. Because this is an idealized spectrogram, the frequency content of each pulse is the same. Figure 2 shows an idealized spectrogram that cannot be calculated in reality from a short-term FFT due to the tradeoff in resolution between time and frequency, but is shown for illustration purposes. For example, to capture only one impulse in one FFT window, the FFT window would have a record length much shorter than one bit interval. The time window width of that record length would not be long enough to accurately measure the low frequency content of the signal. To obtain a higher resolution for determining the effects of jitter and ISI, oversampling would be desirable.

[0038] As the FFT window becomes wider, it begins to include more than one bit interval of the signal. For an ideal impulse at each time location where a bit transition occurs, the magnitude of its spectrum is 1 at all frequencies and the phase of all frequencies is equal to zero with respect to the time location of the impulse for each individual transition.

[0039] However, a wider FFT window will contain multiple frequencies resulting from each impulse within the window whose phase varies with the delay between edges. These are summed during the FFT operation to provide a single composite phase at each frequency. If there are only two impulses within the FFT window, one will be a rising edge and the other a falling edge, by definition, if it is a non-return-to-zero (NRZ) signal. A four-level Pulse Amplitude Modulated (PAM4) signal may have two consecutive rising edges or two consecutive falling edges, or one rising edge and one falling edge. For ideal edges, the phase of the rising edge is 180 degrees away from the phase of the falling edge.

[0040] If multiple edges are within a window, then the phase, jitter, ISI, etc. of each edge will add up. If the pattern and position of the edges differs from one window to the next, then the bispectral sum will average out the differences over a larger range, resulting in a smaller bicoherence value.

[0041] The current oscilloscope has a sampling rate of approximately 200 gigasamples per second (GS / s) to capture a 53 gigabaud (GBd) PAM4 signal with a symbol width of 1 / 26.5 gigahertz (GHz) over a period of 37.7 picoseconds (ps). The period of the sample rate is 5 ps, which results in 7.547 samples per bit interval. The length of the FFT, which contains four symbol periods, is approximately 32 points.

[0042] When the test and measurement equipment channel impulse response to a PRBS signal is convolved with the ideal bit pattern, the edges of past bits affect subsequent bits. This is called inter-symbol interference (ISI). ISI produces non-constant rise and fall times that depend on the preceding bit pattern. ISI affects the phase and magnitude of the spectrum due to the derivative of each edge. This causes jitter on the edges, which is often classified as data-dependent jitter (DDJ). Transmitter clock jitter from the device under test causes the vertical bars in the spectrogram in Figure 2 to become wider.

[0043] FIG. 3 illustrates an exemplary test and measurement system 300 in accordance with an example of the disclosed technique. A signal x(n) is received at an input of a test and measurement instrument. An optional continuous time linear equalizer (CTLE) 302 may be provided to partially de-embed the test instrument or the serial data link. Another optional filter (not shown) may be a de-embed / embed filter to de-embed the test instrument or the channel response. The de-embed / embed filter may be located next to the CTLE 302 (either before or after the CTLE 302).

[0044] The clock recovery 304 is implemented by one or more processors. It takes an input signal, either directly or an equalized input waveform from the CTLE 302, and generates a recovered clock waveform and a list of edge crossings. In some cases, a clock may be used directly (an explicit clock) obtained by an analog-to-digital converter, or a bursty explicit clock such as that available in double data rate (DDR) memory systems.

[0045] The input signal and the recovered clock are sent to a window sequencer 306. In some alternative embodiments, rather than sending the input signal to the window sequencer 306, the equalized signal from the CTLE 302 may be sent. The window sequencer 306 is implemented by one or more processors and determines the location of any short-term FFT windows relative to the recovered clock signal and the input waveform data. As will be described further below, the window sequencer 306 allows the bispectral or bicoherence functions to be used in a meaningful way to extract characteristics of the waveform.

[0046] In some examples, an optional resampler 308 may be provided. The resampler 308 resamples the signal during the FFT window interval so that the start of the FFT is at the center of the unit interval (UI). If the input signal is a PRBS sequence, the resampler 308 removes the effect of the sample rate being asynchronous with respect to the start time of the window. Thus, the resampler 308 removes analog-to-digital sample jitter; that is, after resampling, there is always a data sample exactly at the start of the window. In situations where the input signal is continuous, the resampler 308 can be used to guarantee the phase of the spectrum with respect to an absolute time reference point on the waveform.

[0047] Prior to performing an FFT 312 to convert the signal to the frequency domain, a window function 310, such as a Tukey window, is applied by one or more processors, as shown in Figure 3. The window function 310 can isolate a portion of the signal to perform an FFT 312. The FFT 312 is a short-term FFT performed by one or more processors and applied to the windowed waveform data at each time position determined by a window sequencer 306.

[0048] The phase reference points of the FFT 312 are used in calculating the higher order spectral data 314 for the signal. The recovered clock from the clock recovery 304 is used to place the phase reference points for each FFT. The location of the phase reference points based on the recovered clock allows the edges in the FFT window to be coherent with respect to the clock phase reference positions.

[0049] The high order spectral data 314 may be determined by one or more processors by taking the spectrum from the FFT 312 and converting the data into a bispectrum. The high order spectral data 314 may include a complex valued bispectrum, bispectral phase, or bicoherence. To determine the bicoherence, the bispectrum is normalized to determine a magnitude value. That is, the bispectral data 314 may be represented as magnitude values, phase values, or complex values ​​before providing the high order spectral data 314 to subsequent operations. Throughout this disclosure, the high order spectral data 314 may include the bispectrum, bicoherence, or bispectral phase. The high order spectral data 314 may also include higher order statistics, such as fourth order or higher, determined based on the spectrum.

[0050] A threshold gate 316 is provided in some examples to retain bispectral samples having a magnitude above a specified threshold level. The threshold level may be a magnitude threshold, a phase threshold, or a complex threshold. The threshold level may be set by a user or may be determined by one or more processors. Although not shown, the test and measurement instrument may include a user interface for receiving a user-specified threshold level.

[0051] If the threshold gate 316 is a magnitude threshold gate 316, as shown in Figure 3, the magnitude threshold gate 316 passes the bispectrum if its magnitude is above a threshold and sets the bispectrum to zero or other low level if its magnitude is below the threshold. Alternatively, the phase of the bispectrum may be gated based on a threshold on the magnitude of the bicoherence or the phase of the bispectrum. In some examples, the bispectrum may remain complex valued and a complex threshold may be used.

[0052] The output of the threshold gate 316 is sent to a neural network 318. The neural network 318 receives the bispectrum and / or bicoherence in at least one of the following forms: magnitude, phase, or complex. Through the neural network 318, machine learning is applied to the bispectrum and / or bicoherence to classify and decode the signal. That is, the neural network 318 processes the threshold-gated bispectrum or bicoherence to obtain a desired learned output response. The output of the neural network 318 may be a number of different outputs, such as, but not limited to, bit error rate, distortion, pattern decoding, jitter measurements, ISI measurements, signal to noise ratio, etc. Although a neural network is illustrated, examples of the disclosed technology are not limited to neural networks, and neural networks may refer to any form of machine learning.

[0053] 4 illustrates another example of a test and measurement instrument 400 according to an example of the disclosed technique. In this example, components similar to those in FIG. 3 are given the same reference numerals and will not be described further.

[0054] The test and measurement instrument 400 of FIG. 4 is similar to the test and measurement instrument 300 of FIG. 3, but includes a second threshold gate 402. The threshold of the second threshold gate 402 may be set either by a user or by one or more processors of the test and measurement instrument 400. The second threshold gate 402 receives the output from the FFT 312 and outputs only data that does not violate the threshold setting to the neural network 318. The neural network 318 can then classify the signal using both the output of the FFT and the output of the high order spectral data 314. The second threshold gate 402 may be one of a magnitude, phase, or complex threshold gate and may be of the same type as the first threshold gate 316.

[0055] Figure 5 shows yet another example of a test and measurement system 500. Components similar to Figures 3 and 4 are given the same reference numerals and will not be further described here. The inclusion of the FFT 312 and higher order spectral data 314 incorporates both second and third order statistics, respectively. In some examples, there may be the possibility to add even higher order statistics, either used independently or in combination with lower order statistics.

[0056] In some alternatives, a window of higher order spectral data 314 on the signal may be processed by taking the sum of multiple spectra corresponding to multiple similar patterns in each FFT 312 window and referencing this to the recovered clock. The bispectral windows of similar patterns are then summed. For example, if there are five symbol intervals in one FFT window, then 32 patterns and 32 different bicoherence resulting spectra are possible. The test and measurement instrument 500 provides multiple different neural networks 318, as described in more detail below.

[0057] In the test and measurement system 500, the input signal x(n) is received by both the CTLE 302 and a derivative 502, which may be performed by one or more processors in the test and measurement instrument 500 and provides a derivative of the input signal x(n). The output of the derivative 502 is also sent to the window sequencer 306.

[0058] As mentioned above, if we assume that there are 5 UIs in the FFT window, there will be 32 possible bit patterns. The test and measurement instrument 500 includes a neural network pattern classifier 504. The neural network pattern classifier 504 receives the output of the FFT 312 and classifies the patterns in the FFT window. The neural network pattern classifier 504 can identify the patterns of several bits in the window, even in the presence of noise and ISI.

[0059] The output pattern from the neural network pattern identification unit 504 is sent to a multiplexer switch 506. The multiplexer switch 506 can select one of a number of different paths to send the output of the FFT 312. In the above example, there are 32 different patterns, so in such an example, 32 different paths would be provided. The number of paths provided or used may correspond to the number of possible patterns.

[0060] In each path, the high order spectral data 314 is determined as described above. The output of the high order spectral data 314 is sent to the neural network 318 for classification. In some examples, an optional summing or averaging block 508 may be provided in the paths. The high order spectral data 314 tends to reduce noise towards zero and preserve the coherent components of the spectrum. However, jitter will cause changes in phase from pattern to pattern. The summing block 508 averages the output of each high order spectral data 314 that contains the same bit pattern as the previous bit pattern. This helps to average out and smooth out the jitter that appears in the bicoherence result. The neural network 318 can compare the averaged bicoherence to the instantaneous bicoherence output to measure the amount of jitter.

[0061] 5, each path may in some examples further include a threshold gate 316. The threshold gate may include either an average threshold for each of the instantaneous bicoherence outputs, or a magnitude, phase, or multiple thresholds.

[0062] The output of the neural net pattern classifier 504 may also be fed to a transfer function block 510, which uses one or more processors to determine the transfer function. The output of the FFT 312 is also fed to the transfer function block 510, which may determine the transfer function of the transmitter and the serial data channel. One way to do this would be to divide the spectrum of the ideal pattern waveform by the spectrum of the captured waveform and then perform an inverse FFT (IFFT) to obtain the impulse response. Other known methods may be used as well.

[0063] 6 shows another exemplary test and measurement system 600. Components similar to those in FIGS. 3-5 are given the same reference numerals and will not be further described here. The test and measurement system 600 may include a second multiplexer switch 602 having a number of output paths corresponding to a number of patterns.

[0064] Once a pattern is identified by the neural network pattern identification unit 504, a multiplexer switch 602 determines which path will receive the output from the FFT 312. Each path may have an optional summing block 508 that averages similar patterns in each window of the FFT 312. A transfer function block 510 may take the output from each summing block 508 in addition to the recovered clock signal from the clock recovery 304 and the original input signal x(n) to calculate the transfer function.

[0065] In the following, we describe how to compute the short-term FFT 312 of each segment of the input signal x(n) at multiple different windowed time positions along the time record. The first point of the FFT window is the time position that represents zero phase. To maintain a constant phase reference for the bicoherence calculation, we need to identify and separate the multiple bit patterns in the region gated by each FFT window. In this way, only sequences of the same bit pattern are summed together to average the bispectrum output. The averaged bispectrum is then normalized to obtain the bicoherence, as described above in Equation 8.

[0066] The width of the FFT 312 window is an integer number of bits of the input signal x(n) under analysis. The zero-phase reference point of the FFT 312 is the start time of the window. As shown in FIG. 2 above, the spectrum of the derivative of an ideal step signal at each bit transition has a constant amplitude at all frequencies from DC (i.e., zero) to its Nyquist frequency. The phase of all ideal impulses within the FFT 312 window is zero at all frequencies with the phase reference point at the time position of the impulse. However, the zero-phase reference point of each FFT 312 is at the beginning of the FFT 312, and the entire band of frequencies representing each edge will be in phase with the other edges within the FFT 312 window.

[0067] The result of the FFT has a single set of frequencies and phases resulting from the sum of the frequencies of all the edges. Each combination of bit patterns within the window interval is expected to have a unique set of phases and magnitudes within the spectrum, which allows each pattern to be classified as a different high-frequency Next This can be used in a neural network 318 to decode the bit sequence within the window for distribution and summation over paths of the spectral data 314.

[0068] Figure 7 shows the relationship between a recovered clock 702, an input signal x(n) 704, and a Tukey window function 706. In the illustration of Figure 7, the input signal x(n) 704 is a PRBS data signal. As shown in Figure 7, a unit interval (UI) is the width of one pulse of the recovered clock.

[0069] As shown in Figure 7, the Tukey window 706 is N sample bits long and begins and ends at the center of a clock cycle. The zero phase reference for each FFT is set at the beginning of each window. The FFT window is relatively short compared to the number of samples and UIs, which allows the bispectrum to be a reasonable size for analysis by the neural network 318.

[0070] The windows are an integer number of UIs, with the start of each FFT at the center of a UI, as shown in Figure 7. One way to address the windowing of the FFT 312 and the transients at the edges of the cyclic window is to calculate the derivative of the data, as described in several figures above, and then center the start of the FFT window in the center of the UI where the derivative is zero. Centering the start of the FFT window in the center of the UI where the derivative is zero avoids transient leakage at the beginning and end of the record.

[0071] Furthermore, by applying a Tukey window, both ends of the FFT record can be tapered to zero, so that the same patterns in a window are directed to a specific bispectral path, with the effect of different ISI being reduced by the preceding window. When using 5 bits for the window interval, each path accepts 1 of 32 possible bit patterns.

[0072] Taking the derivative can enhance high frequency noise, which may be undesirable in some instances. However, the bicoherence pass can take the distributed bit pattern and calculate the bicoherence, which helps to remove noise. Additionally, the pattern pass may average similar bit patterns to remove further noise and jitter fluctuations, as shown.

[0073] The resampler block 308 can remove sample clock jitter relative to the zero phase reference point of the FFT window, ensuring that the time domain samples of the waveform always occur at exactly the same time as the zero phase reference.

[0074] Examples of the disclosed techniques are not limited to a 5-bit window interval as described above, but may be smaller or larger depending on the sample rate of the test and measurement equipment. Edges within the window may be jittered due to signal clock jitter or delayed by random amounts, and combinations of these will be similar in average from summing block 508 in terms of one instantiation of the window. The best observation of jitter may be made by creating a smaller FFT window, incorporating fewer edges and a smaller UI. One edge per window would be ideal, but the sample rate may not be high enough to optimally support such a configuration.

[0075] In each of the test and measurement instruments shown in Figures 3-7, different components are shown in different arrangements, and those skilled in the art will appreciate that these components may be arranged in other situations, for example, threshold gates may or may not be placed at each bicoherence output, resamplers may be present in some instances and not in others, bispectrum may be used instead of bicoherence, higher order statistics may be incorporated, machine learning algorithms may be used rather than neural networks, etc.

[0076] As will be readily appreciated by those skilled in the art, each of the test and measurement instruments described above may incorporate additional components not illustrated in the examples. For example, each test and measurement instrument may include one or more processors, additional hardware or firmware, one or more memory components, a user interface, a display, etc.

[0077] Aspects of the disclosed technology can operate on specially created hardware, firmware, digital signal processors, or specially programmed general-purpose computers, including processors that operate according to programmed instructions. The term "controller" or "processor" in this application contemplates microprocessors, microcomputers, ASICs, and dedicated hardware controllers, among others. Aspects of the disclosed technology can be implemented in computer-available data and computer-executable instructions, such as one or more program modules, executed by one or more computers (including a monitoring module) or other devices. Generally, program modules include routines, programs, objects, components, data structures, and the like, which, when executed by a processor in a computer or other device, perform particular tasks or implement particular abstract data formats. The computer-executable instructions may be stored in computer-readable storage media, such as hard disks, optical disks, removable storage media, solid-state memory, RAM, and the like. As will be appreciated by those skilled in the art, the functionality of the program modules may be combined or distributed as desired in various embodiments. Furthermore, such functionality may be embodied in whole or in part in firmware or hardware equivalents, such as integrated circuits, field programmable gate arrays (FPGAs), etc. Certain data structures may be used to more effectively implement one or more aspects of the disclosed techniques, and such data structures are considered to be within the scope of the computer-executable instructions and computer-usable data described herein.

[0078] The disclosed aspects may, in some cases, be implemented in hardware, firmware, software, or any combination thereof. The disclosed aspects may also be implemented as instructions carried by or stored on one or more computer-readable media that may be read and executed by one or more processors. Such instructions may be referred to as computer program products. A computer-readable medium as described herein means any medium that can be accessed by a computing device. By way of example, and not limitation, computer-readable media may include computer storage media and communication media.

[0079] Computer storage media means any medium that can be used to store computer-readable information. By way of example and not limitation, computer storage media may include random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital video disc (DVD) or other optical disk storage, magnetic cassettes, magnetic tape, magnetic disk storage or other magnetic storage devices, and any other volatile or non-volatile removable or non-removable media implemented in any technology. Computer storage media excludes signals themselves and transitory forms of signal transmission.

[0080] A communication medium refers to any medium capable of communicating computer readable information. By way of example, and not limitation, communication media may include coaxial cables, fiber optic cables, air, or any other medium suitable for communicating electrical, optical, radio frequency (RF), infrared, acoustic, or other types of signals. [Example]

[0081] In the following, examples useful for understanding the technology disclosed in the present application are presented. The embodiment of the technology may include one or more of the examples described below and any combination thereof.

[0082] Example 1 teeth , a test and measurement device for analyzing a signal And an input configured to receive a signal; and a digital signal configured to convert the signal into a digital signal. Analog / Digital a converter; and one or more processors configured to determine a recovered clock signal based on the digital signal, set a window position for a fast Fourier transform of the digital signal, separate the digital signal into a series of windowed waveform data based on the window position, convert each of the windowed waveform data into frequency domain windowed waveform data using the fast Fourier transform, and determine higher order spectral data for each of the frequency domain windowed waveform data; and a neural network configured to receive the higher order spectral data of the frequency domain windowed waveform data and classify each windowed waveform data based on the higher order spectral data.

[0083] Example 2 is the test and measurement instrument of Example 1, further comprising a threshold gate configured to receive the high order spectral data, pass the high order spectral data that does not exceed a threshold, and set the high order spectral data that exceeds the threshold to a nominal value.

[0084] Example 3 is the test and measurement device of Example 2, where the minimum value is zero.

[0085] Example 4 is the test and measurement instrument of any of Examples 1 to 3, wherein the higher order spectral data includes at least one of a magnitude value, a phase value, or a complex value.

[0086] Example 5 is the test and measurement instrument of any of Examples 1 to 4, further comprising: a multiplexer switch configured to receive the frequency domain windowed waveform data and send it to a specific path based on a received bit pattern; and a bit pattern recognition neural network configured to receive the frequency domain windowed waveform data and output a bit pattern, wherein the one or more processors include a multiplexer switch configured to multiplex the frequency domain windowed waveform data to each path based on a received bit pattern. Next There is a process to determine the spectral data, and the high Next A separate neural network for each pass is provided to classify the frequency domain windowed waveform data based on the spectral data.

[0087] Example 6 is the test and measurement instrument of example 5, wherein the one or more processors are configured to average higher order spectral data for each of the frequency domain windowed waveform data, and each of the respective neural networks is configured to classify the windowed waveform data based on the average of the higher order spectral data.

[0088] Example 7 is the test and measurement instrument of any of Examples 5 or 6, wherein the one or more processors are further configured to determine a transfer function of a channel of the test and measurement instrument based on an output bit pattern of the bit pattern recognition neural network.

[0089] Example 8 is the test and measurement instrument of any of Examples 1 to 7, wherein the neural network is configured to output at least one of bit error rate, distortion, pattern decording, jitter measurement, signal to noise ratio, and inter-symbol interference based on the process of classifying the windowed waveform data.

[0090] Example 9 is the test and measurement instrument of any of Examples 1-8, wherein the one or more processors are configured to separate the digital signal into a series of windowed waveform data based on a window position using a Tukey window function.

[0091] Example 10 is a method for analyzing an input signal in a test and measurement system, the method including: receiving the input signal; determining a recovered clock signal based on the input signal; setting a window position for a Fast Fourier Transform of a digital signal; separating the input signal into a series of windowed waveform data based on the window position; converting the windowed waveform data into frequency domain windowed waveform data using a Fast Fourier Transform; determining higher order spectral data for each of the frequency domain windowed waveform data; and classifying each windowed waveform data by a neural network based on the higher order spectral data.

[0092] Example 11 is the method of example 10, further comprising the steps of: sending the high-order spectral data that does not exceed the threshold to the neural network; and setting the high-order spectral data that exceeds the threshold to a nominal value.

[0093] Example 12 is the method of any of examples 10 or 11, wherein the higher order spectral data includes at least one of magnitude values, phase values, or complex values.

[0094] Example 13 is a method according to any one of Examples 10 to 12, further comprising the steps of: detecting a bit pattern based on the frequency domain windowed waveform data by a bit pattern recognition neural network; and sending the frequency domain windowed waveform data to a specific neural network based on the bit pattern, and classifying the higher order spectral data of each of the frequency domain windowed waveform data based on the bit pattern.

[0095] Example 14 is the method of example 13, further comprising averaging the high order spectral data of each of the frequency domain windowed waveform data for a particular bit pattern prior to classifying the high order spectral data of each of the frequency domain windowed waveform data based on the bit pattern.

[0096] Example 15 is a method according to any one of Examples 10 to 14, wherein the process of classifying each windowed waveform data includes a process of outputting at least one of a bit error rate, distortion, pattern decoding, jitter measurement value, signal-to-noise ratio, and inter-symbol interference based on the process of classifying the windowed waveform data.

[0097] Example 16 is the method according to any one of Examples 10 to 15, wherein the process of separating the input signal into a series of windowed waveform data based on the window position uses a Tukey window function.

[0098] Example 17 is one or more non-transitory computer-readable storage media including instructions that, when executed by one or more processors of a test and measurement instrument, cause the test and measurement instrument to determine a recovered clock signal based on an input signal, set a window position for a fast Fourier transform of the digital signal, separate the input signal into a series of windowed waveform data based on the window position, convert each of the windowed waveform data into frequency-domain windowed waveform data using a fast Fourier transform, determine high-order spectral data for each of the frequency-domain windowed waveform data, and classify each windowed waveform data based on the high-order spectral data using a neural network.

[0099] Example 18 is the one or more non-transitory computer readable storage media of example 17, further comprising: Device Then, the higher-order spectral data that does not exceed the threshold is sent to the neural network. Rase The method further includes instructions that cause higher order spectral data that exceed the threshold to be set to a nominal value.

[0100] Example 19 is one or more non-transitory computer-readable storage media of any of Examples 17 or 18, further including instructions for causing the test measurement to detect bit patterns based on the frequency domain windowed waveform data by a bit pattern recognition neural network, send the frequency domain windowed waveform data to a specific neural network based on the bit patterns, and classify the higher order spectral data of each frequency domain windowed waveform data based on the bit patterns.

[0101] Example 20 is one or more non-transitory computer-readable storage media of Example 19, further including instructions for averaging the higher order spectral data of the frequency domain windowed waveform data for a particular bit pattern before causing the test measurement to classify the higher order spectral data of each frequency domain windowed waveform data based on the bit pattern.

[0102] The above-described versions of the disclosed subject matter have many advantages that have been described or that will be apparent to those of skill in the art. Nevertheless, not all of these advantages or features are required in every version of a disclosed device, system or method.

[0103] In addition, the description of this application refers to specific features. It should be understood that the disclosure herein includes all possible combinations of those specific features. When a particular feature is disclosed in the context of a particular aspect or embodiment, that feature can also be used in the context of other aspects and embodiments, to the extent possible.

[0104] Furthermore, when the application refers to a method having two or more defined steps or processes, those defined steps or processes may be performed in any order or simultaneously, unless the context precludes such a possibility.

[0105] For convenience of illustration, specific embodiments of the invention have been illustrated and described, but it will be understood that various modifications can be made therein without departing from the spirit and scope of the invention. Accordingly, the invention should not be limited, except as by the appended claims. [Explanation of symbols]

[0106] 300 Test and Measurement System 302 Continuous Time Linear Equalizer (CTLE) 304 Clock Recovery Block 306 Window Sequencer 308 Resampler 310 Window Function Block 312 Fast Fourier Transform 314 Higher Order Spectral Data 316 Threshold Gate 318 Neural Network Blocks 400 Test and Measurement System 402 Second Threshold Gate 500 Test and Measurement System 502 Derivative Function Block 504 Neural Network Pattern Classification Unit 506 Multiplexer Switch 508 Summation or Averaging Block 510 Transfer Function Block 600 Test and Measurement System 602 Second Multiplexer Switch

Claims

1. an input configured to receive a signal; an analog-to-digital converter configured to convert the signal into a digital signal; determining a recovered clock signal based on the digital signal; setting a window position for a fast Fourier transform of said digital signal; separating said digital signal into a series of windowed waveform data based on said window positions; converting each of the windowed waveform data into frequency domain windowed waveform data using a fast Fourier transform; Obtaining higher order spectrum data for each of the frequency domain windowed waveform data one or more processors configured to a neural network configured to receive the higher order spectral data of the frequency domain windowed waveform data and classify each of the windowed waveform data based on the higher order spectral data; 1. A test and measurement instrument for analyzing a signal comprising:

2. 2. The test and measurement instrument of claim 1, further comprising a threshold gate configured to receive said high order spectral data, pass said high order spectral data that does not exceed a threshold, and set said high order spectral data that does exceed said threshold to a minimum value.

3. a multiplexer switch configured to receive the frequency domain windowed waveform data and route it to a particular path based on a bit pattern received; a bit pattern recognition neural network configured to receive the frequency domain windowed waveform data and output the bit pattern; Further comprising:

2. The test and measurement instrument of claim 1, wherein one or more of said processors include a processor for determining said higher order spectral data of said frequency domain windowed waveform data for each of said paths, and a separate neural network is provided for each of said paths for classifying said frequency domain windowed waveform data based on said higher order spectral data for each of said paths.

4. 4. The test and measurement instrument of claim 3, wherein one or more of the processors are configured to average the higher order spectral data of each of the frequency domain windowed waveform data, and each of the individual neural networks is configured to classify the windowed waveform data based on the average of the higher order spectral data.

5. 1. A method for analyzing an input signal in a test and measurement system, comprising: receiving an input signal; determining a recovered clock signal based on the input signal; setting a window position for a fast Fourier transform of the digital signal; separating the input signal into a series of windowed waveform data based on the window position; converting the windowed waveform data into frequency domain windowed waveform data using a Fast Fourier Transform; determining higher order spectral data for each of the frequency domain windowed waveform data; classifying each of the windowed waveform data using a neural network based on the higher order spectral data; A method for analyzing an input signal comprising:

6. 6. The method of claim 5, further comprising the steps of: feeding said higher order spectral data that does not exceed a threshold to a neural network; and setting said higher order spectral data that exceeds said threshold to a local minimum.

7. detecting bit patterns based on the frequency domain windowed waveform data using a bit pattern recognition neural network; sending said frequency domain windowed waveform data to a specific neural network based on said bit pattern, and classifying said higher order spectral data of each of said frequency domain windowed waveform data based on said bit pattern; 6. The method of claim 5 further comprising:

8. 8. The method of claim 7, further comprising the step of averaging said higher order spectral data of each of said frequency domain windowed waveform data for a particular bit pattern prior to the step of classifying said higher order spectral data of each of said frequency domain windowed waveform data based on said bit pattern.

9. A computer program product which, when executed by one or more processors of a test and measurement instrument, causes the instrument to perform the method of any one of claims 5 to 8.

Images