Frequency offset estimation method and device for digital subcarrier coherent optical communication system

Abstract

The invention provides a frequency offset estimation method and device for a digital subcarrier coherent optical communication system, and relates to the technical field of communication. According to the method, when frequency offset estimation is carried out on a received digital subcarrier signal, empirical mode decomposition is carried out on the signal, namely, an eigenmode function derived by a noisy signal is used for carrying out frequency offset estimation on the signal; decomposing the signal into a plurality of eigenmode components and residual components; using different numbers of eigenmode components to reconstruct a plurality of signals with the residual components; through quartile outlier detection and clustering analysis, a reconstruction signal which most meets the requirement is selected in a self-adaptive mode to serve as a final reconstruction result; and finding out the frequency position of a specified point from the reconstructed signal to carry out frequency offset estimation. According to the method, blind frequency offset estimation in a PDM-16 / 64QAM DSCM system can be accurately and stably realized within a certain error allowable range, and the method has the advantages of being small in error, applicable to different roll-off factors, capable of processing large frequency offset and the like.

Description

Technical Field

[0001] The present invention belongs to the field of communication technologies, and particularly relates to a frequency offset estimation method and apparatus for a digital subcarrier coherent optical communication system. Background Art

[0002] Currently, with the rapid development of the digital economy, improving the single-carrier rate is a key measure for the next-generation backbone optical transport network to deeply match the growth demand of the basic network bandwidth. At higher spectral efficiency and transmission rate, traditional single-carrier modulation technologies gradually expose their limitations: signals are vulnerable to non-linear effects and equalization-enhanced phase noise interference in high-speed transmission scenarios, resulting in an increase in bit error rate and deterioration of system performance. Digital Subcarrier Multiplexing (DSCM) technology has received extensive attention from researchers in recent years due to its advantage of enhancing the tolerance of high baud rate signals to dispersion, filtering, and fiber non-linearity. It can be seen that DSCM technology has broad application prospects in next-generation high-speed optical transmission.

[0003] Different from single-carrier systems, DSCM technology combines different electrical signals by frequency shifting and then loads them onto a laser beam transmitted by a transmitter. After transmission through the channel, at the receiving end, the different electrical signals are separated by frequency shifting in the electrical domain and a filter, and then the separated electrical signals are processed by digital signal processing (DSP) algorithms for single-carriers. Due to the special nature of the DSCM spectrum, many DSP methods originally applicable to single-carrier systems are no longer applicable to DSCM systems. Due to the inherent frequency offset characteristics of lasers, the center wavelengths between the transmitter laser and the local oscillator laser cannot be constantly aligned, causing the overall rotation of the constellation diagram and ultimately affecting the transmission performance of the received signal. Excessive frequency offset will cause the filtering operation at the receiving end to be unable to extract a complete single-carrier, having an irreversible adverse impact on subsequent DSP. Therefore, the frequency offset estimation (FOE) algorithm is of prime importance in the DSCM system. However, traditional frequency offset estimation algorithms are not applicable to DSCM systems. Therefore, it is necessary to improve the FOE algorithm in DSCM systems.

[0004] There are many problems with the existing frequency offset estimation algorithms in DSCM systems, such as low spectral efficiency or limited estimation range. For example, in the prior art, a frame header signal is inserted at the very front of each frame of digital subcarrier signals, and it is used as a frequency correction sequence at the receiving end to perform frequency offset estimation and compensation. However, the frame header occupies a certain spectral width, resulting in reduced spectral efficiency. In the prior art, a preamble symbol is inserted into the signal frame to assist the receiving end in performing frequency offset estimation, but this scheme is not a blind equalization and also has the problem of low spectral efficiency. Moreover, none of the existing schemes have studied the range of frequency offset estimation. Summary of the Invention

[0005] Aiming at the above deficiencies in the prior art, a frequency offset estimation method and device for a digital subcarrier coherent optical communication system provided by the present invention solve the problem that the existing schemes are not applicable to all roll-off factors, and have the advantages of small error, applicability to this algorithm under different roll-off factors, and the ability to handle large frequency offsets.

[0006] To achieve the above objectives, the technical solution adopted by the present invention is as follows: A frequency offset estimation method for a digital subcarrier coherent optical communication system, comprising: Signal decomposition stage: Perform empirical mode decomposition on the received PDM-16 / 64QAM DSCM signal to obtain intrinsic mode functions and a residual component; Signal reconstruction stage: Take different numbers of intrinsic mode function components and the residual component, and combine them to obtain multiple reconstructed signals; Optimized reconstruction stage: Use the outlier detection of the interquartile range and density-based spatial clustering analysis to perform clustering analysis on the outliers of multiple reconstructed signals, obtain the optimal starting component of the intrinsic mode function of the reconstructed signal, and determine the optimal reconstructed signal; Frequency offset estimation stage: Based on the optimal reconstructed signal, by taking the frequency position corresponding to the item with the minimum power of each cluster and subtracting the initial frequency position at the transmitting end, obtain the magnitude of the frequency offset, and complete the frequency offset estimation of the digital subcarrier coherent optical communication system.

[0007] The beneficial effects of the present invention are as follows: A frequency offset estimation method applicable to a digital subcarrier 16 / 64 QAM coherent optical communication system provided by the present invention, when performing frequency offset estimation on the received digital subcarrier signal, first performs empirical mode decomposition (EMD) on the signal, that is, uses the intrinsic mode functions derived from the noisy signal itself to decompose the signal into multiple intrinsic mode components and a residual component. The frequencies of these components are arranged from high to low, and finally the overall slow-varying trend of the signal is obtained; secondly, different numbers of intrinsic mode components are used to reconstruct multiple signals with the residual component to achieve the purpose of noise removal; thirdly, through the outlier detection of the interquartile range and density-based spatial clustering analysis, the most compliant reconstructed signal is adaptively selected as the final reconstruction result; finally, the frequency position of the specified point is found for the reconstructed signal to perform frequency offset estimation. The present invention can accurately and stably achieve blind frequency offset estimation in the PDM-16 / 64QAM DSCM system within a certain error tolerance range, and has the advantages of small error, applicability to this algorithm under different roll-off factors, and the ability to handle large frequency offsets.

[0008] Further, the empirical mode decomposition is specifically as follows: A1. Represent the received symbols of each polarization state of the PDM-16 / 64QAM DSCM signal as R ; A2. Perform Fourier transform on the received symbol R to obtain the power spectral density , where represents the power at the m th frequency point; A3. When the power spectrum does not satisfy the definition of the intrinsic mode function, calculate the remaining signal ; A4. Based on the remaining signal , calculate the stopping criterion , set the threshold of the stopping criterion , and assign to ; A5. Judge whether the stopping criterion is greater than the threshold . If so, re-judge whether the power spectrum satisfies the definition of the intrinsic mode function, and return to A3. Otherwise, take the remaining signal as the j th intrinsic mode function, denoted as , and enter A6; A6. Use to perform on the power spectrum Update and determine the updated power spectrum to check if it conforms to the definition of the intrinsic mode function. If not, recalculate the intrinsic mode function. If it conforms, obtain multiple intrinsic mode function components and denote the updated power spectrum as the residual component .

[0009] The beneficial effect of the above further solution is that by performing EMD adaptive decomposition on the received signal, the noisy signal is divided into different IMF components, and multiple decomposition components of the signal are obtained.

[0010] Furthermore, the stop criterion and the expression of the remaining signal are as follows respectively:

[0011]

[0012] where represents the total number of all sample values of the power spectrum , m represents the power index of the power spectrum , represents the upper envelope sample value at the m -th frequency point, represents the lower envelope sample value at the m -th frequency point; The expression of the power spectrum is as follows: .

[0013] The beneficial effect of the above further solution is that by updating the power spectrum through the stop criterion, multiple IMF components are adaptively obtained.

[0014] Furthermore, the expression of the reconstructed signal is as follows:

[0015] where represents the reconstructed signal starting with the -th intrinsic mode function, represents the index number of the starting IMF component when performing signal reconstruction, N represents the number of intrinsic mode functions, represents the power of the j -th intrinsic mode function at the m -th frequency point, represents the power of the m -th frequency point of the residual component.

[0016] The beneficial effect of the above further solution is that the received signal is reconstructed by EMD, achieving the purpose of filtering the high-frequency noise attached to the received signal.

[0017] Furthermore, the determination of the optimal reconstructed signal is specifically as follows: B1. Determine the power peak point, and denote the power at the peak point as , with the unit of dBm; B2. Take out the starting index number and the ending index number of all points within the range of . Take out all points in the reconstructed signal whose index numbers are within the range of . Denote the power of all the taken-out points as , and denote the position index of all the taken-out points as , where ; B3. Sort from small to large according to the principle of interquartile range, and take out the power values at the 25% and 75% positions and denote them as and . Based on the taken-out values, obtain the interquartile range value . The expression of the interquartile range value is as follows: ; B4. Set the normal value range according to the interquartile range value , and determine that the points in whose power is less than the lower bound of the normal value range are outlier points. The calculation expression of the lower bound is as follows:

[0018] where takes the value of 2.5; B5. For the reconstructed signal , traverse within the range of to obtain groups of outlier points. Denote the indexes of the groups of outlier points as , where , , represents the total number of points in the q th group of outlier points; B6. Perform density-based spatial clustering processing on the indexes of the obtained groups of outlier points to obtain the clustering result , where represents the total number of categories after clustering, and represents the number of elements in each category; B7. Extract the clustering result , for a coherent optical communication system with M sub - carriers, only when , the optimal reconstructed signal is obtained. At this time, the q corresponding to the group of outliers is used as the index number of the optimal initial intrinsic mode function component of the reconstructed signal. Denote the index number at this time as . When and only when , is the optimal reconstructed signal for frequency offset estimation.

[0019] The beneficial effect of the above - mentioned further solution is: outlier detection is performed through the inter - quartile range to locate the depression position between sub - carriers. The density - based clustering processing method is used for the located depression position, and the points representing the depression position are classified to obtain the frequency of the depression position between sub - carriers.

[0020] Furthermore, the method for obtaining the frequency offset magnitude is as follows: C1. According to the optimal intrinsic mode function component of the reconstructed signal, determine the frequency components obtained in the reconstructed signal , M represents the total number of digital sub - carriers, represents the index number of the depression position between adjacent sub - carriers, ; C2. In a digital sub - carrier coherent optical communication system, for the signal transmitted by the transmitter, calculate the depression position between each adjacent sub - carrier; C3. Estimate the frequency offset magnitude based on the frequency components and the ideal frequency magnitude at the depression position of the transmitter, and complete the frequency offset estimation of the digital sub - carrier coherent optical communication system.

[0021] Furthermore, the obtained frequency components are as follows: D1. Extract the corresponding to the index number of the optimal intrinsic mode function component of the reconstructed signal, and divide them into groups according to the clustering result, where represents the clustering result after performing density - based spatial clustering analysis; ​D2. Determine the power value corresponding to the index number of each group of points in the group, denoted as , where represents the total number of categories after clustering; D3. Extract the index number corresponding to the minimum value in each group of power values ; D4. Determine the frequency component obtained in the power spectrum according to the value of the index number ;

[0022] The beneficial effect of the above further solution is: By using the result of density-based clustering analysis, the frequency of each depression position at the receiving end is obtained.

[0023] Furthermore, the ideal frequency at the depression position between all adjacent subcarriers at the transmitting end is expressed as follows:

[0024] where represents the baud rate, represents the roll-off factor.

[0025] The beneficial effect of the above further solution is: The frequency of the initial depression position at the transmitting end is calculated by the formula.

[0026] Furthermore, the expression of the frequency offset magnitude is as follows: .

[0027] The beneficial effect of the above further solution is: By subtracting the frequency component at the receiving end from the frequency component at the transmitting end, the frequency offset magnitude is obtained, and the frequency offset estimation is completed.

[0028] The present invention provides a frequency offset estimation device for a digital subcarrier coherent optical communication system, including: a signal decomposition module, configured to perform empirical mode decomposition on the received PDM-16 / 64QAM DSCM signal to obtain intrinsic mode functions and a residual component; a signal reconstruction module, configured to take different numbers of intrinsic mode function components and the residual component, and combine them to obtain multiple reconstructed signals; an optimized reconstruction module, configured to perform clustering analysis on the outliers of the multiple reconstructed signals by using the outlier detection of the interquartile range and density-based spatial clustering analysis, obtain the optimal starting component of the intrinsic mode function of the reconstructed signal, and determine the optimal reconstructed signal; ​​The frequency offset estimation module is used to perform frequency offset estimation on the digital subcarrier coherent optical communication system. Based on the optimal reconstructed signal, it obtains the frequency offset magnitude by taking the difference between the frequency positions corresponding to the minimum power terms of each cluster and the initial frequency position at the transmitting end.

[0029] The beneficial effects of the present invention are as follows: A frequency offset estimation method applicable to a digital subcarrier 16 / 64 QAM coherent optical communication system. When performing frequency offset estimation on the received digital subcarrier signal, first, the signal is subjected to EMD, that is, using the intrinsic mode functions derived from the noisy signal itself, the signal is decomposed into multiple intrinsic mode components and a residual component; second, different numbers of intrinsic mode components are used to reconstruct multiple signals with the residual component; third, through the outlier detection of the interquartile range and DBSCAN clustering analysis, the most suitable reconstructed signal is adaptively selected as the final reconstruction result; finally, the frequency position of the specified point is found for the reconstructed signal to perform frequency offset estimation. The present invention can accurately and stably achieve blind frequency offset estimation in the DSCM-PDM-16 / 64 QAM system within a certain error tolerance range, and has the advantages of small error, applicability of this algorithm under different roll-off factors, and the ability to handle large frequency offsets. Brief Description of the Drawings

[0030] Figure 1 It is the flowchart of the method of the present invention.

[0031] Figure 2 It is the schematic diagram of the simulation system.

[0032] Figure 3 It is the flowchart taking four subcarriers as an example.

[0033] Figure 4 It is the schematic diagram of the estimated frequency offset results under different numbers of subcarriers and different FFT lengths.

[0034] Figure 5 It is the relationship diagram between the set frequency offset and the estimated frequency offset of 16QAM with different numbers of subcarriers.

[0035] Figure 6 It is another relationship diagram between the set frequency offset and the estimated frequency offset of 16QAM with different numbers of subcarriers.

[0036] Figure 7 It is the performance diagram of the AEMD scheme and the comparison scheme with different numbers of subcarriers under a low roll-off factor.

[0037] Figure 8 The schematic diagram of the estimation error of the AEMD scheme and the comparison scheme with different numbers of subcarriers.

[0038] Figure 9 It is the schematic diagram of the system structure of the present invention. Detailed Embodiments

[0039] The specific embodiments of the present invention will be described below to facilitate those skilled in the art to understand the present invention. However, it should be clear that the present invention is not limited to the scope of the specific embodiments. For those of ordinary skill in the art, as long as various changes are within the spirit and scope of the present invention defined and determined by the appended claims, these changes are obvious, and all inventions made using the concept of the present invention are within the scope of protection.

[0040] Embodiment 1 In this embodiment, DSCM (Digital Subcarrier Multiplexing Systems) is digital subcarrier multiplexing, and PDM is polarization multiplexing.

[0041] In this embodiment, the present invention provides a frequency offset estimation method applicable to a digital subcarrier 16 / 64 QAM coherent optical communication system. This method is applicable to various roll-off factors and different numbers of subcarrier systems. The frequency offset estimation range is [-5 GHz, 5 GHz], and the estimation error is within 50 MHz. As Figure 1 shown, the present invention provides a frequency offset estimation method for a digital subcarrier coherent optical communication system, and its implementation method is as follows: S1. Signal decomposition stage: Perform empirical mode decomposition on the received PDM-16 / 64QAM DSCM signal to obtain intrinsic mode functions and residual components; Performing empirical mode decomposition specifically includes: A1. Represent the received symbols of each polarization state of the PDM-16 / 64QAM DSCM signal as R ; A2. Perform Fourier transform on the received symbol R to obtain the power spectral density , where represents the power at the m th frequency point; A3. When the power spectrum does not meet the definition of the intrinsic mode function, calculate the remaining signal ; A4. Based on the remaining signal , calculate the stop criterion , set the threshold of the stop criterion , and assign to ; A5. Determine whether the stop criterion is greater than the threshold . If so, re-judge the power spectrum Whether it meets the definition of the intrinsic mode function, and return A3. Otherwise, the remaining signal as the j th intrinsic mode function, denoted as , and enter A6; A6. Use to update the power spectrum , and determine whether the updated power spectrum meets the definition of the intrinsic mode function. If not, recalculate the intrinsic mode function. If it meets, obtain multiple intrinsic mode function components, and denote the updated power spectrum as the residual component .

[0042] In this embodiment, first, the received 16 / 64 QAM signal is subjected to empirical mode decomposition. Specifically, the intrinsic mode function derived from the signal itself is used as the basis function, and the remaining signal is obtained by subtracting the average of the upper and lower envelopes from the current signal. The remaining signal is continuously updated according to the definition of the intrinsic mode function, and finally multiple intrinsic mode function components and 1 residual component are obtained.

[0043] In this embodiment, S1 is specifically as follows: S101. Calculate the intrinsic mode function: Express the received symbol of each polarization state of DSCM-16 / 64QAM as: (1) where represents the transmitted symbol, represents the frequency offset to be estimated, is the symbol period.

[0044] First, perform a Fourier transform on the received symbol to obtain the power spectrum . Assume that the total number of samples of is . Judge whether it meets the definition of the intrinsic mode function. If not, calculate the temporary remaining signal, and the specific calculation formula is as follows: (2) where represents the total number of all samples of the power spectrum , m represents the power index of the power spectrum , represents the upper envelope sample value of the m rd frequency point, represents the mThe lower envelope sample values at a frequency point.

[0045] Finally, calculate the stopping criterion using the following formula, set the threshold of the stopping criterion, and then assign to .

[0046] (3) Where, represents the stopping criterion. Denote as the threshold of the stopping criterion. If > , then re - judge whether satisfies the definition of the intrinsic mode function, and repeat the corresponding steps of formula (2) and formula (3) above. Otherwise, take as the j th intrinsic mode function, denoted as .

[0047] S102. Obtain all intrinsic mode components and the residual component: Use the obtained in step S101 to update , and the specific calculation formula is: (4) Judge whether conforms to the definition of the intrinsic mode function. If not, repeat the operations in step S101. If it conforms to the definition, or if 10 intrinsic mode function components have been obtained, then take as the residual component .

[0048] S2. Signal reconstruction stage: Take different numbers of intrinsic mode function components and the residual component, and combine them to obtain multiple reconstructed signals; In this embodiment, signal reconstruction is a process of determining the index number of the starting component, adding the intrinsic mode function components after this index number, and adding them to the residual component to obtain the reconstructed signal. Multiple reconstruction results are obtained by changing the index number of the starting component.

[0049] In this embodiment, signal reconstruction is performed on the intrinsic mode function components and the residual component obtained in S1. The specific calculation formula is: (5) Where, represents the reconstructed signal with the th intrinsic mode function as the starting component, represents the index number of the starting IMF component when performing signal reconstruction, represents the mThe power of a frequency point, denotes the j th intrinsic mode function at the m th frequency point, N denotes the number of intrinsic mode functions. By changing the value, multiple reconstructed signals can be obtained.

[0050] S3. Optimization and reconstruction stage: Using the outlier detection of the interquartile range and density-based spatial clustering analysis, perform clustering analysis on the outliers of multiple reconstructed signals to obtain the starting component of the optimal intrinsic mode function of the reconstructed signal and determine the optimal reconstructed signal. Specifically: B1. Determine the highest power point and denote the power of the highest point as , with the unit of dBm; B2. Take out the starting index number and the ending index number of all points within the range of . Take out all points in the reconstructed signal whose index numbers are within the range of . Denote the power of all taken-out points as , and denote the position index of all taken-out points as , where ; B3. Use the principle of the interquartile range to sort from smallest to largest, and take out the power values at the 25% and 75% positions and denote them as and . Based on the taken-out values, obtain the interquartile difference ; The expression of the interquartile difference is as follows: ; B4. Set the normal value range according to the interquartile difference , and determine that the points in with power less than the lower bound of the normal value range are outliers. The calculation expression of the lower bound is as follows:

[0051] where takes the value of 2.5; B5. For the reconstructed signal , traverse within the range of to obtain groups of outliers, and denote the indices of the groups of outliers as , where , , , represents the total number of points in the q th group of outliers; B6. For the obtained indexes of the groups of outliers perform density-based spatial clustering processing respectively to obtain clustering results , where represents the total number of categories after clustering, represents the number of elements in each category; B7. Take out the total number of categories in the clustering results . For a coherent optical communication system with M subcarriers, only when , the optimal reconstructed signal is obtained. At this time, the q corresponding to the th group of outliers is used as the index number of the optimal initial intrinsic mode function component of the reconstructed signal. The index number at this time is denoted as . When and only when , is the optimal reconstructed signal for frequency offset estimation.

[0052] In this embodiment, for the obtained multiple reconstruction results, the outlier detection using the interquartile range and the density-based spatial clustering analysis are used to obtain the index number of the optimal intrinsic mode function component of the reconstructed signal, so as to obtain the optimal reconstruction result. Specifically: S301. Calculate the interquartile range to find outliers: First, find the point with the highest power, and denote the power of this point as (dBm). Take out the starting index number and the ending index number of all points within the power range of . Take out all points in the reconstructed signal whose index numbers are within the range of . Denote the power of the taken-out points as , and the index number as , where represents the power values of all taken-out points, represents the position indexes of all taken-out points, . Using the principle of the interquartile range, after arranging from small to large, take out the values at the 25% and 75% positions and denote them as and . The expression of the interquartile difference is as follows: (6) where Denote the interquartile range. According to set the normal value range, and the points outside the range are considered outliers. The lower bound of this range is expressed as: (7) where takes the value of 2.5. The points with medium power less than are the outliers. By changing the value, the following groups of outliers can be obtained. Denote the indices of the obtained outliers as , . Denote the total number of points in the q th group of outliers as , ; S302. Perform clustering analysis according to DBSCAN: Perform density-based spatial clustering processing on the obtained in the above step S301 respectively. Denote the clustering result as , . Take out the number of different classifications included in . For a digital subcarrier system with subcarriers, if and only if , it is considered that the optimal reconstructed signal is obtained. At this time, the q th group of outliers corresponds to which is the index number of the optimal initial intrinsic mode function component of the reconstructed signal. Substitute it into formula (5) to obtain the optimal reconstruction result.

[0053] S4. Frequency offset estimation stage: Based on the optimal reconstructed signal, by taking out the frequency position corresponding to the minimum power term of each cluster and subtracting it from the ideal frequency position at the transmitting end, the frequency offset magnitude is obtained, and the frequency offset estimation of the digital subcarrier coherent optical communication system is completed. Specifically: C1. Determine the frequency components obtained in the reconstructed signal according to the optimal intrinsic mode function component of the reconstructed signal , M represents the total number of digital subcarriers, represents the index number of the depression position between adjacent subcarriers, ; The obtained frequency components , specifically: D1. Take out the corresponding to the index number of the optimal intrinsic mode function component of the reconstructed signal and divide it into groups according to the clustering result. Among them, Indicates the clustering result after performing density-based spatial clustering analysis; D2. Determine The power value corresponding to the index number of each group of points in the group, denoted as , where Indicates the total number of categories after clustering; D3. Take out the power value of each group The index number corresponding to the minimum value in ; D4. According to the index number value, determine the frequency component obtained in the power spectrum ; ; C2. In a digital subcarrier coherent optical communication system, for the signal transmitted by the transmitter, calculate the depression position between each adjacent subcarrier; C3. Based on the frequency component and the ideal frequency magnitude at the depression position of the transmitter , estimate the frequency offset magnitude , and complete the frequency offset estimation of the digital subcarrier coherent optical communication system.

[0054] The obtained frequency component , specifically: take out the corresponding to the index number of the optimal intrinsic mode function component of the reconstructed signal, and divide it into groups according to the clustering result, where represents the classification result after density-based spatial clustering analysis, which will rearrange the original data according to the classification result, and determine The power value corresponding to the index number of each group of points in the group, denoted as ; where represents the total number of categories after density-based spatial clustering. For example, if the original data is divided into three categories, then = 3, take out the index number corresponding to the minimum value in each group of power values ; According to the index number value, determine the frequency component obtained in the reconstructed signal ; ; ;

[0055] In this embodiment, for the obtained reconstructed signal, while obtaining the index of the optimal component, the result of density-based spatial clustering analysis corresponding to this index is also exactly the result of optimally finding the frequency offset estimation point. By taking out the frequency position corresponding to the power minimum term in each cluster and subtracting the initial frequency position of the transmitter, the magnitude of the frequency offset can be estimated. Specifically: S401. Calculate the frequency components: Take out the optimal index number obtained in the step S302 corresponding to , and divide them into groups according to the clustering results. In the reconstructed signal , find the power values corresponding to the index numbers of each group of points in the groups, denoted as (dBm), where . Then take out the index number corresponding to the minimum value in each group , and according to the value of , the corresponding frequency components can be obtained from the frequency spectrum diagram of . .

[0056] S402. Calculate the estimated frequency offset value: In a digital subcarrier system, for the signal transmitted by the transmitter, the depression position between every two adjacent subcarriers can be calculated by the following formula: (8) where, represents the frequency magnitude of the depression positions of all adjacent subcarriers, represents the baud rate, represents the roll-off factor. Subtract the frequency position obtained in the step S401 from , and the frequency offset magnitude can be estimated. The specific calculation formula is: (9) where, represents the frequency magnitude of the depression positions of all adjacent subcarriers, k represents the index number of the depression position between adjacent subcarriers. For example, for 4 subcarriers, there are 3 depression positions, so k = 1, 2, 3, M represents the total number of digital subcarriers.

[0057] In this embodiment, in order to verify the effectiveness of the present invention, in this embodiment, by using the optical communication simulation software VPI Design Suite 11.1 and MATLAB, the present invention verifies the feasibility of the proposed method. Hereinafter, AEMD is used as the abbreviation of the present invention. The schematic diagram of the simulation system is as shown in Figure 2As shown, where the transmitter is used to generate a 64 GBaud PMD-16 / 64 QAM signal. The roll-off factor is set in the range of 0.01 to 0.05 with a step of 0.01, and in the range of 0.1 to 0.25 with a step of 0.5. The central wavelength of the laser emission end is 1550 nm. During the simulation, the linewidth of the laser is set to 100 kHz. To verify the performance of the algorithm, the frequency offset fluctuation range used in the simulation experiment is set in [-5 GHz, 5 GHz], and the step is set to 1 GHz. The simulation explores the effectiveness of the proposed frequency offset estimation method by adding Amplifier Spontaneous Emission (ASE) noise. Among them, an optical signal-to-noise ratio module is set to introduce different ASE noises into the transmission link. The transmitter uses subcarrier multiplexing to implement digital subcarrier signals.

[0058] In this embodiment, the proposed method is described in detail taking four subcarriers as an example. Figure 3 (a)is the spectrogram of DSCM-16 QAM with a 2 GHz frequency offset added. First, the signal spectrogram is decomposed by empirical mode decomposition to obtain 10 IMF components and a residual component. Subsequently, change the index number of the initial reconstruction component, such as Figure 3 (b)shows that from left to right are the reconstruction results with , and as the starting IMF components. Next, IQR quartile outlier detection is performed on these reconstructed signals, and the found outliers are subjected to DBSCAN clustering analysis. Finally, the optimal initial reconstruction component is found according to the number of classifications for signal reconstruction. As Figure 3 (c)shows that from left to right, the clustering analysis results divide the outliers into 1 class, 3 classes, and 5 classes respectively. For the DSCM signal with four subcarriers, it has 3 depression positions, and it is most desirable to get a clustering result of 3 classes. Thus, in this case, j =5 can obtain the optimal reconstruction result. The reconstruction result is as shown in Figure 3 (d). Taking the outlier detection result corresponding to this index number, the adjacent subcarrier boundary depression position required for calculating the frequency offset can be located.

[0059] In this embodiment, the algorithm parameters are first optimized. The accuracy of the present invention depends on the size of the spectral resolution. To reduce the complexity without affecting the performance of the proposed scheme, 64 GBaud PDM-64QAM DSCM constellation symbols are generated, pulse shaping is performed using a square root raised cosine filter with a roll-off factor of 0.1, the transmission distance is 350 km, the cumulative dispersion is 5600 ps / nm, the center wavelength of the laser is 1550 nm. At the receiver, the frequency offset and linewidth of the local oscillator laser are set to 2 GHz and 200 kHz respectively. Under the above conditions, the system performance is tested, and the results are as Figure 4 shown. It can be found that: in the process of increasing the FFT length from 512 to 2048, the estimated frequency offset results of systems with different numbers of subcarriers are getting closer and closer to the set frequency offset of 2 GHz. When the FFT length reaches 2048 and above, the frequency offset estimation accuracy of systems with different numbers of subcarriers will no longer change significantly. To reduce the system complexity while ensuring the accuracy, the FFT length is set to 2048.

[0060] To test the performance upper limit of the proposed scheme, the FO in the range of [-5 GHz, 5 GHz] is traversed with a step of 1 GHz. In this embodiment, the roll-off factor is set to 0.01. Figure 5 For the verification of the frequency offset estimation range, the estimated frequency offset of the present invention can better fit the curve of the set frequency offset in the range of [-5 GHz, 5 GHz]. Considering that the frequency offset of current commercial lasers affected by temperature is within 3 GHz, the present invention has strong applicability. Figure 6 For the relationship between the frequency offset estimation error of different subcarrier systems and the set frequency offset, it can be found that the frequency offset estimation error of the present invention is within 50 MHz.

[0061] In this embodiment, the present invention compares the proposed AEMD scheme with the traditional scheme of finding the depression position, and names the comparison scheme the Dip scheme. Figure 7 are the results of the FO estimated by the two schemes under the roll-off factor [0.01, 0.05]. It can be found that due to the inability of the traditional scheme to accurately find the position of the depression point, when the roll-off factor is 0.05, the maximum error reaches 200 MHz, and when the roll-off factor is 0.01, the maximum error is close to 1 GHz. It can be seen that the traditional scheme even has the problem of algorithm failure as the roll-off factor decreases. For the AEMD scheme of the present invention, the performance is better than that of the traditional scheme under different roll-off factors, and the maximum error is within 50 MHz.

[0062] In this embodiment, to further illustrate the robustness of the present scheme to the roll-off factor, the frequency offset estimation error in the range of the roll-off factor [0.01, 0.25] is traversed and compared with the traditional scheme. Figure 8As the estimated results, it can be found that the traditional scheme is greatly affected by the transformation of the roll-off factor, and FO estimation errors above 200 MHz occur when the roll-off factor is small or large. For the AEMD scheme of the present invention, the algorithm is applicable under different roll-off factors, showing strong robustness to the roll-off factor.

[0063] Embodiment 2 As Figure 9 shown, the present invention provides a frequency offset estimation device for a digital subcarrier coherent optical communication system, which is used to execute the frequency offset estimation method of the digital subcarrier coherent optical communication system described in Embodiment 1, including: A signal decomposition module, configured to perform empirical mode decomposition on the received PDM-16 / 64QAM DSCM signal to obtain intrinsic mode functions and a residual component; A signal reconstruction module, configured to take different numbers of intrinsic mode function components and the residual component, and combine them to obtain multiple reconstructed signals; An optimized reconstruction module, configured to analyze the multiple reconstructed signals by using the outlier check of the interquartile range and cluster analysis, and determine the optimal reconstructed signal by obtaining the optimal intrinsic mode function component of the reconstructed signal; A frequency offset estimation module, configured to, based on the optimal reconstructed signal, obtain the frequency offset magnitude by taking the difference between the frequency position corresponding to the minimum power term of each cluster and the initial frequency position at the transmitting end, and complete the frequency offset estimation of the digital subcarrier coherent optical communication system.

[0064] In this embodiment, the present application can divide the functional units according to the frequency offset estimation method. For example, each function can be divided into each functional unit, or two or more functions can be integrated into one processing unit. The above integrated unit can be implemented in the form of hardware or in the form of a software functional unit. It should be noted that the division of units in the present invention is illustrative, only a logical division, and there may be other division methods in actual implementation.

[0065] In this embodiment, in order to implement the principle and beneficial effects of the frequency offset estimation method, the frequency offset estimation device includes the corresponding hardware structure and / or software module for executing each function. Those skilled in the art should easily realize that, in combination with the schematic units and algorithm steps described in the embodiments disclosed in the present invention, the present invention can be implemented in the form of hardware and / or the combination of hardware and computer software. Whether a certain function is executed in the way of hardware or computer software driving depends on the specific application and design constraints of the technical solution. Different methods can be used for each specific application to implement the described function, but such implementation should not be considered to exceed the scope of the present application.

[0066] In summary, a frequency offset estimation device proposed by the present invention is applicable to a digital subcarrier 16 / 64 QAM coherent optical communication system, and this device is applicable to systems with various roll-off factors and different numbers of subcarriers. The frequency offset estimation range is [-5 GHz, 5 GHz], and the estimation error is within 50 MHz.

Claims

1. A frequency offset estimation method for a digital subcarrier coherent optical communication system, characterized in that Including: Signal decomposition stage: Perform empirical mode decomposition on the received PDM-16 / 64QAM DSCM signal to obtain intrinsic mode functions and a residual component; Signal reconstruction stage: Take different numbers of intrinsic mode function components and the residual component, and combine them to obtain multiple reconstructed signals; Optimized reconstruction stage: Use the outlier detection of the interquartile range and density-based spatial clustering analysis to perform clustering analysis on the outliers of multiple reconstructed signals, obtain the optimal starting component of the intrinsic mode function of the reconstructed signal, and determine the optimal reconstructed signal; Frequency offset estimation stage: Based on the optimal reconstructed signal, by taking the frequency position corresponding to the minimum power term of each cluster and subtracting it from the initial frequency position at the transmitting end, obtain the frequency offset magnitude, and complete the frequency offset estimation of the digital subcarrier coherent optical communication system.

2. The frequency offset estimation method for a digital subcarrier coherent optical communication system according to claim 1, characterized in that, The specific implementation of the empirical mode decomposition is as follows: A1. Represent the received symbols of each polarization state of the received PDM-16 / 64QAM DSCM signal as R ; A2. Fourier transform the received symbols R to obtain the power spectral density , where represents the power at the m -th frequency point; A3. Regarding the power spectrum When it does not satisfy the definition of the intrinsic mode function, the residual signal is calculated ; A4. Based on the residual signal , calculate the stop criterion , set the threshold of the stop criterion , and assign to and assign it to ; A5. Judgment Stop Criterion Is it greater than the threshold , if so, re-judge the power spectrum Whether it meets the definition of the intrinsic mode function, and return A3, otherwise, the remaining signal As the j th intrinsic mode function, denoted as , enter A6; A6. Using to update the power spectrum and determine whether the updated power spectrum meets the definition of the intrinsic mode function. If not, recalculate the intrinsic mode function. If so, obtain multiple intrinsic mode function components and denote the updated power spectrum as the residual component .

3. The frequency offset estimation method for a digital subcarrier coherent optical communication system according to claim 2, characterized in that, The stop criterion and the remaining signal are expressed as follows respectively: Among them, represents the total number of all sample values of the power spectrum , m represents the power index of the power spectrum , represents the upper envelope sample value of the m th frequency point , m represents the lower envelope sample value of the th frequency point; The power spectrum has the following expression: 。 4. The frequency offset estimation method for a digital subcarrier coherent optical communication system according to claim 1, wherein The expression of the reconstructed signal is as follows: Among them, represents the reconstructed signal with the th intrinsic mode function as the starting component, represents the index number of the starting IMF component when performing signal reconstruction, N represents the number of intrinsic mode functions, represents the j th intrinsic mode function at the m th frequency point's power, represents the m th frequency point's power of the residual component.

5. The frequency offset estimation method of the digital subcarrier coherent optical communication system according to claim 1, wherein The specific implementation of determining the optimal reconstructed signal is as follows: B1. Determine the power peak point and record the power at the peak point as , with the unit of dBm; B2. Retrieve the starting index number of all points within the range, as well as the ending index number , and retrieve the reconstructed signal . For all points within the index number range of , record the power of all retrieved points as , and record the position index of all retrieved points as , where ; B3. Using the principle of interquartile range, sort them in ascending order, and take the power values at the 25% and 75% positions, denoted as and , and based on the extracted values, obtain the interquartile range value ; the expression of the interquartile range value is as follows: ； B4. According to the interquartile range Set the normal value range and determine Points where the medium power is less than the lower bound of the normal value range are outlier points, and the calculation expression of the lower bound is as follows: Among them, The value is 2.5; B5. For the reconstructed signal , traverse within the range of to obtain groups of outliers, and denote the indices of the groups of outliers as , where , , , represents the total number of points in the q th group of outliers; B6. For the obtained indexes of the outlier groups perform density-based spatial clustering processing respectively to obtain clustering results , where represents the total number of categories after clustering, represents the number of elements in each category; B7. Extract the clustering results The total number of categories , for a coherent optical communication system with M sub - carriers, only when the case, the optimal reconstructed signal is obtained. At this time, the q group of corresponding outlier points is used as the index number of the optimal initial intrinsic mode function component of the reconstructed signal. The index number at this time is denoted as . When and only when , is the optimal reconstructed signal for frequency offset estimation.

6. The frequency offset estimation method for a digital subcarrier coherent optical communication system according to claim 5, characterized in that The specific implementation of obtaining the frequency offset magnitude is as follows: C1. Determine the frequency components obtained in the reconstructed signal according to the optimal intrinsic mode function components of the reconstructed signal in the reconstructed signal , M denotes the total number of digital subcarriers denotes the index number of the depression position between adjacent subcarriers ; C2. In the digital subcarrier coherent optical communication system, for the signal transmitted by the transmitting end, calculate the depression position between each adjacent subcarrier; C3. According to the frequency components and the ideal frequency magnitude at the depression position of the transmitting end , the frequency offset magnitude is estimated , completing the frequency offset estimation of the digital subcarrier coherent optical communication system.

7. The frequency offset estimation method for a digital subcarrier coherent optical communication system according to claim 6, characterized in that, The obtained frequency components , specifically: D1. Take out the one corresponding to the index number of the optimal intrinsic mode function component of the reconstructed signal, and divide it into groups according to the clustering result, where represents the clustering result after performing density-based spatial clustering analysis; ​ D2. Determine the power value corresponding to the index number of each group of points in the group, denoted as , where represents the total number of categories after clustering; D3. Extract the index number corresponding to the minimum value in each set of power values ; ; D4. Determine the frequency component obtained in the power spectrum according to the value of the index number .

8. The frequency offset estimation method for a digital subcarrier coherent optical communication system according to claim 6, characterized in that The ideal frequency at the depression position between all adjacent subcarriers at the transmitting end has the following expression: Among them, represents the baud rate, represents the roll-off factor.

9. The frequency offset estimation method for a digital subcarrier coherent optical communication system according to claim 6, wherein The magnitude of the frequency offset has the following expression: 。 10. A frequency offset estimation device for a digital sub-carrier coherent optical communication system, which is used to perform the frequency offset estimation method of the digital sub-carrier coherent optical communication system according to any one of claims 1-9, characterized in that, Including: A signal decomposition module, which is used to perform empirical mode decomposition on the received PDM-16 / 64QAM DSCM signal to obtain intrinsic mode functions and a residual component; A signal reconstruction module, which is used to take different numbers of intrinsic mode function components and the residual component, and combine them to obtain multiple reconstructed signals; An optimized reconstruction module, which is used to use the outlier detection of the interquartile range and density-based spatial clustering analysis to perform clustering analysis on the outliers of multiple reconstructed signals, obtain the optimal starting component of the intrinsic mode function of the reconstructed signal, and determine the optimal reconstructed signal; A frequency offset estimation module, which is used to based on the optimal reconstructed signal, by taking the frequency position corresponding to the minimum power term of each cluster and subtracting it from the initial frequency position at the transmitting end, obtain the frequency offset magnitude, and complete the frequency offset estimation of the digital subcarrier coherent optical communication system.

Images