Carrier frequency offset estimation method for dual-polarization coherent light communication

Abstract

The invention belongs to the technical field of coherent light communication, and relates to a carrier frequency offset estimation method for dual-polarization coherent light communication, which comprises the following steps of: performing DSP (Digital Signal Processor) processing on original data at a transmitting end, and modulating the data processed by the DSP by using an IQ (In-phase Quadrature) modulator to generate a dual-polarization coherent light signal; performing power compensation on the dual-polarization coherent optical signal; a receiving end receives the signal and performs coherent demodulation on the received signal; performing DSP processing on the signal after coherent demodulation to obtain a restored signal; in the DSP processing, a coarse and fine positioning algorithm is adopted to process the signals, and a training sequence is obtained; carrying out carrier frequency offset estimation on the optimized signal according to the training sequence; performing phase offset compensation and data recovery on the signal according to a carrier frequency offset compensation result; according to the invention, the coarse positioning algorithm and the fine positioning algorithm are adopted to accurately position the training sequence in the received signal, and the carrier frequency offset estimation is carried out on the signal based on the training sequence, so that the estimation accuracy is improved.

Description

Technical Field

[0001] The invention belongs to the technical field of coherent optical communication, and in particular relates to a carrier frequency offset estimation method for dual-polarization coherent optical communication. Background Art

[0002] With the rapid growth of global data traffic, the basic requirements for optical fiber transmission are long distance, high speed and large capacity. Coherent optical communication provides a solution for high-speed, long-distance and large-capacity transmission due to its high receiving sensitivity and support for high-order modulation formats such as QPSK and 16-QAM. However, in coherent optical transmission systems, due to the influence of the manufacturing process, it is impossible to make the frequency of the transmitting laser and the local oscillator (LO) laser exactly the same. The frequency offset of the carrier mainly comes from the fact that the frequency of the transmitting and receiving lasers is not exactly the same. This offset makes demodulation difficult, which in turn leads to inter-symbol interference (ISI) and synchronization errors. Therefore, how to achieve high-precision and low-complexity carrier frequency offset estimation in the receiving DSP has become a key issue in improving the performance of the transmission system.

[0003] At present, a variety of algorithms have been proposed for estimating carrier frequency offset. For example, the fourth power frequency offset estimation algorithm that can only be used for QPSK signals, this algorithm multiplies the received signal with the complex conjugate of its previous signal, then performs a fourth power operation, and then sums a large number of results, takes the phase angle and divides by 4 to obtain the frequency offset. However, the frequency offset range that can be estimated by this algorithm is small. Based on the fourth power frequency offset estimation algorithm, a 16-QAM feedforward carrier frequency recovery algorithm based on the QPSK segmentation scheme is proposed for the 16QAM modulation format. The constellation points of the 16QAM signal are divided into two categories, and the corresponding QPSK constellation points are selected from the 16QAM constellation points for fourth power frequency offset estimation. Since only a part of the points are used when estimating the frequency offset, the performance of the algorithm is reduced. Another commonly used method is to use fast Fourier transform (FFT) to estimate the carrier frequency offset. The basic idea is to perform FFT transformation on the phase angle of the received signal, and divide the obtained peak frequency by 4 to obtain the frequency offset value. However, since FFT transformation is required, the computational complexity of this algorithm is relatively high. Summary of the invention

[0004] In order to solve the problems existing in the above prior art, the present invention proposes a carrier frequency offset estimation method for dual-polarization coherent optical communication, which comprises: the system of dual-polarization coherent optical communication comprises a transmitting end, an optical fiber channel and a receiving end; performing DSP processing on original data at the transmitting end, and using an IQ modulator to modulate the data after DSP processing to generate a dual-polarization coherent optical signal; performing power compensation on the dual-polarization coherent optical signal in the optical fiber channel and transmitting it to the receiving end; the receiving end receives the signal and performs coherent demodulation on the received signal; performing DSP processing on the coherently demodulated signal to obtain a restored signal;

[0005] The DSP processing of the coherently demodulated signal at the receiving end includes: compensating and optimizing the coherently demodulated signal to obtain an optimized signal; calculating a training sequence in the optimized signal; estimating the carrier frequency offset of the optimized signal according to the training sequence; compensating the signal for frequency offset according to the carrier frequency offset estimation result, and performing phase offset compensation and data recovery according to the result of the frequency offset compensation.

[0006] Beneficial effects of the present invention:

[0007] The present invention adopts a coarse positioning algorithm and a fine positioning algorithm to accurately position a training sequence in a received signal, and estimates a carrier frequency offset of the signal based on the training sequence, thereby improving the accuracy of the estimation and reducing the complexity compared to other algorithms. BRIEF DESCRIPTION OF THE DRAWINGS

[0008] Figure 1 It is a principle framework structure diagram of the carrier frequency offset estimation method for dual-polarization coherent optical communication of the present invention;

[0009] Figure 2 It is a structural block diagram of the simulation system of the present invention;

[0010] Figure 3 This is a result diagram of the coarse positioning algorithm of the present invention;

[0011] Figure 4 This is a result diagram of the fine positioning algorithm of the present invention;

[0012] Figure 5 The phase of the training sequence of the present invention is the result before and after the first-order linear fitting based on the least square method;

[0013] FIG6 is a diagram showing the carrier frequency offset estimation accuracy and range results of the algorithm of the present invention and the comparative algorithm; wherein (a) is the X polarization state of the comparative algorithm; (b) is the Y polarization state of the comparative algorithm; (c) is the X polarization state of the algorithm of the present invention; and (d) is the Y polarization state of the algorithm of the present invention;

[0014] Figure 7The figure is a relationship diagram between BER and OSNR of the algorithm of the present invention and the comparative algorithm in a simulation system. DETAILED DESCRIPTION

[0015] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0016] A carrier frequency offset estimation method for dual-polarization coherent optical communication, the method comprising: a dual-polarization coherent optical communication system comprising a transmitting end, an optical fiber channel and a receiving end; comprising: performing DSP processing on original data at the transmitting end, and using an IQ modulator to modulate the data after DSP processing to generate a dual-polarization coherent optical signal; performing power compensation on the dual-polarization coherent optical signal in the optical fiber channel and transmitting it to a receiving end; the receiving end receives the signal and performs coherent demodulation on the received signal; performing DSP processing on the coherently demodulated signal to obtain a restored signal.

[0017] The DSP processing of the coherently demodulated signal at the receiving end includes: compensating and optimizing the coherently demodulated signal to obtain an optimized signal; calculating a training sequence in the optimized signal; estimating the carrier frequency offset of the optimized signal according to the training sequence; compensating the signal for frequency offset according to the carrier frequency offset estimation result, and performing phase offset compensation and data recovery according to the result of the frequency offset compensation.

[0018] Figure 1 This is a principle structure diagram of a carrier frequency offset estimation algorithm based on a training sequence according to the present invention. The algorithm structure consists of a training sequence positioning module and a frequency offset estimation module.

[0019] Dual-polarization coherent optical communication includes: a transmitter, an optical fiber channel, and a receiver. At the transmitter, the original data is processed by the transmitter digital signal processor (DSP) and a dual-polarization coherent optical signal is generated by an IQ modulator provided by a laser. In the optical fiber channel, the signal is transmitted through a long-distance single-mode optical fiber and an erbium-doped fiber amplifier (EDFA) is used for power compensation. At the receiver, a coherent receiver is used to perform coherent demodulation through a local oscillator laser and a photodetector, and then the data is processed by the receiver DSP to restore the original data.

[0020] The training sequence used for frequency offset estimation needs to meet the requirements that the modulation phase cannot be too large, and the real part or imaginary part of the training sequence cannot remain unchanged for a long time; the total length of the training sequence is 256 symbols, the second symbol and the 255th symbol sequence are 1000, which are used for the coarse positioning and fine positioning algorithms of the training sequence; other training sequences used for carrier frequency offset estimation are composed of 1111 and 1010.

[0021] Calculating the training sequence based on the received signal includes: using a coarse positioning algorithm to process the received signal to obtain a starting position range of the training sequence; using a fine positioning algorithm to accurately locate the starting position range obtained by the coarse positioning algorithm to obtain a starting position of the training sequence; and obtaining a complete training sequence based on the starting position and the set length of the training sequence.

[0022] The goal of the algorithm is to calculate the range of the starting position of the training sequence through a coarse positioning algorithm, including: setting the window size to 256 symbols, the window continuously moves backward from the first bit of the received data, and taking the sum of the absolute values ​​of the differences before and after the phase values ​​of all symbols in the window are fitted as the judgment criterion; the specific process of the algorithm includes the following steps:

[0023] S01: Obtain complex data of X polarization state from the received data;

[0024] S02: Set the maximum value of the window start position sequence number to IndexRange_X, that is, the value range of the window start position is 1 to IndexRange_X;

[0025] S03: starting from the 1st bit to the IndexRange_Xth bit of the received data, the following 256 bits of complex data are sequentially obtained;

[0026] S04: Take the argument of the complex data in the window to obtain the corresponding phase value;

[0027] S05: performing phase unwrapping on the result obtained in S04;

[0028] S06: Subtract the phase value of the local training sequence from the result of S05 and store it in the CoarsePhaseDifference_X array;

[0029] S07: Perform a first-order linear fit on the data in the CoarsePhaseDifference_X array based on the least square method to obtain a fitted straight line expression;

[0030] S08: Calculate the absolute value of the difference between each data in the CoarsePhaseDifference_X array and its corresponding fitted data;

[0031] S09: Execute S08 for all data in the window, and sum all the results obtained in S08 and store them in the array WindowAbsDiffSum_X;

[0032] S10: Repeat S04 to S09 until the window start position reaches IndexRange_X;

[0033] S11: Find the minimum value of the data in the array WindowAbsDiffSum_X, and the index value corresponding to the minimum value is the positioning value of the rough estimation of the X polarization state, Index_Rough_X;

[0034] S12: Obtaining complex data of the Y polarization state from the received data;

[0035] S13: Set the maximum value of the window start position sequence number to IndexRange_Y, that is, the value range of the window start position is 1 to IndexRange_Y;

[0036] S14: starting from the 1st bit to the IndexRange_Yth bit of the received data, the following 256 bits of complex data are sequentially obtained;

[0037] S15: Take the argument of the complex data in the window to obtain the corresponding phase value;

[0038] S16: performing phase unwrapping on the result obtained in S15;

[0039] S17: Subtract the phase value of the local training sequence from the result of S16 and store it in the CoarsePhaseDifference_Y array;

[0040] S18: Perform the first-order linear fitting based on the least square method similar to S07 on the data in the CoarsePhaseDifference_Y array to obtain the fitted straight line expression;

[0041] S19: Calculate the absolute value of the difference between each data in the CoarsePhaseDifference_Y array and its corresponding fitted data;

[0042] S20: Execute S19 for all data in the window, and sum all the results obtained by S19 and store them in the array WindowAbsDiffSum_Y;

[0043] S21: Repeat S15 to S20 until the window start position reaches IndexRange_Y;

[0044] S22: Find the minimum value of the data in the array WindowAbsDiffSum_Y, and the index value corresponding to the minimum value is the positioning value of the rough estimation of the Y polarization state, Index_Rough_Y.

[0045] In this embodiment, a first-order linear fitting algorithm is performed based on the least square method, including: assuming that the data before fitting is (x1, y1)…(x i ,y i )…(x m ,y m ), x i and i are the serial number of the PhaseDifference_X array and the value in the array corresponding to the serial number, and m is the length of the data to be fitted; the expression of the fitted straight line is y=kx+b, where k and b are the slope and intercept of the fitted straight line, respectively, and the corresponding slope k expression is:

[0046]

[0047] The intercept b expression corresponding to the fitted straight line is:

[0048]

[0049] The first-order linear fitting algorithm based on the least squares method includes:

[0050] S0701: Calculate the sum of the serial number values ​​of the data array to be fitted and store it in the variable sum_x;

[0051] S0702: Calculate the sum of the values ​​in all arrays of the data array to be fitted and store it in the variable sum_y;

[0052] S0703: Calculate the sum of the product of the sequence number and the corresponding value in the data array to be fitted, and store it in the variable sum_xy;

[0053] S0704: Calculate the sum of the square values ​​of the sequence number values ​​of the data array to be fitted and store it in the variable sum_x_squared;

[0054] S0705: Calculate the length of the data array to be fitted and assign it to the variable m;

[0055] S0706: multiply the result m of S0705 by the result sum_xy of S0703, and then subtract the value of the product of the result sum_x of S0701 and the result sum_y of S0702, and store it in the variable Slope_Numerator;

[0056] S0707: Subtract the squared value of the result sum_x of S0701 from the value obtained by multiplying the result m of S0705 by the result sum_x_squared of S0704, and store it in the variable Slope_Denominator;

[0057] S0708: Divide the value of Slope_Numerator by the value of Slope_Denominator to obtain the slope value of the fitted straight line, which is stored in the variable k;

[0058] S0709: Divide the result sum_y of S0702 by the result m of S0705, and then subtract the result k of S0708 multiplied by the result sum_x of S0701 divided by the result m of S0705 to obtain the intercept value of the fitted straight line, which is stored in the variable b; return the fitted straight line expression y=kx+b.

[0059] The number of the window start position is reduced to the area near Index_Rough_X (or Index_Rough_Y) obtained by rough positioning, and the difference in the serial number values ​​corresponding to the data with two consecutive phase mutations in the window is detected. When the difference is 254, the start position of the window is the accurate start position of the training sequence; the specific process of the fine positioning algorithm includes the following steps:

[0060] S01: Input the parameters required by the algorithm, including the rough estimated positioning value, Index_Rough_X and Index_Rough_Y;

[0061] S02: Set the fluctuation size StartPosVariation_X of the starting position of the window compared to Index_Rough_X, that is, the value range of the starting position of the window is from Index_Rough_X-StartPosVariation_X to Index_Rough_X+StartPosVariation_X;

[0062] S03: Take the received X-polarization state complex data with a length of 256 from the starting position of the window;

[0063] S04: Take the argument of the complex data in the window to obtain the corresponding phase value;

[0064] S05: performing phase unwrapping on the result obtained in S04;

[0065] S06: Subtract the phase value of the local training sequence from the result of S05 and store it in the FinePhaseDifference_X array;

[0066] S07: Perform a first-order linear fit on the data in the FinePhaseDifference_X array based on the least square method to obtain a fitted straight line expression;

[0067] S08: Calculate the distance from each point in the window to the fitted line and store it in the Distance_X array;

[0068] S09: Check the data in the array Distance_X in turn. If the data is greater than 1.2, store the corresponding serial number in the array Point_index_X.

[0069] S10: Subtract the value of the first element from the value of the second element in the Point_index_X array and add 1, and store it in the variable L_X;

[0070] S11: If the value of L_X is equal to 254, the exact starting position Index_X of the training sequence in the X polarization state is equal to the starting position of the window at this time; otherwise, repeat S03 to S10;

[0071] S12: setting the fluctuation size StartPosVariation_Y of the starting position of the window compared to Index_Rough_Y, that is, the value range of the starting position of the window is from Index_Rough_Y-StartPosVariation_Y to Index_Rough_Y+StartPosVariation_Y;

[0072] S13: Take the received Y polarization state complex data with a length of 256 from the start position of the window;

[0073] S14: taking the argument of the complex data in the window to obtain the corresponding phase value;

[0074] S15: performing phase unwrapping on the result obtained in S14;

[0075] S16: Subtract the phase value of the local training sequence from the result of S15 and store it in the FinePhaseDifference_Y array;

[0076] S17: Perform a first-order linear fit on the data in the FinePhaseDifference_Y array based on the least square method to obtain a fitted straight line expression;

[0077] S18: Calculate the distance from each point in the window to the fitted line and store it in the Distance_Y array;

[0078] S19: Check the data in the array Distance_Y in turn. If the data is greater than 1.2, store the corresponding serial number in the array Point_index_Y.

[0079] S20: Subtract the value of the first element from the value of the second element in the Point_index_Y array and add 1, and store it in the variable L_Y;

[0080] S21: If the value of L_Y is equal to 254, the exact starting position Index_Y of the training sequence in the Y polarization state is equal to the starting position of the window at this time; otherwise, S13 to S20 are repeated.

[0081] The phase of the signal can be expressed as:

[0082] R k =exp{j(θ s (k)+ΔωkT+θ L (k)+θ n (k))}

[0083] Among them, θ s (k) represents the modulation phase of the signal with sequence number k. L (k) is the phase noise caused by the laser line width. Since the change of phase noise is slow compared to the high-speed symbol, the phase noise between adjacent symbols can be assumed to be unchanged. The ASE noise of the amplifier is expressed as θ n (k), and the mean ASE noise phase is 0. ΔωkT is the phase error caused by the carrier frequency offset to the kth signal, and ΔωT is the carrier frequency deviation value.

[0084] Assume RS k represents the received kth training symbol, RS k The phase of the signal can be expressed by the phase formula of the signal; TS k Represents the original k-th training symbol of the transmitter. TS k Phase:

[0085] TS k =exp{j(θ s (k))}

[0086] Figure 1 At point a, the received training sequence symbol RS k and local training sequence symbol TS k Take the phase angle. The phase of the received training sequence symbol minus the phase of the local training sequence symbol, remove the modulation phase θ s (k), this process is completed at point b.

[0087]

[0088] The phase of the training sequence symbol is subtracted from the modulation phase as shown above. The phase is fitted by the least square method in a first-order linear manner, and the estimated value of the carrier frequency offset can be obtained. This process is completed at point c.

[0089] In this embodiment, after the phase value of the local training sequence is subtracted from the phase value of the training sequence, a first-order linear fitting algorithm based on the least square method is used to obtain a fitting straight line, and the slope of the fitting straight line is divided by 2π and then multiplied by the symbol rate to obtain the carrier frequency offset estimation value; the specific process of the algorithm includes the following steps:

[0090] S01: Input the parameters required by the algorithm, including: obtaining the accurate starting position Index_X and Index_Y of the training sequence;

[0091] S02: fetch 256-bit complex data starting from the Index_X position of the received X polarization state data;

[0092] S03: Take the phase angle of the result of S02 to obtain the phase value corresponding to the received training sequence;

[0093] S04: performing phase unwrapping on the result obtained in S03;

[0094] S05: Subtract the phase value of the local training sequence from the result of S04 and store it in the TrainPhaseDifference_X array;

[0095] S06: Perform a first-order linear fit on the data in the TrainPhaseDifference_X array based on the least square method to obtain the slope Xfe of the fitted straight line;

[0096] S07: Divide Xfe obtained in S06 by 2π and then multiply by the corresponding symbol rate to obtain the estimated carrier frequency offset of the X polarization state, EstFreqOffset_X;

[0097] S08: fetch 256-bit complex data starting from the Index_Y position of the received Y polarization state data;

[0098] S09: taking the phase angle of the result of S08 to obtain the phase value corresponding to the received training sequence;

[0099] S10: performing phase unwrapping on the result obtained in S09;

[0100] S11: Subtract the phase value of the local training sequence from the result of S10 and store it in the TrainPhaseDifference_Y array;

[0101] S12: Perform a first-order linear fit on the data in the TrainPhaseDifference_Y array based on the least square method to obtain the slope Yfe of the fitted straight line;

[0102] S13: Divide Yfe obtained in S12 by 2π and then multiply by the corresponding symbol rate to obtain the estimated value of the carrier frequency offset of the Y polarization state, EstFreqOffset_Y.

[0103] In this embodiment, the present invention is verified by a specific example, the total number of bits to be transmitted is set to 524288, the transmission rate is 200Gbit / s, the modulation format of the data is 16QAM, dual polarization is used for transmission, and the length of the training sequence for positioning and frequency offset estimation algorithm is set to 256 symbols. And the maximum values ​​of the window start position sequence number IndexRange_X and IndexRange_Y in the coarse positioning algorithm of the present invention are set to 200; the fluctuation size StartPosVariation_X and StartPosVariation_Y of the window start position in the fine positioning algorithm of the present invention compared with the coarse positioning result are set to 4. Figure 2 is the structural block diagram of the simulation system; Table 1 is the specific parameters of the simulation system.

[0104] Table 1 Parameters of simulation system based on training sequence positioning and frequency offset estimation

[0105]

[0106] Figure 3 is the result of the rough positioning algorithm for the training sequence. Figure 3 As shown, the phase peaks appear at the 1st symbol, the 254th symbol, and the 256th symbol. Compared with the phase of the local training sequence, it can be seen that the starting position of the training sequence obtained by the coarse positioning algorithm is moved forward by one position compared with the accurate starting position.

[0107] Figure 4 for Figure 3 The result of the fine positioning algorithm after the coarse positioning. Figure 4 As shown, peaks appear at the 2nd symbol and the 255th symbol, which are consistent with the positions where the local training sequence peaks appear. Therefore, the coarse positioning and fine positioning algorithms in the algorithm of the present invention can obtain the accurate starting position of the training sequence, realizing the function of training sequence positioning.

[0108] Figure 5In the frequency offset estimation algorithm of the algorithm of the present invention, the phase of the training sequence minus the local phase is subjected to a first-order linear fitting result based on the least squares method. It can be seen that the phase value of the training sequence is a straight line after the first-order linear fitting, and the slope value of the straight line divided by 2π and then multiplied by the corresponding symbol rate is the estimated value of the carrier frequency offset.

[0109] In order to illustrate the advantages of the algorithm of the present invention in this embodiment, a frequency offset estimation algorithm based on quartic FFT is used for comparison. The number of points used for FFT estimation in this algorithm is 4096, and this algorithm is called the comparison algorithm.

[0110] FIG6 shows the specific performance of the two algorithms in terms of the accuracy and range of estimating the carrier frequency deviation. It can be seen that when the carrier frequency deviation is 1 GHz to 6 GHz, the accuracy of the algorithm of the present invention is better than that of the comparison algorithm. In terms of the estimable range, the simulation results given in FIG6 (a) and (b) show that when the carrier frequency deviation is greater than 6 GHz, the estimated frequency deviation value of the comparison algorithm is significantly different from the actual frequency deviation value; while the algorithm of the present invention can still accurately estimate the carrier frequency deviation value when it is greater than 6 GHz, and through simulation tests, the estimation range of the algorithm of the present invention is -17 GHz to 17 GHz.

[0111] Figure 7 The graphs of the bit error rate (BER) and the injected optical signal-to-noise ratio (OSNR) of two different frequency offset estimation algorithms under X polarization and Y polarization in simulation are shown. It can be seen that the algorithm of the present invention is better than the comparative algorithm at 3.8×10 -3 At the HD-FEC hard decision threshold, an OSNR improvement of about 0.3dB can be achieved.

[0112] Table 2 shows the computational complexity of the two algorithms. In the comparison algorithm, a single symbol requires 1.5 complex multiplications for a fourth power operation, and an N-point FFT calculation requires (N / 2)log2N complex multiplications. Where N is the number of symbols used for the fourth power FFT to estimate the frequency offset, which is set to 4096 in the simulation; N tr is the length of the training sequence used for positioning and frequency offset estimation, which is set to 256 in the simulation. It can be calculated that the comparison algorithm requires 159744 real addition calculations and 122880 real multiplication calculations; while the algorithm of the present invention only requires 2297 real addition calculations, 2570 real multiplication calculations and 256 phase angle operations. This shows that the computational complexity of the algorithm of the present invention is reduced by about 98.1% compared with the comparison algorithm.

[0113] Table 2 Comparison of computational complexity

[0114]

[0115] In summary, the algorithm for positioning based on training sequence and carrier frequency offset estimation for dual-polarization coherent optical communication proposed by the present invention can accurately position the training sequence, and its computational complexity is lower than that of other algorithms. This illustrates the advantages of the algorithm of the present invention and also provides a selection basis for frequency offset estimation in coherent optical communication.

[0116] The above embodiments further illustrate the purpose, technical solutions and advantages of the present invention in detail. It should be understood that the above embodiments are only preferred implementation modes of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made to the present invention within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A carrier frequency offset estimation method for dual-polarization coherent optical communication, wherein the dual-polarization coherent optical communication system comprises a transmitting end, an optical fiber channel and a receiving end, wherein: include: At the transmitting end, the original data is processed by DSP, and the data after DSP processing is modulated by IQ modulator to generate dual-polarization coherent optical signals; Power compensation is performed on the dual-polarization coherent optical signal in the optical fiber channel and the signal is transmitted to the receiving end; the receiving end receives the signal and performs coherent demodulation on the received signal; Perform DSP processing on the coherently demodulated signal to obtain the restored signal; The DSP processing of the coherently demodulated signal at the receiving end includes: compensating and optimizing the coherently demodulated signal to obtain an optimized signal; calculating a training sequence in the optimized signal; estimating the carrier frequency offset of the optimized signal according to the training sequence; compensating the signal for frequency offset according to the carrier frequency offset estimation result, and performing phase offset compensation and data recovery according to the result of the frequency offset compensation.

2. The method for estimating carrier frequency offset for dual-polarization coherent optical communication according to claim 1, characterized in that: The compensation optimization of the coherently demodulated signal includes performing IQ compensation, dispersion compensation, polarization mode dispersion compensation, matched filtering, clock recovery, and channel equalization processing on the coherently demodulated signal.

3. The method for estimating carrier frequency offset for dual-polarization coherent optical communication according to claim 1, characterized in that: Calculating a training sequence based on a received signal includes: setting a training sequence length; using a coarse positioning algorithm to process the received signal to obtain a starting position range of the training sequence; using a fine positioning algorithm to accurately locate the starting position range obtained by the coarse positioning algorithm to obtain a starting position of the training sequence; and obtaining a complete training sequence based on the starting position and the set length of the training sequence.

4. The method for estimating carrier frequency offset for dual-polarization coherent optical communication according to claim 3, characterized in that: The total length of the training sequence is 256 symbols, wherein the sequences of the second symbol and the 255th symbol are 1000, and the sequences of other symbols are composed of 1111 or 1010.

5. The method for estimating carrier frequency offset for dual-polarization coherent optical communication according to claim 3, characterized in that: The received signal is processed using a coarse positioning algorithm including: Step 1, set the window size to 256 symbols and the maximum value of the window starting position sequence number IndexRange_X; initialize the CoarsePhaseDifference_X array and the array WindowAbsDiffSum_X; obtain the complex data of the X polarization state from the received signal; Step 2: Starting from the first bit of the received data to the IndexRange_X bit, take the following 256 bits of complex data in sequence to obtain the window complex data; Step 3, taking the phase angle of the window complex data to obtain the phase value corresponding to the window complex data; Step 4: perform phase unwrapping on the phase value, subtract the phase unwrapping result from the phase value of the local training sequence, and store the subtraction result in the CoarsePhaseDifference_X array; Step 5, perform a first-order linear fit based on the least squares method on the data in the CoarsePhaseDifference_X array; Step 6, calculate the difference between each data in the CoarsePhaseDifference_X array and its corresponding first-order linear fitting data, and take the absolute value of the difference; Step 7, when all the data in the window have completed the absolute value calculation, all the absolute values ​​are summed up and the summation result is stored in the array WindowAbsDiffSum_X; Step 8. Repeat steps 3 to 7 until the window starts at IndexRange_X. Step 9, finding the minimum value of the data in the array WindowAbsDiffSum_X, and the index value corresponding to the minimum value is the positioning value Index_Rough_X of the rough estimation of the X polarization state; Step 10, set the maximum value of the window start position sequence number to IndexRange_Y, initialize the CoarsePhaseDifference_Y array and the array WindowAbsDiffSum_Y, and obtain the complex data of the Y polarization state from the received data; Step 11: The complex data of the Y polarization state is processed in the same manner as the complex data of the X polarization state, to obtain a positioning value Index_Rough_Y of a rough estimate of the Y polarization state.

6. The method for estimating carrier frequency offset for dual-polarization coherent optical communication according to claim 5, characterized in that: The first-order linear fit of the data based on the least squares method includes: Step 51, calculate the sum of the sequence number values ​​in the fitting data group to obtain sum_x; Step 52, calculate the sum of all data in the fitting data group to obtain sum_y; Step 53, calculate the sum of the products of the sequence numbers and the corresponding values ​​in the data array to be fitted, and obtain sum_xy; Step 54, calculate the sum of the square values ​​of the sequence number values ​​of the data array to be fitted, and obtain sum_x_squared; Step 55, calculating the length m of the data array to be fitted; Step 56, calculate the slope k of the fitted straight line according to the length m, sum_x, sum_y, sum_xy and sum_x_squared; Step 57, calculate the intercept b of the fitting line according to the slope value k, sum_x, sum_y and length m; obtain the fitted straight line expression according to the slope and intercept of the fitting line, and complete the first-order linear fitting based on the least squares method.

7. The method for estimating carrier frequency offset for dual-polarization coherent optical communication according to claim 6, characterized in that: The expression of the slope k of the fitted line is: The intercept b of the fitted line is expressed as: Among them, m is the length of the data array to be fitted, x i is the i-th sequence value in the fitting data set, y i is the i-th data value in the fitted data set, and k is the slope of the fitted line.

8. The method for estimating carrier frequency offset for dual-polarization coherent optical communication according to claim 3, characterized in that: The fine positioning algorithm is used to accurately locate the starting position range obtained by the coarse positioning algorithm, including: S1, initial FinePhaseDifference_X array, Distance_X array, Point_index_X array, set the fluctuation value of the window start position relative to Index_Rough_X; calculate the value range of the initial position of the window according to the roughly estimated positioning value Index_Rough_X and the fluctuation value; S2, taking the received X-polarization state complex data with a length of 256 from the starting position of the window; S3, taking the phase angle of the complex data in the window to obtain the phase value corresponding to the complex data in the window; S4, performing phase unwrapping on the phase value, subtracting the phase unwrapping result from the phase value of the local training sequence, and storing the subtraction result in the FinePhaseDifference_X array; S5. Perform a first-order linear fit on the data in the FinePhaseDifference_X array based on the least square method to obtain a fitting straight line; S6. Calculate the distances from all points in the window to the fitted straight line and store the distances in the Distance_X array; S7, set a threshold, and compare the data in the Distance_X array with the threshold. If the data is greater than the set threshold, store its corresponding serial number in the Point_index_X array; S8. Subtract the value of the first element from the value of the second element in the Point_index_X array and add 1, and store it in the variable L_X; S9, if the value of L_X is equal to 254, then the exact starting position Index_X of the training sequence in the X polarization state is equal to the starting position of the window at this time; otherwise, repeat S2 to S8; S10, initial FinePhaseDifference_Y array, Distance_Y array, Point_index_Y array, set the fluctuation value of the window start position relative to Index_Rough_Y; calculate the value range of the initial position of the window according to the roughly estimated positioning value Index_Rough_Y and the fluctuation value; S11, taking the received Y polarization state complex data with a length of 256 from the starting position of the window; S12. The processing method for the Y polarization state complex data is the same as the processing method for the X polarization state complex data, so as to obtain the accurate starting position Index_Y of the training sequence of the Y polarization state.

9. The method for estimating carrier frequency offset for dual-polarization coherent optical communication according to claim 3, characterized in that: Carrier frequency offset estimation of the optimized signal includes: Step 1, initialize TrainPhaseDifference_X array and TrainPhaseDifference_Y array; Step 2, fetch 256-bit complex data starting from the Index_X position of the received X polarization state data; Step 3, take the phase angle of the 256-bit complex data to obtain the phase value corresponding to the received training sequence; Step 4: Perform phase unwrapping on the phase value, subtract the phase value of the local training sequence from the phase unwrapping result, and store it in the TrainPhaseDifference_X array; Step 5: Perform a first-order linear fit on the data in the TrainPhaseDifference_X array based on the least square method to obtain the slope Xfe of the fitted straight line; Step 6: Xfe is divided by 2π and then multiplied by the corresponding symbol rate to obtain the carrier frequency offset estimate EstFreqOffset_X of the X polarization state; Step 7, fetch 256-bit complex data starting from the Index_Y position of the received Y polarization state data; Step 8, taking the phase angle of the complex data of the Y polarization state data to obtain the phase value corresponding to the received training sequence; Step 9, perform phase unwrapping on the phase value corresponding to the Y polarization state data angle, and subtract the phase value of the local training sequence from the phase unwrapping, and store it in the TrainPhaseDifference_Y array; Step 10, perform a first-order linear fit on the data in the TrainPhaseDifference_Y array based on the least square method to obtain the slope Yfe of the fitted straight line; Step 11: Yfe is divided by 2π and then multiplied by the corresponding symbol rate to obtain the carrier frequency offset estimate of the Y polarization state, EstFreqOffset_Y.

Images