JP5116636B2 - Optical signal receiving method and optical signal receiving apparatus - Google Patents

Description

  The present invention relates to an optical signal receiving method and an optical signal receiving apparatus in optical communication.

Until now, in optical communication, signal delay due to chromatic dispersion caused by an optical fiber as a propagation path has been compensated by using a dispersion compensating fiber having inverse characteristics (see, for example, Non-Patent Document 1).
"Next Generation Ultra-High-Speed Optical Communication Technology -Technical Issues and Overcoming Solutions for Optical Device Development", First Edition, Technical Information Association, Inc., June 27, 2003, p.112-p.118

  However, in the above-described prior art, it is necessary to install an optimum dispersion compensation fiber for each optical communication path, which hinders flexible network design such as path change. In addition, an approach to prevent characteristic degradation due to chromatic dispersion by signal processing such as OFDM (Orthogonal Frequency Division Multiplexing) has been considered. However, chromatic dispersion with a large temporal spread but small temporal variation and a temporal spread. If signal processing is performed at the same time for polarization mode dispersion that is small but has a large time fluctuation, it is necessary to perform high-speed computation that can follow the time fluctuation for signals in a long time domain, resulting in signal quality degradation and There is a problem in that the calculation load increases. That is, in the prior art, signal quality deterioration due to chromatic dispersion and polarization mode dispersion has been a problem in optical communication.

  The present invention has been made in consideration of such circumstances, and its purpose is chromatic dispersion having a large temporal spread but a small temporal variation, and a polarized wave having a small temporal spread but a large temporal variation. An optical signal receiving method and an optical signal receiving apparatus capable of compensating for large temporal fluctuations while reducing the calculation load by performing signal processing on a mode suitable for each mode dispersion independently. It is to provide.

  In order to solve the above-described problems, the present invention provides an optical signal receiving method in optical communication, an analog-digital conversion step for converting a received optical signal into a digital signal, and the received digitally converted signal. A chromatic dispersion compensation step for performing signal processing for compensating the effect of chromatic dispersion according to the transmission distance based on predetermined transmission distance correction information, and a weight calculation for calculating a reception weight using the compensated received signal. A path length determination step for outputting the predetermined transmission distance correction information used in the chromatic dispersion compensation step based on either the reception weight or the channel information used for calculating the reception weight; And a decoding step for performing decoding using the compensated received signal and the received weight. Is an issue receiving method.

  In order to solve the above-described problems, the present invention provides an optical signal receiving method in optical communication, in which a polarization separation step of separating a received optical signal on a plane of polarization and each of the separated optical signals is a digital signal. A wavelength for performing an analog / digital conversion step for converting the signal into a signal, and performing signal processing for compensating the effect of chromatic dispersion in accordance with the transmission distance on each of the digitally converted reception signals based on predetermined transmission distance correction information A dispersion compensation step, a weight calculation step of calculating a reception weight using each of the digitally converted reception signals, a decoding step of performing decoding using the digitally converted reception signals and the reception weights, The reception weight, channel information used for calculating the reception weight, or the decoded signal Wherein based on any of the compensated received signal, an optical signal receiving method which comprises a path length determination step of outputting the predetermined transmission distance correction information used by the chromatic dispersion compensation step.

  According to the present invention, in the above invention, the decoding step feeds back either the decoded signal or reception weight or channel information obtained from the decoded signal, or error information thereof to the weight calculation step, and the weight The calculation step updates the reception weight based on one of the decoded signal fed back from the decoding step, the reception weight or channel information obtained from the decoded signal, and error information thereof. To do.

  According to the present invention, in the above invention, the chromatic dispersion compensation step performs signal processing for compensating for the effect of chromatic dispersion according to a plurality of transmission distances in parallel, and the compensated received signal corresponding to each transmission distance. The weight calculation step calculates a reception weight using the compensated reception signals corresponding to the plurality of transmission distances, and the decoding step uses the transmission distance used for each chromatic dispersion compensation. In this case, decoding is performed using a reception weight and a reception signal corresponding to the received signal, and a decoded signal that can be decoded is output.

  The present invention is characterized in that, in the above invention, the chromatic dispersion compensation step compensates for the effect of chromatic dispersion and removes a signal corresponding to a DC component of the digital signal.

  In order to solve the above-described problems, the present invention provides an optical signal receiving apparatus in optical communication, an analog-digital conversion unit that converts a received optical signal into a digital signal, and the digitally converted received signal On the other hand, based on predetermined transmission distance correction information, a chromatic dispersion compensation unit that performs signal processing for compensating for the effect of chromatic dispersion according to the transmission distance, and a reception weight is calculated using the compensated received signal. Based on either the weight calculation unit and the reception weight or the channel information used to calculate the reception weight, the path length determination for outputting the predetermined transmission distance correction information used by the chromatic dispersion compensation unit And a decoding unit that performs decoding using the compensated reception signal and the reception weight.

  Further, in order to solve the above-described problem, the present invention provides an optical signal receiving apparatus in optical communication, in which a polarization separation unit that separates a received optical signal in a polarization plane, and the separated optical signal An analog / digital converter for converting into a digital signal, and signal processing for compensating the effect of chromatic dispersion according to the transmission distance based on predetermined transmission distance correction information for each of the digitally converted received signals A chromatic dispersion compensation unit for performing, a weight calculation unit for calculating a reception weight using each of the digitally converted reception signals, and a decoding unit for performing decoding using the digitally converted reception signals and the reception weights The reception weight, or channel information used to calculate the reception weight, or the decoded signal and the compensated reception signal. Based on one of an optical signal receiving apparatus, characterized in that it comprises a path length determination unit for outputting a predetermined transmission distance correction information used by the wavelength dispersion compensator.

  According to the present invention, after performing chromatic dispersion compensation in accordance with the transmission distance, the reception weight is calculated, so that the chromatic dispersion having a large temporal spread but a small temporal variation is small and the temporal spread is small. Compensation against large polarization mode dispersion for each time variation is signal-processed independently and in a suitable manner, so that communication that is resistant to large time variations can be realized while reducing the computation load. The advantage is obtained.

  Hereinafter, an embodiment of the present invention will be described with reference to the drawings.

A. First Embodiment First, a first embodiment of the present invention will be described.
FIG. 1 is a block diagram showing a configuration of a receiving apparatus according to the first embodiment of the present invention. In the figure, 101 is an analog / digital conversion unit, 102 is a chromatic dispersion compensation unit, 103 is a weight calculation unit, and 104 is a decoding unit. When the optical signal is received, the analog / digital conversion unit 101 converts the received optical signal into a digital signal and inputs the digital signal to the chromatic dispersion compensation unit 102. At this time, the optical signal may be directly converted into a digital signal, or the optical signal may be converted into an electric signal and then converted into a digital signal. The chromatic dispersion compensation unit 102 removes the influence of chromatic dispersion determined by the transmission distance and the characteristics of the optical fiber from the input signal by signal processing, and supplies the signal to the weight calculation unit 103.

  The weight calculation unit 103 calculates the reception weight using the pilot signal included in the input signal or the characteristics of the modulation method of the input signal. The weight calculation unit 103 supplies the input signal and the calculated reception weight to the decoding circuit 104. The decoding circuit 104 decodes the input signal using the calculated reception weight, and outputs the decoded signal.

  A reception signal received via a transmission line by an optical fiber causes a difference in arrival time due to frequency components due to chromatic dispersion. It is known that this phenomenon can be expressed by the chromatic dispersion value, chromatic dispersion slope value, and optical fiber length of an optical fiber (Reference 1: Govind P. Agrawal, “Nonlinear fiber optics,” Academic press, 2006 p.63. -65, 76-77). The chromatic dispersion compensation unit 102 compensates the chromatic dispersion amount using the transmission distance information of the optical signal and the information on the type of the optical fiber. For example, when expressed as a phase shift in the frequency domain, the phase rotation in the frequency domain caused by chromatic dispersion can be defined by the following equation (1) for a carrier of frequency f.

Here, L is the transmission distance [km], λ is the wavelength [nm], c is the speed of light 3 × 10 ⁻⁷ [km / ps], D is the chromatic dispersion coefficient [ps / nm / km], and D _slope is the dispersion slope. The coefficient [ps / nm ² / km], fc is the frequency of the optical carrier. Since the chromatic dispersion and the dispersion slope can be given specific values depending on the type of the optical fiber, the phase rotation can be estimated if the type of the optical fiber and the distance L are known. In the formula (1), for a fiber that is greatly affected by chromatic dispersion, the chromatic dispersion slope can be ignored by setting D _slope = 0. The chromatic dispersion compensation unit 102 can remove or reduce the influence of the chromatic dispersion expressed by the mathematical formula (1) by multiplying the inverse characteristic.

  Specifically, the inverse characteristic of Equation (1), that is, the complex conjugate of g (f) is changed to the overlap-save method or overlap-add method (reference document 2: John J. Shynk, “Frequency-domain and multirate adaptive filtering”. ”Using Signal Processing Magazine, IEEE, 1992, p.14-37.), The received signal is transformed into the frequency domain by Fourier transform, multiplied by the coefficient in Equation (1), and again inverse Fourier transformed into the time domain. Then, the influence of chromatic dispersion can be eliminated by using a suitable location or performing a convolution operation on the received signal with a coefficient representing the complex conjugate of Equation (1) in time series.

  As described above, if the influence of chromatic dispersion is removed in advance, the influence of the circuit characteristics and polarization mode dispersion is dominant on the signal. Here, the circuit characteristics and the influence due to the polarization mode dispersion are significantly smaller in time spread than the influence due to the chromatic dispersion.

  The signal multiplied by the inverse characteristic of Equation (1) in the chromatic dispersion compensation unit 102 is input to the weight calculation unit 103. The weight calculation unit 103 calculates a reception weight for decoding using a known signal included in the signal and information on the modulation scheme of the transmission signal.

  An example in which a signal transmitted at 10 Gsample / sec is received at (R × 10) Gsample / sec is shown below. Note that R ≧ 1. Here, if the transmission signal is transmission by the single carrier scheme, it is 10 Gbaud, and if transmission by the OFDM scheme is 10 / (K + G) Gbaud (the OFDM symbol is regarded as one modulation).

  K and G are the number of points in the Fourier transform of OFDM and the number of symbols in the guard interval, respectively, and baud is a unit of signal rate, which is the number of modulations in the carrier per second. Assume that a signal having a known signal of s (1) to s (K) and a time length of K × 100 psec is transmitted M times repeatedly. M is a number of 1 or more and does not need to be an integer. Considering the case where the signal has a spread of ± 500 psec due to the influence of circuit characteristics and polarization mode dispersion, the received signal corresponding to the known signal is represented as r (1) to r (R × M × K + R × 10). be able to. Since reception is R × 10 Gsample / sec, one signal is obtained every (100 / R) psec.

  Here, since the chromatic dispersion compensation is performed in the preceding stage, the length of the received signal to be considered is short, and the calculation load is remarkably reduced. The effect of chromatic dispersion can have signal characteristics that are significantly larger than circuit characteristics and signal spread due to polarization mode dispersion. Therefore, a received signal that is significantly longer than the above is stored and calculated as a received signal corresponding to a known signal. There must be.

  The weight calculation unit 103 stores in advance information obtained based on s (1) to s (K). For example, the signals sf (1) to sf (K) obtained by converting the known signals s (1) to s (K) into the frequency domain by Fourier transform, or the known signals s (1) to s (K) are oversampled. (100 / R) Oversampled known signals s ′ (1) to s ′ (R × K) as data every psec, or signal sf ′ (1) obtained by Fourier transforming the oversampled known signals and converting them to the frequency domain ~ Sf ′ (R × K) can be stored each time.

  The receiving unit extracts R × K data of r (x + 1) to r (x + R × K) from the received signals r (1) to r (R × M × K + 20 × R), performs Fourier transform, and performs frequency conversion. Region signals rf (1) to rf (R × K) are obtained. x represents a signal acquisition position. By taking a value larger than the time spread of the propagation path (here, a maximum of 500 psec), x can be made less susceptible to inter-signal interference. Since rf (1) to rf (R × K) are signals after being affected by the propagation channel generated in the transmission path, the propagation channel coefficient hf ′ can be expressed by the following equation (2). .

  When the transmission signal is single carrier transmission, the reception weight in the frequency domain can be calculated by the following equation (3) using this propagation channel coefficient (Reference 3: K. Takeda, F. Adachi). , “Frequency-domain MMSE channel estimation for frequency-domain equalization of DS-CDMA Signals,” IEICE Trans. Commun., Vol.E90-B, No.7, pp.1746-1753, July 2007.).

Here, σ ² is a fractional value of thermal noise, and * represents a complex conjugate.
In the case of a signal based on OFDM, simply calculate wf ′ (i) = hf ′ (i) ^* / | hf ′ (i) | ² and multiply the received signal rf (i) in each frequency band. Thus, the transmitted signal can be estimated as rf (i) · wf ′ (i). Also, only the weight of Kd subcarriers carrying the transmission signal or pilot signal in wf ′ (i) is calculated, or the received signals r (x + 1) to r (x + R × K) are downsampled. Further, rdf (1) to rdf (K) can be calculated by performing Fourier transform on the down-sampled received signals rd (xd + 1) to rd (xd + K). In this case, the propagation channel coefficient can be expressed by Equation (4), and the reception weight can be obtained by setting σ ² to 0 in Equation (3) or Equation (3).

  In this case, only the weight of Kd subcarriers carrying signals or pilot signals can be calculated.

  Alternatively, the reception weight can be calculated by configuring a digital filter using information such as the center frequency, modulation method, and polarization of the received signal without using a known signal (Reference 4: Nobuyoshi Kikuma, "Adaptive signal processing by array antenna", Science and Technology Publishing, pp.17-21, 1998). Alternatively, the reception weight can be calculated by configuring a digital filter using information such as the known signal and the center frequency, modulation method, and polarization of the received signal. The operation principle of the digital filter is, for example, the least square error method (MMSE), the maximum SNR method (MSN), the constrained output power minimization method (CMP), the constant envelope signal algorithm (CMA), etc. Is mentioned.

  The reception weight does not need to be in the frequency domain, and the reception weight in the time domain can be used by the algorithm using the digital filter described above. In the case of the reception weight in the time domain, it can be decoded by performing a convolution operation on the reception signal.

  The decoding unit 104 decodes the signal from the input reception signal and reception weight. If the received reception weight is in the frequency domain, the received signal can be decoded by converting the received signal into the frequency domain by Fourier transform and multiplying the frequency domain signal by the reception weight. In this case, in the decoding using the frequency domain such as the above-described overlap-save or overlap-add, or decoding using the guard interval, Fourier transform is performed by performing Fourier transform at a fixed signal position using the guard interval. It is also possible to perform the decoding without performing the Fourier transform.

  If the input reception weight is in the time domain, the reception weight is convolved with the reception signal to perform decoding. When performing decision feedback, information on the decoded signal and information such as channel error, propagation channel coefficient, and phase error calculated using the decoded signal are supplied to the weight calculation unit 103 to perform weight calculation. The reception weight in the unit 103 can also be updated.

  Further, the chromatic dispersion compensation unit 102 can have compensation weights for a plurality of path lengths. In this case, only the signal sequence compensated with the correct path length is decoded. Distributes received signals, performs chromatic dispersion compensation corresponding to a plurality of transmission distances in parallel, selects a signal sequence that has been correctly compensated in weight calculation unit 103 or decoding unit 104, and outputs a decoding result be able to.

  By performing decoding in this way, it becomes possible to set the time length of the reception weight calculated in the weight calculation unit 103, or the number of points of Fourier transform in the frequency domain, and reduce the calculation load, The following performance can be improved with respect to channel time fluctuation.

Further, the chromatic dispersion compensator 102 can estimate the path length of the optical fiber from the received signal.
For example, among the known signals Sf (1) to Sf (K) in the frequency domain, the signals of the known signals sf (A1) to sf (A2) in a certain section A represented by A1 to A2 are extracted, and the remaining K- (A2− A time domain frequency interval A known signals sA (1) to sA (K) obtained by inserting 0 into the signal of A1 + 1) and performing inverse Fourier transform, as well as the known signal sf (B1) of interval B in the frequency domain Extract the sf (B2) signal, insert 0 into the remaining K− (B2−B1 + 1) signal, and perform inverse Fourier transform to obtain the frequency domain B known signal sB (1) to sB (K) in the time domain. Can be stored in advance, and the correlation between these known signals and received signals can be evaluated by a convolution operation. If the frequency sections A and B are set as different regions, the temporal shift in the position of the obtained correlation peak represents the shift in arrival time due to frequency, that is, the magnitude of chromatic dispersion. The correspondence of the path length L corresponding to this time shift can be stored, and the phase rotation represented by the formula (1) can be corrected.

  Alternatively, the received weight is calculated for the received signal using information such as the center frequency, modulation scheme, and polarization of the received signal according to Reference 4, and the path length is evaluated using the received weight information. You can also However, the reception weights wc (1),..., Wc (T) obtained here are weights that consider chromatic dispersion unlike the above-described reception weights, and the weight size must be set large (T ≧ K). .

In order to evaluate the path length from the reception weight, for example, by selecting a reception weight corresponding to a different frequency section in the reception weight in the frequency domain, inserting 0 in each of the other frequency domains, and performing an inverse Fourier transform. The magnitude of chromatic dispersion can be estimated from the obtained peak position shift between the frequency section A reception weight and the frequency section B reception weight in the time domain, and the corresponding path length L can be determined.
Alternatively, using the fact that the phase condition of the reception weight in the frequency domain has a high correlation with the phase condition represented by Expression (1), the path length L giving the highest correlation with the phase condition of the reception weight in the phase of Expression (1). Can also be selected.

B. Second Embodiment Next, a second embodiment of the present invention will be described.
FIG. 2 is a block diagram showing a configuration of a receiving apparatus according to the second embodiment of the present invention. In the figure, 201 is an analog / digital conversion unit, 202 is a chromatic dispersion compensation unit, 203 is a weight calculation unit, 204 is a decoding unit, and 205 is a path length determination unit. When the optical signal is received, the analog / digital conversion unit 201 converts the optical signal into a digital signal and inputs the digital signal to the chromatic dispersion compensation unit 202. At this time, the optical signal may be directly converted into a digital signal, or the optical signal may be converted into an electric signal and then converted into a digital signal.

  After being converted to a digital signal, the chromatic dispersion compensator 202 removes the influence of chromatic dispersion determined by the transmission distance and the characteristics of the optical fiber from the input signal by signal processing, and outputs it to the weight calculator 203. The weight calculation unit 203 calculates a reception weight using the pilot signal included in the input signal or the characteristics of the modulation method of the input signal. The weight calculation unit 203 outputs the input signal and the calculated reception weight to the decoding circuit 204. The decoding circuit 204 performs decoding using the reception weight calculated on the input signal, and outputs the decoded signal.

  The chromatic dispersion compensation unit 202 compensates the chromatic dispersion amount using the transmission distance information of the optical signal and the information on the type of the optical fiber. For example, it is expressed as a phase shift in the frequency domain, and the influence of chromatic dispersion is compensated by multiplying by the inverse characteristic of Equation (1).

  A signal that is multiplied or convolved with the inverse characteristic of Equation (1) in the chromatic dispersion compensation unit 202 and compensated for the influence of chromatic dispersion is input to the weight calculation unit 203. The weight calculation unit 203 uses the known signal included in the signal and information on the modulation scheme of the transmission signal to calculate a reception weight for decoding using Equation (3) or an adaptive digital filter, The compensated received signal is output to the decoding unit 204, and the reception weight, information on the propagation channel used to obtain the received weight, and some or any of the compensated received signal are path lengths. It supplies to the determination part 205.

  The decoding unit 204 decodes a signal from the input reception signal and reception weight. If the input reception weight is in the frequency domain, the received signal is converted to the frequency domain by Fourier transform, and the frequency domain signal is multiplied by the reception weight, and the input reception weight is in the time domain. If it is, the reception weight is convolved with the reception signal.

  In addition, when performing decision feedback, the information of the decoded signal and information such as channel error, propagation channel coefficient, and phase error calculated using the decoded signal are supplied to the weight calculation unit 203. You can also.

  The path length determination unit 205 uses the propagation channel coefficient calculated by the weight calculation unit 203, the reception weight, or the information of the received signal that has been subjected to chromatic dispersion compensation to use the residual wavelength included in the received signal that has been compensated for chromatic dispersion. Evaluate the amount of dispersion. In other words, the path length L used in the chromatic dispersion compensation unit 202 is corrected.

  Here, a method for determining whether the path length is estimated to be long or short using the reception weight or the propagation channel coefficient will be described. The influence of chromatic dispersion can be expressed as in Expression (1) in the frequency domain, but the chromatic dispersion slope of the second term on the right side of Expression (1) is smaller than the influence of chromatic dispersion on the first term on the right side. Can be ignored. When the influence of polarization mode dispersion or circuit characteristics is small, and the influence of chromatic dispersion remains, the phase characteristic of the measured propagation channel coefficient is a quadratic function. The phase rotation measured when chromatic dispersion remains can be expressed by the following equation (5).

Here, α = πλ ² D / c, and ΔL is obtained by subtracting the actual path length from the path length L used for correction. Therefore, when D> 0, ΔL <0 when the path length L used is shorter than the actual path length, and ΔL> 0 when the actual path length is better than L. Therefore, if the quadratic function diverges in the negative direction, the path length L used in Equation (1) is short, so it is necessary to increase ΔL and substitute it into Equation (1). If it is a shape that diverges into λ, it is possible to improve the compensation effect of chromatic dispersion by shortening by ΔL and substituting it into Equation (1).

  As a method of knowing the sign of ΔL, for example, a change in phase gradient with respect to frequency can be used. The phase channel coefficients θf (1) to θf (K) obtained by eliminating the amplitude information from the propagation channel coefficients hf (1) to hf (K) are defined as the following equation (6).

  When the slope of the phase channel coefficient with respect to the frequency increases from a low frequency region toward a high frequency region, ΔL <0, and when it decreases, ΔL> 0. The phase gradient Φ can be calculated, for example, by the following equation (7).

  Here, angle (A) is a function for extracting an angle from a real axis or an imaginary axis on the complex plane of the complex number A, and arctan, arcsin, arccos, etc. can be used. y is a starting point for obtaining an average value, and Q is the number of samples to be averaged. In equation (7), the angle may be weighted so that the greater the weight near the center portion y + Q / 2 is.

  Whether the phase gradient Φ thus obtained is positive or negative is determined. If Φ is positive, the higher the frequency on the θf frequency axis, the larger the gradient, and ΔL <0, which is used for correction. If the path length is short and Φ is negative, the path length for correction is estimated to be long. The path length determination unit 205 supplies the chromatic dispersion compensation unit 202 with information on the deviation of the path length used in the obtained chromatic dispersion compensation unit 202 from the actual path length.

  The chromatic dispersion compensation unit 202 corrects the path length used in Equation (1) using the input path length deviation information, and outputs a correction coefficient. For example, when Φ is input, a table of ΔL to be corrected is stored in advance for the magnitude of Φ. When −30 ° ≦ Φ <30 °, ΔL = 0, 30 ≦ Φ <90, and ΔL A table such as = 100 [km] is maintained, and a new path length can be used as L = L + ΔL.

  Alternatively, the phase channel coefficients θf (1) to θf (K) can be defined as the following equation (8) using the reception weight.

  Further, the residual effect of chromatic dispersion compensation can be evaluated by dividing the reception weight or channel information by frequency components and converting them into time-series signals.

  An example using the propagation channel coefficient hf (i) will be described. When inverse Fourier transform is performed on the frequency domain information hf (1) to hf (K), a propagation channel coefficient in the time domain is obtained. However, when the frequency domain is limited, a high frequency component and a low frequency are obtained. It is also possible to obtain the difference in arrival time with respect to the component. For example, [hf (1), hf (2), ..., hf (K / 2), 0, 0, ..., 0] and [0, 0, ..., 0, hf (K / 2 + 1), hf ( K / 2 + 2),..., Hf (K)]. Each of K / 2 zero signals is inserted in the second half or the first half instead of the propagation channel coefficient. The peak positions of the powers of the signals ha (1),..., Ha (K) and hb (1),.

  Here, when the peak position a of the low-frequency component propagation channel coefficient ha (i) in the time domain and the peak position b of the high-frequency component propagation channel coefficient hb (i) in the time domain deviate, residual chromatic dispersion exists. To do. If Δt = b−a is positive, L is estimated to be short for an optical fiber with D> 0, and therefore, a positive path length correction is performed such that L = L + | ΔL |. , Δt <0, negative path length correction can be performed as L = L− | ΔL |. About the magnitude | size of (DELTA) L, the table | surface can be previously memorize | stored with respect to (DELTA) t similarly to the case of (PHI), and (DELTA) L can be set according to the magnitude | size of (DELTA) t.

  In addition, using the frequency domain reception weights, for example, [wf (1), wf (2),..., Wf (K / 2), 0, 0,..., 0] and [0, 0,. 0, wf (K / 2 + 1), wf (K / 2 + 2),..., Wf (K)] and performing inverse Fourier transform to evaluate the peak position shift in these time domains as Δt. You can also.

  In addition, it is not always necessary to take half the frequency domain, and the arrival time difference Δt of the signal can be evaluated by performing inverse Fourier transform using a plurality of propagation channel coefficients or reception weights corresponding to different frequency domains.

  Further, the arrival time difference Δt in the frequency domain can be evaluated using the known signal sf (i) or sf ′ (i) stored in advance for the received signal. For example, only signals in different frequency regions of known signals are used, and [sf (1), sf (2),..., Sf (K / 2), 0, 0,..., 0] and [0, 0,. , Sf (K / 2 + 1), sf (K / 2 + 2),..., Sf (K)], and inverse Fourier transforms into signals in the time domain, resulting signals sa (1),. ) And sb (1),..., Sb (K) can be used by performing a convolution operation on the received signal that has received the known signal and defining the shift of the peak position of the correlation value as Δt.

  In addition, ΔL is evaluated by feeding back the information of the signal decoded by the decoding unit 204 and the information such as the channel error, propagation channel coefficient, and phase error calculated using the decoded signal, and performing the weight calculation again. The reception weight estimated by the unit 203 and the propagation channel coefficient can also be used.

Further, the chromatic dispersion compensator 202 can estimate the path length of the optical fiber from the received signal.
For example, the aforementioned known signals sa (1),..., Sa (K) and sb (1),..., Sb (K) are stored, and a convolution operation is performed on the received signal using the complex conjugate of the known signals. The deviation of the peak position of the correlation value is calculated as Δt, and the path length L can be determined from the correspondence with the path length L stored in advance from Δt.

Alternatively, the received weight is calculated for the received signal using information such as the center frequency, modulation scheme, and polarization of the received signal according to Reference 4, and the path length is evaluated using the received weight information. You can also. However, the reception weights wc (1),..., Wc (T) obtained here are weights that consider chromatic dispersion unlike the above-described reception weights, and the weight size must be set large (T ≧ K). .
For example, using the reception weights wc (1),..., Wc (T) in the frequency domain of the obtained reception weights, phase information is calculated as in Equation (8), and Φ obtained from Equation (7) is calculated. Can be used to determine the path length L.

  Alternatively, it is also possible to estimate the path length from the shift Δt of the peak position of the reception weight in the time domain by leaving 0 in the remaining frequency domain while leaving the information of the frequency domain having different reception weights in the frequency domain. For example, [wcf (1), wcf (2), ..., wcf (T / 2), 0,0, ..., 0] and [0,0, ..., 0, wcf (T / 2 + 1), wcf ( T / 2 + 2),..., Wcf (T)] and inverse Fourier transform are performed, so that the peak position shift in these time regions can be evaluated as Δt.

C. Third Embodiment Next, a third embodiment of the present invention will be described.
FIG. 3 is a block diagram showing a configuration of a receiving apparatus according to the third embodiment of the present invention. In the figure, 300 is a polarization separation unit, 301-1 and 301-2 are analog / digital conversion units, 302-1 and 302-2 are chromatic dispersion compensation units, 303 is a weight calculation unit, 304 is a decoding unit, and 305 is A path length determination unit.

  The optical signal is first separated into two orthogonal polarization planes in the polarization separation unit 300. The separated signals are converted into digital signals by the analog / digital converters 301-1 and 301-2, respectively, and output to the chromatic dispersion compensators 302-1 and 302-2. At this time, the optical signal may be directly converted into a digital signal, or the optical signal may be converted into an electric signal and then converted into a digital signal. After being converted into a digital signal, the chromatic dispersion compensators 302-1 and 302-2 remove the influence of chromatic dispersion determined by the transmission distance and the characteristics of the optical fiber from the input signals by signal processing, and perform weight calculation. The data is output to the unit 303.

  The weight calculation unit 303 calculates a reception weight using a pilot signal included in the input signal or a characteristic of the modulation method of the input signal. The weight calculation unit 303 outputs the input signal and the calculated reception weight to the decoding circuit 304. The decoding circuit 304 performs decoding using the reception weight calculated on the input signal, and outputs the decoded signal.

  The chromatic dispersion compensators 302-1 and 302-2 further compensate for the number of wavelengths by using the transmission distance information of the optical signal and the information on the type of the optical fiber. For example, it is expressed as a phase shift in the frequency domain, and the influence of chromatic dispersion is compensated by multiplying by the inverse characteristic of Equation (1).

  In the chromatic dispersion compensation units 302-1 and 302-2, a signal that has been multiplied or convolved with the inverse characteristic of Equation (1) and compensated for the influence of chromatic dispersion is input to the weight calculation unit 303. The weight calculation unit 303 uses the known signal included in the signal and information on the modulation scheme of the transmission signal to calculate the reception weight for decoding using Equation (3) or an adaptive digital filter, and receives and compensates the reception weight. The received reception signal is supplied to the decoding unit 304, and the reception length, the propagation channel information used for obtaining the reception weight, or a part of the compensated reception signal, or any one of them is the path length. The data is output to the determination unit 305.

  The decoding unit 304 decodes the signal from the input reception signal and reception weight. If the input reception weight is in the frequency domain, the received signal is converted to the frequency domain by Fourier transform, and the frequency domain signal is multiplied by the reception weight, and the input reception weight is in the time domain. If it is, the reception weight is convolved with the reception signal.

  In addition, when performing decision feedback, information about the decoded signal and information such as channel error, propagation channel coefficient, and phase error calculated using the decoded signal are supplied to the weight calculation unit 303. You can also.

  As described above, even when there are two inputs, it is possible to decode one or two series of signals transmitted by the digital filter described in Reference 4. As the operation principle of the digital filter, for example, the least square error method (MMSE), the maximum SNR method (MSN), the constrained output power minimization method (CMP), the constant envelope signal algorithm (CKA), and the like can be cited. It is done.

  Alternatively, the reception weight can be calculated as follows using a known signal. Hereinafter, an example in which a signal transmitted at 10 Gsample / sec is received at (R × 10) Gsample / sec will be described. Note that R ≧ 1. Here, if the transmission signal is transmission by the single carrier scheme, it is 10 Gbaud, and if transmission by the OFDM scheme is 10 / (K + G) baud (the OFDM symbol is regarded as one modulation). Note that K and G are the number of points in the Fourier transform of OFDM and the number of symbols in the guard interval, respectively, and baud is a unit of the signal rate, which is the number of modulations in the carrier per second.

  Assume that a signal having a known signal of s (1) to s (K) and a time length of K × 100 psec is transmitted M times repeatedly. M is a number of 1 or more and does not need to be an integer. When transmitting two series of transmission signals at the same time by polarization polarization multiplexing communication, it is possible to transmit known signals alternately in the two series, or perform orthogonal coding so that they can be separated on the receiving side. Signal processing can be performed and transmitted.

  Considering the case where the signal has a spread of ± 500 psec due to the influence of circuit characteristics and polarization mode dispersion, r1 (1) to r1 (R × M × K + R × 20) are received signals corresponding to known signals. R2 (1) to r2 (R × M × K + R × 10). Here, r1 is a received signal corresponding to the signal from the analog / digital converter 301-1, and r2 is a received signal corresponding to the signal from the analog / digital converter 301-2.

  The weight calculation unit 303 stores information obtained based on s (1) to s (K) in advance. For example, the signals sf (1) to sf (K) obtained by converting the known signals s (1) to s (K) into the frequency domain by Fourier transform, or the known signals s (1) to s (K) are oversampled. (100 / R) Oversampled known signals s ′ (1) to s ′ (R × K) as data every psec, or signal sf ′ (1) obtained by Fourier transforming the oversampled known signals into the frequency domain ˜sf ′ (R × K) can be stored.

  In the weight calculation unit 303, r1 (x + 1) to r1 (x + R × K) out of the received signals r1 (1) to r1 (R × M × K + 20) and r2 (1) to r2 (R × M × K + 20), R × K data of r2 (x + 1) to r2 (x + R × K) are respectively extracted, Fourier-transformed, and frequency domain signals rf1 (1) to rf1 (R × K), rf2 (1) to rf2 (R XK) is obtained. x represents a signal acquisition position. By taking a value larger than the time spread of the propagation path (here, a maximum of 500 psec), x can be made less susceptible to inter-signal interference. Since rf (1) to rf (R × K) are signals after being influenced by the propagation channel generated in the transmission path, the propagation channel coefficients hf11 ′ (i) and hf12 ′ ( i) can be obtained by the following mathematical formula (9).

  A channel matrix composed of propagation channel coefficients obtained in this way is defined as the following formula (10).

  The reception weight can be calculated using the channel matrix defined as described above. Alternatively, when two series of signals are transmitted simultaneously using two polarization planes, the channel matrix can be similarly estimated by transmitting the known signal series from the transmission ports corresponding to the two polarization planes. As an example, a channel matrix estimation method when a known signal is transmitted from each polarization plane transmission port at different times is shown.

  R × K data of received signals r1 (x + 1) to r1 (x + R × K) and r2 (x + 1) to r2 (x + R × K) is a received signal corresponding to a known signal transmitted from the first polarization plane. , R1 (y + 1) to r1 (y + R × K), r2 (y + 1) to r2 (y + R × K) R × K data is a received signal corresponding to a known signal transmitted from the second polarization plane The signals when converted into the frequency domain are rf11 (1) to rf11 (R × K), rf12 (1) to rf12 (R × K), and rf21 (1) to rf2 (R × K), rf22, respectively. (1) to rf22 (R × K). Then, the propagation channel coefficients hf11 ′ (i), hf12 ′ (i), hf21 ′ (i), hf22 ′ (i) of the i-th frequency channel can be expressed as the following equation (11), and the channel The matrix is obtained as shown in Equation (12).

  Here, the known signals transmitted from the respective polarization planes are transmitted at different times. However, if the orthogonal signals satisfy the orthogonal condition, the known signals can be transmitted simultaneously from the respective polarization planes.

Further, the reception weight can be determined using the channel matrix obtained by Expression (10) or (12). When the transmission signal is single carrier transmission, using this channel matrix, the reception weight in the frequency domain can be calculated by the following equation (13) using the MMSE weight (Reference 5: Akinori Nakajima, Gargdeep) Sika, Fumiyuki Adachi, “Transmission characteristics of single carrier MIMO multiplexing”, IEICE Technical Report, RCS2004-107, pp.13-18, August 2004). Here, σ ² is a fractional value of thermal noise, and I is a 2 × 2 unit matrix.

  Alternatively, in the case of an OFDM signal, the weight can also be determined using the thermal noise dispersion value independent in each frequency band as the following equation (14).

By multiplying the received signal Rf (i) = (rf1 (i) rf2 (i)) ^T in each frequency band, the transmitted signal can be estimated as Wf ′ (i) Rf (i). Also, only the weight of Kd subcarriers carrying the transmission signal or pilot signal in Wf ′ (i) is calculated, or the reception signal R (x + 1) = (r1 (x + 1) r2 (x + t)) ^T ~ R (x + R * K) = (r1 (x + R * K) r2 (x + R * K)) ^T is downsampled, and Fourier transform is performed on the downsampled received signals Rd (xd + 1) to Rd (xd + K). , Rdf (1) to Rdf (K) can also be calculated. In addition, said W, H, and R represent a matrix, respectively (it represents with bold type in a numerical formula).

  In this case, the propagation channel coefficient can be expressed by the following equation (15), and the reception weight can be obtained from equation (13). Also in this case, only the weight of Kd subcarriers carrying signals or pilot signals can be calculated. The same applies to R = 1 without downsampling.

Here, rdf1 (i) and rdf2 (i) are elements of Rdf (i), and Rdf (i) = ((rdf1 (i) rdf2 (i)) ^T. Alternatively, signals differ in the two polarization planes. When multiplexing and transmitting a sequence, similarly to Equation (13), a received signal corresponding to a known signal transmitted from each polarization plane at a different time is converted into the frequency domain, thereby obtaining the following Equation (16). The propagation channel coefficient can be obtained as shown in FIG.

  The path length determination unit 305 uses the propagation channel coefficient calculated by the weight calculation unit 303, the reception weight, or the information of the received signal that has been subjected to chromatic dispersion compensation to use the residual wavelength included in the received signal that has been compensated for chromatic dispersion. Evaluate the quantity. In other words, the path length L used in the chromatic dispersion compensation units 302-1 and 302-2 is corrected.

  Here, a method for determining whether the path length is estimated to be long or short using the reception weight or the propagation channel coefficient will be described. The influence of chromatic dispersion can be expressed as in Expression (1) in the frequency domain, but the chromatic dispersion slope of the second term on the right side of Expression (1) is smaller than the influence of chromatic dispersion on the first term on the right side. Can be ignored. When the influence of polarization mode dispersion or circuit characteristics is small, and the influence of chromatic dispersion remains, the phase characteristic of the measured propagation channel coefficient is a quadratic function. The phase rotation measured when chromatic dispersion remains can be expressed as Equation (5).

  However, in the third embodiment, two transmission channel coefficients are obtained when transmitted in one sequence, and four propagation channel coefficients are obtained when transmitted in two sequences. From these plural pieces of information, a plurality of equations (6), (7) and a plurality of Δt in the second embodiment can be obtained, and averaging processing, weighting according to the magnitude of received power, and selection are performed. Thus, the accuracy of path length evaluation can be increased. The evaluation of the obtained path length is supplied to the chromatic dispersion compensation units 302-1 and 302-2, and the path length value used for chromatic dispersion compensation can be changed.

  Further, the correction value of the path length can be estimated from the two or four propagation channel coefficients as compared with the phase distribution expressed by Equation (1) due to the influence of chromatic dispersion. For example, by calculating the phase gradient Φ in the second embodiment, averaging this value, adding after weighting, or making a selection, the corresponding path length can be used as the correction value. .

  Further, the chromatic dispersion compensation units 302-1 and 302-2 can have compensation weights for a plurality of path lengths. In this case, the decoding circuit 304 only decodes the signal sequence compensated with the correct path length. It is possible to select a signal sequence that has been correctly compensated in the weight calculation unit 303 or the decoding unit 304 and output a decoding result.

In addition, the chromatic dispersion compensation units 302-1 and 302-2 can estimate the path length of the optical fiber from the received signal.
For example, the known signals sa (1),..., Sa (K) and sb (1),..., Sb (K) described in the second embodiment are stored, and the received signal is used as the received signal using the complex conjugate of the known signals. It is also possible to perform a convolution operation, calculate the shift of the peak position of the correlation value as Δt, and determine the path length L from this Δt by correspondence with the path length L stored in advance. At this time, Δt can be determined by averaging, weighted addition, or selection using two or four Δt, and the path length L can be determined by determining the path length L from this value. The estimation accuracy of can be improved.

Alternatively, the received weight is calculated for the received signal using information such as the center frequency, modulation scheme, and polarization of the received signal according to Reference 4, and the path length is evaluated using the received weight information. You can also. However, the reception weight matrix Wc (1),..., Wc (T) obtained here is a weight matrix considering wavelength dispersion unlike the above-described reception weight W, and it is necessary to set the size of the weight matrix large ( T ≧ K).
For example, using the reception weight matrix Wc (1),..., Wc (T) in the frequency domain of the reception weight obtained, the phase information is calculated as in Equation (8), and the coefficients constituting Wc (i) Can be used to determine the path length L using two or four Φs obtained from equation (7). Also in this case, Φ can be determined by averaging, weighted addition, or selection, and by determining the path length L from this value, the accuracy of estimating the path length can be improved.

  Alternatively, it is also possible to estimate the path length from the shift Δt of the peak position of the reception weight in the time domain by leaving information on the different frequencies of the reception weight in the frequency domain and inserting 0 in the remaining frequency domain. For example, if the coefficient of 1 row and 1 column of Wcf (i) is wcf11 (i), [wcf11 (1), wcf11 (2),..., Wcf11 (T / 2), 0, 0,. , [0, 0,..., 0, wcf11 (T / 2 + 1), wcf11 (T / 2 + 2),..., Wcf11 (T)], and performing inverse Fourier transform to obtain peak positions in these time domains. The deviation can be evaluated as Δt. Further, since this calculation can be performed for wcf12 (i), wcf21 (i), and wcf22 (i), respectively, averaging or weighted addition of up to four Δt, or By performing the selection, it is possible to improve the estimation accuracy of the path length by determining Δt and determining the path length L.

  In the first to third embodiments, the chromatic dispersion compensator can also have a function of removing a direct current component. The DC component can be calculated as follows.

Here, a (k) is a forgetting factor, and the direct current component can be estimated from the reception signal before the X point using the reception signals after the Y point. Further, Y> −X is satisfied, and X and Y can be negative numbers. The DC component can be removed by subtracting Dc (x) thus obtained from r (x). By performing this calculation before frequency synchronization is achieved in the chromatic dispersion compensator, interference power can be removed. In addition, when a frequency shift occurs due to chromatic dispersion compensation, the DC component with interference power can be removed by performing it before chromatic dispersion compensation.
Alternatively, when performing chromatic dispersion compensation in the frequency domain, after converting the received signal to the frequency domain, the component of one or more frequency channels corresponding to the DC component is replaced with 0, and then the received signal in the time domain The direct current component can be removed by converting to.
By doing so, it is possible to prevent the characteristics from deteriorating due to the interference power on the DC component in the subsequent reception weight calculation and decoding, and to improve the communication quality.

  FIG. 4 is a diagram showing a result of plotting the Q value with respect to the distance when the long distance transmission is actually performed using the single mode fiber (SMF) and the signal is received by the method of the present invention. In the experiment, 12.5 Gbaud single carrier transmission was used, and the modulation method QPSK and 7% of all signals were used as known signals. The overlap-save method was used as the chromatic dispersion compensation units 302-1 and 302-2, and chromatic dispersion was compensated by overlapping half of the FFT window size. The FFT window size used is 1024, 512, and 256 samples, and it can be seen that the larger the window size, the less the degradation of characteristics with respect to long-distance transmission and the greater the chromatic dispersion tolerance. The reception weight used in the weight calculation unit 303 in the subsequent stage is 16 samples or 4 samples, the reception weight of 16 samples is multiplied in the frequency domain, and the reception weight of 4 samples is convolved in the time domain.

  From FIG. 4, it can be seen that the number of samples for reception weight can be significantly reduced compared to the sample size used for chromatic dispersion compensation. In the chromatic dispersion compensation, it is possible to simultaneously perform the decoding, but in this case, the decoding is performed using the number of samples exceeding 100, and it is difficult to follow the time variation of the channel. When time changes occur, the characteristics are greatly degraded. If decoding is performed with the configuration of the present invention, the number of received weight samples can be made extremely small, the calculation load for reception weight calculation can be reduced, and the channel time fluctuation can be easily followed. .

  As described above, according to the present invention, it is possible to remove the influence of chromatic dispersion in optical communication by digital calculation processing, and while reducing the calculation load of polarization mode dispersion compensation with time fluctuation, The tolerance to fluctuations can be improved.

It is a block diagram which shows the structural example of the receiver by the 1st Embodiment of this invention. It is a block diagram which shows the structural example of the receiver by the 2nd Embodiment of this invention. It is a block diagram which shows the structural example of the receiver by the 3rd Embodiment of this invention. It is a figure which shows the result of having plotted Q value with respect to distance at the time of performing long-distance transmission using a single mode fiber (SMF) and receiving by the method of this invention.

Explanation of symbols

DESCRIPTION OF SYMBOLS 101 Analog / digital conversion part 102 Wavelength dispersion compensation part 103 Weight calculation part 104 Decoding part 201 Analog / digital conversion part 202 Wavelength dispersion compensation part 203 Weight calculation part 204 Decoding part 205 Path length determination part 300 Polarization separation part 301-1, Reference numeral 301-2 Analog / digital converters 302-1, 302-2 Wavelength dispersion compensation unit 303 Weight calculation unit 304 Decoding unit 305 Path length determination unit

Claims (7)

An optical signal receiving method in optical communication,
An analog-digital conversion step for converting the received optical signal into a digital signal;
A chromatic dispersion compensation step for performing signal processing for compensating the effect of chromatic dispersion according to transmission distance based on predetermined transmission distance correction information for the digitally converted received signal;
A weight calculation step of calculating a reception weight using the compensated reception signal;
Based on either the reception weight or the channel information used to calculate the reception weight, a path length determination step for outputting the predetermined transmission distance correction information used in the chromatic dispersion compensation step;
And a decoding step of performing decoding using the compensated reception signal and the reception weight. An optical signal receiving method in optical communication,
A polarization separation step for separating the received optical signal at the polarization plane;
An analog-to-digital conversion step of converting each of the separated optical signals into a digital signal;
A chromatic dispersion compensation step for performing signal processing for compensating the effect of chromatic dispersion in accordance with the transmission distance based on predetermined transmission distance correction information for each digitally converted reception signal;
A weight calculation step of calculating a reception weight using each of the digitally converted reception signals;
A decoding step of decoding using each of the digitally converted reception signals and the reception weights;
The predetermined transmission distance used in the chromatic dispersion compensation step based on either the reception weight, the channel information used to calculate the reception weight, or the decoded signal and the compensated reception signal And a path length determination step for outputting correction information. The decoding step includes
Either the decoded signal, or the reception weight or channel information obtained from the decoded signal, and the error information thereof are fed back to the weight calculation step.
The weight calculation step includes:
The reception weight is updated based on one of a decoded signal fed back from the decoding step, a reception weight or channel information obtained from the decoded signal, and error information thereof. Alternatively, the optical signal receiving method according to claim 2 . The chromatic dispersion compensation step includes
Perform signal processing to compensate for the effects of chromatic dispersion according to multiple transmission distances in parallel, and output a compensated received signal corresponding to each transmission distance,
The weight calculation step includes:
Each of the received weights is calculated using the compensated received signals corresponding to the plurality of transmission distances,
The decoding step includes
Performs decoding using respective and receiving weight and the received signal corresponding to the transmission distance used for wavelength dispersion compensation, claims 1 to 3 and outputs the decoded signal was decodable The optical signal receiving method according to any one of the above. The chromatic dispersion compensation step includes
As well as compensate for the effects of chromatic dispersion, the optical signal receiving method according to any one of claims 1 to 4, wherein removing the signal corresponding to the DC component of the digital signal. An optical signal receiving device in optical communication,
An analog-digital converter that converts the received optical signal into a digital signal;
A chromatic dispersion compensator that performs signal processing for compensating the effect of chromatic dispersion according to the transmission distance based on predetermined transmission distance correction information for the digitally converted received signal;
A weight calculation unit for calculating a reception weight using the compensated reception signal;
Based on either the reception weight or the channel information used to calculate the reception weight, a path length determination unit that outputs the predetermined transmission distance correction information used in the chromatic dispersion compensation unit;
An optical signal receiving apparatus comprising: a decoding unit that performs decoding using the compensated reception signal and the reception weight. An optical signal receiving device in optical communication,
A polarization separator that separates the received optical signal at the polarization plane;
An analog-to-digital converter that converts the separated optical signals into digital signals, and
A chromatic dispersion compensator that performs signal processing for compensating the effect of chromatic dispersion according to the transmission distance based on predetermined transmission distance correction information for each digitally converted reception signal;
A weight calculation unit for calculating a reception weight using each of the digitally converted reception signals;
A decoding unit that performs decoding using each of the digitally converted reception signals and the reception weight;
The predetermined transmission distance used in the chromatic dispersion compensator based on either the reception weight, channel information used to calculate the reception weight, or the decoded signal and the compensated reception signal An optical signal receiving apparatus comprising: a path length determination unit that outputs correction information.

Images