CN103095614B - Joint equalization and frequency offset estimation device in proruption coherent optical fiber communications - Google Patents

Abstract

The invention discloses a joint equalization and frequency offset estimation device in proruption coherent optical fiber communications. Combined utilization of electric channel balance, frequency offset estimation and frequency offset compensation enables iteration step length needed by channel equalization to be reduced under the condition of the same frequency offset, and the condition that iteration is not restrained is avoided. Simultaneously, in the process of self-adaption update of a balance tap weight coefficient, equilibrium value and decision value which removes frequency offset compensation are adopted by an error signal to calculate, therefore, effect of the frequency offset compensation on accuracy of the error signal is removed, and an equalization module can accurately play a role. The joint equalization and frequency offset estimation device not only can estimate and compensate for proruption offset frequency in a certain range, but also can guarantee convergence of electric channel equalization. A proruption coherent optical fiber communication receiving system which is built by the joint equalization and frequency offset estimation device can keep good performance under the condition of big offset frequency.

Description

A Joint Equalization and Frequency Offset Estimation Device in Burst Coherent Optical Fiber Communication

technical field

The invention belongs to the technical field of optical communication, and more specifically relates to a joint equalization and frequency offset estimation device in burst coherent optical fiber communication.

Background technique

With the advancement of high-speed digital signal processing technology and the improvement of people's communication requirements, high-speed coherent optical communication has become a research hotspot. At present, many coherent optical communication systems are in a continuous working state. Once the system is established, the signal will continue to occur. However, in some communication systems, signals occur intermittently or even suddenly, which is called burst communication. This burst mode is a covert communication that is short-lived and difficult to detect and interfere with.

The coherent detection communication system receiver uses a local oscillator laser to coherent with the received carrier modulation signal to obtain a baseband signal. In theory, the oscillation frequency of the local oscillator laser is required to be exactly the same as the frequency of the signal carrier, but in fact each laser is There is a certain amount of oscillation frequency offset. Assuming that the possible oscillation frequency offset range of each laser is [-X,X]Hz, the relative frequency offset range of the two lasers may be [-2X,2X]Hz, so the carrier Frequency offset is one of the main factors affecting the performance of high-speed coherent optical communication systems. It will largely determine whether signals can be demodulated accurately and information can be completely restored.

Fig. 1 is a schematic diagram of frequency drift of burst communication. In the initial stage of burst communication, frequency instability will occur, and the instability of burst frequency will have a great impact on the judgment of subsequent signals. Therefore, it is necessary to estimate and compensate the burst frequency offset. On the other hand, as the transmission rate increases, the impact of fiber dispersion on the system will be more serious. Therefore, in addition to the estimation and compensation of the burst frequency offset in the high-speed burst coherent optical fiber communication system, channel equalization is also essential. . In the existing continuous coherent optical communication system research, in order to perform frequency offset estimation and compensation, it is assumed that dispersion equalization has been performed in the optical domain, so the influence of dispersion does not need to be considered when performing frequency offset estimation and compensation.

When the existing optical fiber communication system performs dispersion equalization, the tap coefficient of the electrical channel equalizer adopts adaptive adjustment, which not only can realize dispersion compensation, but also can track the phase jitter caused by frequency offset in the received signal to a certain extent, so in When the frequency offset is relatively small, the signal can be correctly demodulated without additional frequency offset compensation. However, as the frequency offset in the system increases, the equalizer iteration step size also needs to be increased to track the signal phase change caused by the frequency offset; however, an excessively large iteration step size will cause the equalizer tap coefficients to be unstable. It even does not converge, resulting in serious degradation of system performance.

Contents of the invention

The purpose of the present invention is to overcome the deficiencies of the prior art, and provide a joint equalization and frequency offset estimation device in burst coherent optical fiber communication. By combining the electrical channel equalization with frequency offset estimation and frequency offset compensation, the frequency offset is the same Under certain conditions, the iteration step size required for channel equalization is reduced, so as to avoid the situation where the iteration does not converge, so that the equalization module can work stably, and at the same time, the system can estimate and compensate for a large range of sudden frequency offsets. can also have better performance.

In order to achieve the above-mentioned purpose of the invention, the joint equalization and frequency offset estimation device in the burst coherent optical fiber communication of the present invention is characterized in that it includes:

An equalization module, for receiving the signal r _k quantized by AD sampling and performing equalization processing, that is, according to the judgment value of the previous signal r _k-1 fed back by the judgment module The cumulative frequency offset θ _{k-1 of the previous signal r k-1 fed} back by the frequency offset estimation module performs iterative update of the equalization tap weight coefficient:

${\mathrm{<span\; class="notranslate">\; C</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; =</span>{\mathrm{<span\; class="notranslate">\; C</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; \&Delta;\&epsiv;</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\mathrm{<span\; class="notranslate">\; r</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}^{<span\; class="notranslate">\; *</span>}$

Among them, C _k and C _k-1 are the equalization tap weight coefficients when the kth and k-1th signals pass through the equalization module respectively; r _k-1 is the received signal sample value of the k-1th signal, is the conjugate value of the signal r _k-1 , Δ is the iteration step size, which is a positive number, and is set by the user, ε _k-1 is the error signal obtained from the k-1th signal, and its calculation formula is:

${\mathrm{<span\; class="notranslate">\; \&epsiv;</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; =</span>{\stackrel{<span\; class="notranslate">\; ~</span>}{\mathrm{<span\; class="notranslate">\; m</span>}}}_{\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; -</span>{\stackrel{<span\; class="notranslate">\; ~</span>}{\mathrm{<span\; class="notranslate">\; m</span>}}}_{\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; \&CenterDot;</span>{\mathrm{<span\; class="notranslate">\; e</span>}}^{{\mathrm{<span\; class="notranslate">\; j\&theta;</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}}$

in, is the equalized value of the k-1th signal after passing through the equalized module, is the decision module pair The decision value of the signal obtained after frequency offset compensation, θ _k-1 represents the cumulative frequency offset obtained by the k-1th signal through the frequency offset estimation module;

The equalization module equalizes the signal r _k according to the equalization tap weight coefficient C _k , and outputs the equalized value Send to the judgment module and the frequency offset module;

A judgment module, receiving the equalization value And calculate the equalization value according to the cumulative frequency offset θ _k-1 obtained by the k-1th signal through the frequency offset estimation module The frequency offset compensation signal And make a judgment, output the judgment value Send to the frequency offset estimation module and feed back to the equalization module;

A frequency offset estimation module, receiving the equalization value transmitted by the equalization module and the judgment value transmitted by the judgment module Equilibrium value The signal phase is expressed as φ _d,k +kΔωT+φ _n,k , where Represents the modulation phase, kΔωT represents the phase error introduced by the frequency offset, is the phase error caused by Gaussian noise, and the decision value The signal phase is Compute the cumulative phase error estimate z_tmp _k for the first k signals:

$\mathrm{<span\; class="notranslate">\; z</span>}<span\; class="notranslate">\; \_</span>\mathrm{<span\; class="notranslate">\; t</span>}{\mathrm{<span\; class="notranslate">\; mp</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; =</span><span\; class="notranslate">\; (</span>{\stackrel{<span\; class="notranslate">\; ~</span>}{\mathrm{<span\; class="notranslate">\; m</span>}}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; \&Center\; Dot;</span>{{\stackrel{<span\; class="notranslate">\; ^</span>}{\mathrm{<span\; class="notranslate">\; m</span>}}}_{\mathrm{<span\; class="notranslate">\; k</span>}}}^{<span\; class="notranslate">\; *</span>}<span\; class="notranslate">\; )</span>$

$<span\; class="notranslate">\; =</span>{\mathrm{<span\; class="notranslate">\; \&phi;</span>}}_{\mathrm{<span\; class="notranslate">\; d</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; k\&Delta;\&omega;T</span>}<span\; class="notranslate">\; +</span>{\mathrm{<span\; class="notranslate">\; \&phi;</span>}}_{\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; \&phi;</span>}}_{\mathrm{<span\; class="notranslate">\; d</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; k</span>}}$

$<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; k\&Delta;\&omega;T</span>}<span\; class="notranslate">\; +</span>{\mathrm{<span\; class="notranslate">\; \&phi;</span>}}_{\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; k</span>}}$

Compute the phase error estimate α _k for the kth signal:

α _k = z_tmp _k ·(z_tmp_delay _k-1 ) ^*

=kΔωT+φ _n,k -[(k-1)ΔωT+φ _n,k-1 ]

=ΔωT+φ _n,k -φ _n,k-1

Among them, z_tmp_delay _k-1 is obtained by delaying one beat of the cumulative phase error estimate z_tmp _k-1 of the previous k-1 signals;

Perform an angle calculation operation on the phase error estimate α _k to obtain the angle value β _k , if the angle value β _k is within the range of [-π, π], then send the phase error estimate α _k to the filter to suppress the noise phase, otherwise the phase The error estimate α _k is discarded; the frequency offset estimate ΔωT is obtained after the noise phase is suppressed by the filter, and the cumulative frequency offset θ _k =θ _k-1 +ΔωT of the first k signals is calculated;

The frequency offset estimation module transmits and feeds back the accumulated frequency offset θ _k to the equalization module and the judgment module respectively, for the equalization processing and judgment of the next signal.

Among them, the equalization module is an adaptive equalizer based on the LMS algorithm, including a series of FIR filters.

Further, the frequency offset estimation module includes a memory for storing the cumulative phase error estimated value z_tmp _k , first take out z_tmp _k-1 and perform a one-beat delay processing to obtain z_tmp_delay _k-1 to synchronize with z_tmp _k , and then write into z_tmp _k signal.

Wherein, the initial values of the equalization tap weight coefficient C _k-1 , error signal ε _k-1 , cumulative phase error estimate z_tmp _k-1 , and cumulative frequency offset θ _k-1 are obtained by initializing the training sequence, and the initialization process for:

The length of the training sequence pilot _k is L. L is set by the user and must be greater than the length l of the sliding window of the filter in the frequency offset estimation module. Considering the requirement for the number of divergent symbols of the burst frequency offset, L must be long enough to ensure Make the equalization module converge; the training sequence pilot _k obtains the signal sequence R _k after being processed by analog interference;

The equalization module receives the signal sequence R _k , and the equalization tap weight coefficient is always C _k =1; the signal R ₁ generates the first equalization value through the equalization module Signal R ₂ passes through the equalization module to generate the second equalized value Delay z_tmp ₁ by one beat to generate z_tmp_delay ₁ , phase error estimation value α ₂ =z_tmp ₂ ·(z_tmp_delay ₁ ) ^* , take the angle of α ₂ to get the first angle value β ₂ ; repeat the signal sequence R _k with the second The signals are processed in the same way until the signal R _l+1 , the angle value is accumulated to l, and the l angle values enter the filter to suppress the noise phase together to obtain the first frequency offset estimation value θ _l+1 , and the error signal is obtained ${\mathrm{\&epsiv;}}_{l+1}={\tilde{m}}_{l+1}-{\mathrm{pilot}}_{l+1}\&Center\; Dot;e^{{\mathrm{j\&theta;}}_{l+1}};$

The equalization tap weight coefficient starts to iteratively update C _l+2 =C _l+1 -Δε _l+1 R _l+1 ^* =1-Δε _l+1 R _l+1 ^* , and the signal R _l+2 is equalized by the equalization module value Perform frequency offset estimation to obtain the first cumulative frequency offset θ _l+2 . Then, the iteratively updated equalization tap weight coefficient is used to perform joint equalization and frequency offset estimation on the signal sequence R _k , where C _k =C _k-1 -Δε _k-1 R _k-1 ^* , The initialization of the equalization tap weight coefficient is completed until the end of the training sequence. At this time, the equalization module converges, and the current C _L , ε _L , z_tmp _L , and θ _L are used as the equalization tap weight coefficient C _k-1 and the error signal ε _{k- 1.} Initial values of the cumulative phase error estimate z_tmp _k-1 and the cumulative frequency offset θ _k-1 .

The purpose of the invention of the present invention is achieved in this way: in the self-adaptive update process of the equalization tap weight coefficient, the error signal is calculated by using the equalization value and the judgment value of removing the frequency offset compensation, thereby eliminating the frequency offset compensation's influence on the error signal The impact of accuracy enables the equalization module to function accurately.

The joint equalization and frequency offset estimation device in the burst coherent optical fiber communication of the present invention can not only estimate and compensate the burst frequency offset within a certain range, but also ensure the convergence of the electrical channel equalization, so that the joint equalization and frequency offset estimation device of the present invention can The burst coherent optical fiber communication receiving system formed by the offset estimating device can also maintain good performance when there is a burst frequency offset or the frequency offset is large.

Description of drawings

Fig. 1 is a schematic diagram of frequency drift of burst communication;

Fig. 2 is a schematic structural diagram of the joint equalization and frequency offset estimation device in the burst coherent optical fiber communication of the present invention applied to the burst coherent optical fiber communication system;

FIG. 3 is a structural diagram of a specific embodiment of the joint equalization and frequency offset estimation device in the burst coherent optical fiber communication of the present invention;

Fig. 4 is a diagram of a burst frequency offset simulation result of a specific embodiment of the joint equalization and frequency offset estimation device in the burst coherent optical fiber communication of the present invention;

Fig. 5 is a simulation result diagram of the symbol error rate of the burst coherent optical fiber communication receiving system using the joint equalization and frequency offset estimation device in the burst coherent optical fiber communication of the present invention.

Detailed ways

Specific embodiments of the present invention will be described below in conjunction with the accompanying drawings, so that those skilled in the art can better understand the present invention. It should be noted that in the following description, when detailed descriptions of known functions and designs may dilute the main content of the present invention, these descriptions will be omitted here.

Fig. 2 is a schematic structural diagram of the joint equalization and frequency offset estimation device in the burst coherent optical fiber communication of the present invention applied to the burst coherent optical fiber communication system. As shown in Figure 2, at the transmitting end, differential quadrature phase shift keying (QPSK) modulation is performed on two optical signals whose polarization directions are perpendicular to each other, and the polarization coupler implements polarization multiplexing of the two optical signals to obtain PDM -QPSK transmission signal, when the signal is transmitted on a single-mode fiber, effects such as chromatic dispersion and polarization mode dispersion will cause the broadening of the optical pulse and form inter-symbol interference.

At the receiving end, the polarization multiplexed signal is split into two polarization directions by a polarization beam splitter for independent coherent demodulation. The 4 electrical signals after coherent demodulation need to be AD sampled and quantized first, and then the electrical channel equalization and frequency offset compensation are further performed. Finally, the phase judgment restores the sent data, and performs statistics on the symbol error rate.

Fig. 3 is a structural diagram of a specific embodiment of a joint equalization and frequency offset estimation device in burst coherent optical fiber communication according to the present invention. As shown in FIG. 3 , the joint equalization and frequency offset estimation device in the burst coherent optical fiber communication of the present invention includes an equalization module 1 , a decision module 2 , and a frequency offset estimation module 3 .

Since the error signal used for updating the equalization tap weight coefficient needs to use the decision signal, so before transmitting the signal, it is necessary to calculate the equalization tap weight coefficient C _k-1 , error signal ε _k-1 , and cumulative phase error estimate z_tmp _{k- 1.} The cumulative frequency offset θ _k-1 is initialized. In this embodiment, a training sequence is used for initialization, and the initialization process is as follows:

The length of the training sequence pilot _k is L. L is set by the user and must be greater than the length l of the sliding window of the filter 5 in the frequency offset estimation module 3. Considering the requirement for the number of divergent symbols of the burst frequency offset, L needs to be long enough to Ensure that the equalization module 1 can be converged; the training sequence pilot _k is processed by analog interference to obtain a signal sequence R _k of length L;

The equalization module 1 receives the signal sequence R _k , and the equalization tap weight coefficient is always C _k =1; the signal R ₁ generates the first equalization value through the equalization module 1 Signal R ₂ passes through the equalization module 1 to generate the second equalized value Delay z_tmp ₁ by one beat to generate z_tmp_delay ₁ , phase error estimate α ₂ =z_tmp ₂ ·(z_tmp_delay ₁ ) ^* , take the angle of α ₂ to get the first angle value β ₂ ; repeat the signal sequence R _k with the second The signals are processed in the same way. Since the length of the filter sliding window is l, the angle values need to be accumulated to l. When the signal R _l+1 is reached, the angle value is accumulated to l, and the l angle values enter the filter to suppress the noise phase together to obtain the first frequency offset estimation value θ _l+1 , and the error signal is obtained The calculation of the error signal ε _k in the initialization process does not use the judgment value, but uses the training sequence pilot _k instead of the judgment value.

The equalization tap weight coefficient starts to be iteratively updated C _l+2 =C _l+1 -Δε _l+1 R _l+1 ^* =1-Δε _l+1 R _l+1 ^* , and the signal R _l+2 is obtained through the equalization module 1 Equilibrium value Perform frequency offset estimation to obtain the first cumulative frequency offset θ _l+2 . Judgment module 2 pairs Perform frequency offset compensation according to θ _l+1 to obtain the signal Judgment Gets Judgment Value Although the judgment value is obtained, it is not used in the initialization process.

The subsequent signal sequence R _k uses iteratively updated equalization tap weight coefficients for joint equalization and frequency offset estimation, where C _k =C _k-1 -Δε _k-1 R _k-1 ^* , The initialization of the equalization tap weight coefficient is completed until the end of the training sequence. At this time, the equalization module 1 converges, and the current C _L , ε _L , z_tmp _L , and θ _L are used as the equalization tap weight coefficient C _k-1 and the error signal ε _{k -1} , the estimated value of the cumulative phase error z_tmp _k-1 , and the initial value of the cumulative frequency offset θ _k-1 .

After the initialization is completed, the joint equalization and frequency offset estimation device receives the signal r _k generated after coherent demodulation, AD sampling and quantization, and starts joint equalization and frequency offset estimation.

The equalization module 1 receives the signal r _k and performs equalization processing, firstly, according to the decision value of the previous signal r _k-1 fed back by the decision module 2 Calculate the equalization tap weight coefficient of this equalization with the cumulative frequency offset θ k _{-1 of the previous signal r k- 1} fed back by the frequency offset estimation module 3:

${\mathrm{<span\; class="notranslate">\; C</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; =</span>{\mathrm{<span\; class="notranslate">\; C</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; \&Delta;\&epsiv;</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\mathrm{<span\; class="notranslate">\; r</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}^{<span\; class="notranslate">\; *</span>}$

It can be seen that the equalization tap weight coefficients are updated iteratively. Among them, C _k and C _k-1 are the equalization tap weight coefficients when the kth and k-1th signals pass through the equalization module 1 respectively; r _k-1 is the received signal sample value of the k-1th signal, is its conjugate value; Δ is the iterative step size, which is a positive number, set by the user, and needs to be set small enough to ensure that the iterative process can converge; ε _k-1 is the error signal obtained from the k-1th signal, and its calculation The formula is:

${\mathrm{<span\; class="notranslate">\; \&epsiv;</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; =</span>{\stackrel{<span\; class="notranslate">\; ~</span>}{\mathrm{<span\; class="notranslate">\; m</span>}}}_{\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; -</span>{\stackrel{<span\; class="notranslate">\; ~</span>}{\mathrm{<span\; class="notranslate">\; m</span>}}}_{\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; \&Center\; Dot;</span>{\mathrm{<span\; class="notranslate">\; e</span>}}^{{\mathrm{<span\; class="notranslate">\; j\&theta;</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}}$

in, is the equalization value of the k-1th signal after passing through the equalization module 1, Is the judgment module 2 pairs The signal obtained after frequency offset compensation The decision value of , θ _k-1 represents the cumulative frequency offset obtained by passing the k-1th signal through the frequency offset estimation module 3 . Since the frequency offset compensation of θ _k-1 is performed before the signal judgment, and the estimated value after equalization does not have such compensation, it is necessary to remove the frequency offset compensation of the judgment value when calculating the error signal, so that the error signal ε _{k -1} is more accurate.

The equalization module 1 equalizes the signal r _k according to the calculated equalization tap weight coefficient C _k , and outputs the equalized value Send it to the decision module 2 and the frequency offset estimation module 3.

In this embodiment, the equalization module 1 is an adaptive equalizer based on the LMS algorithm, which is composed of a series of FIR filters.

Judgment module 2 receives the equalization value transmitted by equalization module 1 And calculate the equalization value according to the cumulative frequency offset θ _k-1 obtained by the k-1th signal through the frequency offset estimation module The frequency offset compensation signal And make a judgment, output the judgment value It is transmitted to the frequency offset estimation module 3 and fed back to the equalization module 1 for updating the weight coefficients of the equalization taps next time.

The frequency offset estimation module 3 receives the equalization value transmitted by the equalization module 1 and the decision value transmitted by decision module 2 Equilibrium value The signal phase is expressed as φ _d,k +kΔωT+φ _n,k , where φ _d,k represents the modulation phase, kΔωT represents the phase error introduced by frequency offset, φ _n,k is the phase error caused by Gaussian noise, and the decision value The signal phase is φ _d,k . Equilibrium value and judgment value The conjugate multiplication of can eliminate the influence of the modulation phase, and obtain the cumulative phase error estimate z_tmp _k of the first k signals:

$\mathrm{<span\; class="notranslate">\; z</span>}<span\; class="notranslate">\; \_</span>\mathrm{<span\; class="notranslate">\; t</span>}{\mathrm{<span\; class="notranslate">\; mp</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; =</span><span\; class="notranslate">\; (</span>{\stackrel{<span\; class="notranslate">\; ~</span>}{\mathrm{<span\; class="notranslate">\; m</span>}}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; \&Center\; Dot;</span>{{\stackrel{<span\; class="notranslate">\; ^</span>}{\mathrm{<span\; class="notranslate">\; m</span>}}}_{\mathrm{<span\; class="notranslate">\; k</span>}}}^{<span\; class="notranslate">\; *</span>}<span\; class="notranslate">\; )</span>$

$<span\; class="notranslate">\; =</span>{\mathrm{<span\; class="notranslate">\; \&phi;</span>}}_{\mathrm{<span\; class="notranslate">\; d</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; k\&Delta;\&omega;T</span>}<span\; class="notranslate">\; +</span>{\mathrm{<span\; class="notranslate">\; \&phi;</span>}}_{\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; \&phi;</span>}}_{\mathrm{<span\; class="notranslate">\; d</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; k</span>}}$

$<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; k\&Delta;\&omega;T</span>}<span\; class="notranslate">\; +</span>{\mathrm{<span\; class="notranslate">\; \&phi;</span>}}_{\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; k</span>}}$

Compute the phase error estimate α _k for the kth signal:

α _k = z_tmp _k ·(z_tmp_delay _k-1 ) ^*

=kΔωT+φ _n,k -[(k-1)ΔωT+φ _n,k-1 ]

=ΔωT+φ _n,k -φ _n,k-1

Among them, z_tmp_delay _k-1 is obtained by delaying one beat of the cumulative phase error estimation value z_tmp _k-1 of the previous k-1 signals, and (z_tmp_delay _k-1 ) ^* is its conjugate value. It can be seen that α _k only includes the frequency offset estimation component ΔωT of the kth signal and the phase change component φ _n,k -φ _n,k-1 caused by noise.

The memory in the frequency offset estimation module 3 is used to store the cumulative phase error estimate z_tmp _k , first take out z_tmp _k-1 and perform one-beat delay processing to obtain z_tmp_delay _k-1 to synchronize with z_tmp _k , and then write the z_tmp _k signal.

The phase component ΔωT of frequency offset estimation can only be between [-π, π], otherwise it will cause phase ambiguity, making frequency offset estimation non-deterministic, so the present invention limits Between [-π,π]. The frequency offset estimation module 3 performs an angle calculation operation on the phase error estimated value α _k to obtain an angle value β _k , and if the angle value β _k is within the range of [-π, π], the phase error estimated value α _k is sent to the filter to suppress noise Otherwise, the estimated phase error value α _k is discarded; the frequency offset estimated value ΔωT of the kth signal is obtained after the filter suppresses the noise phase; since the frequency offset is a cumulative effect, ΔωT must be accumulated to the previous k-1 signal On the cumulative frequency offset, the cumulative frequency offset θ _k =θ _k-1 +ΔωT of the k signals is calculated.

The frequency offset estimation module 3 sends and feeds back the accumulated frequency offset θ _k to the equalization module 1 and the decision module 2 respectively, for the equalization processing and decision of the next signal.

Example

Carry out simulation experiments on the joint equalization and frequency offset estimation device in the burst coherent optical fiber communication of the present invention, the parameters are set as: the attenuation factor of the single-mode optical fiber is 0.2dB/km, the second order dispersion coefficient is -20ps^2/km, the optical fiber transmission The distance is 100km, the receiver bandwidth is set to 50GHz, the length of the filter sliding window in the frequency offset estimation module 3 is 128, and the iteration step of the equalization module 1 is the optimal iteration step of 0.001 when the transmission distance is 100 kilometers.

Table 1 shows the lengths of training sequences used for different burst ranges of frequency offset.

Fig. 4 is a diagram of a burst frequency offset simulation result of a specific embodiment of the joint equalization and frequency offset estimation device in burst coherent optical fiber communication of the present invention. As shown in Figure 4, the curve shows that the joint equalization and frequency offset estimation device in the burst coherent optical fiber communication of the present invention can estimate and compensate for different burst frequency offset ranges, the larger the burst frequency offset range, the required initialization training sequence The longer the length is, the more the equalization module 1 can be converged during the training sequence initialization phase, so that the equalization module 1 can function stably. It can be seen from Fig. 4 that as the frequency offset burst range increases, the performance of the optical fiber communication receiving system will be affected to a certain extent, but it is also acceptable. Even when the burst frequency offset increases to ±6 GHz, the system performance loss is not very large, indicating that the joint equalization and frequency offset estimation device in burst coherent optical fiber communication of the present invention is suitable for optical fiber communication with burst frequency offset receiving system.

Fig. 5 is a simulation result diagram of the symbol error rate of the receiving system using the joint equalization and frequency offset estimation device in the burst coherent optical fiber communication of the present invention. As shown in Figure 5, the dotted line represents the case of equalization only, and the solid line represents the case of joint use of equalization and frequency offset estimation, and the equalization step size used in each curve is the optimal step size.

Table 2 shows the optimal step size of the equalization module 1 used under different frequency offsets when the optical fiber transmission distance is 100 km.

Frequency offset (Hz) 1M 10M 100M 500M balanced 0.002 0.05 0.1 - Equalization and Frequency Offset Estimation 0.001 0.001 0.001 0.001

Table 2

As shown in Figure 5, in the case of equalization only, the frequency offset that the receiving system can tolerate is very small. When the frequency offset value increases to 100M, even if the optical signal-to-noise ratio OSNR is high, the symbol error rate of the receiving system SER are very high; while using the joint equalizer and frequency offset estimation device, when the frequency offset increases to 200M, under the optimal iteration step size, the symbol error rate SER of the receiving system is significantly reduced, and with the optical signal noise With the increase of the ratio, the symbol error rate SER of the receiving system is further reduced. It can be seen that the joint equalization and frequency offset estimation device can greatly improve the performance of the burst coherent optical fiber communication receiving system.

Although the illustrative specific embodiments of the present invention have been described above, so that those skilled in the art can understand the present invention, it should be clear that the present invention is not limited to the scope of the specific embodiments. For those of ordinary skill in the art, As long as various changes are within the spirit and scope of the present invention defined and determined by the appended claims, these changes are obvious, and all inventions and creations using the concept of the present invention are included in the protection list.

Claims (4)

1. A joint equalization and frequency offset estimation device in burst coherent optical fiber communication, characterized in that it comprises: An equalization module, which receives the signal r _k quantized by AD sampling and carries out equalization processing, according to the judgment value of the previous signal r _k-1 fed back by the judgment module The cumulative frequency offset θ _{k-1 of the previous signal r k-1 fed} back by the frequency offset estimation module performs iterative update of the equalization tap weight coefficient: ${\mathrm{<span\; class="notranslate">\; C</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; =</span>{\mathrm{<span\; class="notranslate">\; C</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; \&Delta;</span>}{\mathrm{<span\; class="notranslate">\; \&epsiv;</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\mathrm{<span\; class="notranslate">\; r</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}^{<span\; class="notranslate">\; *</span>}$ Among them, C _k and C _k-1 are the equalization tap weight coefficients when the kth and k-1th signals pass through the equalization module respectively; r _k-1 is the received signal sample value of the k-1th signal; Δ is the iteration step size, which is a positive number and is set by the user. It needs to be set small enough to ensure that the iteration process can converge; ε _k-1 is the error signal obtained from the k-1th signal, and its calculation formula is: ${\mathrm{<span\; class="notranslate">\; \&epsiv;</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; =</span>{\stackrel{<span\; class="notranslate">\; ~</span>}{\mathrm{<span\; class="notranslate">\; m</span>}}}_{\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; -</span>{\stackrel{<span\; class="notranslate">\; ^</span>}{\mathrm{<span\; class="notranslate">\; m</span>}}}_{\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; \&CenterDot;</span>{\mathrm{<span\; class="notranslate">\; e</span>}}^{\mathrm{<span\; class="notranslate">\; j</span>}{\mathrm{<span\; class="notranslate">\; \&theta;</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}}$ in, is the equalized value of the k-1th signal after passing through the equalized module, is the decision module pair The decision value of the signal obtained after frequency offset compensation, θ _k-1 represents the cumulative frequency offset obtained by the k-1th signal through the frequency offset estimation module; The equalization module equalizes the signal r _k according to the equalization tap weight coefficient C _k , and outputs the equalized value Send to the judgment module and the frequency offset module; A judgment module, receiving the equalization value And read the cumulative frequency offset θ _k-1 from the memory, and calculate the equalization value The frequency offset compensation signal And make a judgment, output the judgment value Send to the frequency offset estimation module and feed back to the equalization module; A frequency offset estimation module, receiving the equalization value transmitted by the equalization module and the judgment value transmitted by the judgment module Equilibrium value The signal phase is expressed as φ _d,k +kΔωT+φ _n,k , where φ _d,k represents the modulation phase, kΔωT represents the phase error introduced by frequency offset, φ _nk is the phase error caused by Gaussian noise, and the decision value The signal phase is φ _d,k , calculate the cumulative phase error estimate z_tmp _k of the first k signals: $\begin{array}{c}\mathrm{<span\; class="notranslate">\; z</span>}<span\; class="notranslate">\; \_</span>{\mathrm{<span\; class="notranslate">\; tmp</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; =</span><span\; class="notranslate">\; (</span>{\stackrel{<span\; class="notranslate">\; ~</span>}{\mathrm{<span\; class="notranslate">\; m</span>}}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; \&Center\; Dot;</span>{{\stackrel{<span\; class="notranslate">\; ^</span>}{\mathrm{<span\; class="notranslate">\; m</span>}}}_{\mathrm{<span\; class="notranslate">\; k</span>}}}^{<span\; class="notranslate">\; *</span>}<span\; class="notranslate">\; )</span>\\ <span\; class="notranslate">\; =</span>{\mathrm{<span\; class="notranslate">\; \&phi;</span>}}_{\mathrm{<span\; class="notranslate">\; d</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; k\&Delta;\&omega;T</span>}<span\; class="notranslate">\; +</span>{\mathrm{<span\; class="notranslate">\; \&phi;</span>}}_{\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; \&phi;</span>}}_{\mathrm{<span\; class="notranslate">\; d</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; k</span>}}\\ <span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; k\&Delta;\&omega;T</span>}<span\; class="notranslate">\; +</span>{\mathrm{<span\; class="notranslate">\; \&phi;</span>}}_{\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; k</span>}}\end{array}$ Compute the phase error estimate α _k for the kth signal: α _k ＝z_tmp _k ·(z_tmp_delay _k-1 ) ^* ＝kΔωT+φ _n,k -[(k-1)ΔωT+φ _n,k-1 ]＝ΔωT+φ _n,k -φ _n,k-1 Among them, z_tmp_delay _k-1 is obtained by delaying one beat of the cumulative phase error estimate z_tmp _k-1 of the previous k-1 signals; Perform an angle calculation operation on the phase error estimate α _k to obtain the angle value β _k , if the angle value β _k is within the range of [-π, π], then send the phase error estimate α _k to the filter to suppress the noise phase, otherwise the phase The error estimate value α _k is discarded; the frequency offset estimate value ΔωT is obtained after the noise phase is suppressed by the filter, and the cumulative frequency offset θ _k of the first k signals is calculated = θ _k-1 + ΔωT; The frequency offset estimation module transmits the accumulated frequency offset θ _k to the memory and feeds it back to the equalization module; A memory for storing cumulative frequency offsets, first fetching the cumulative frequency offsets θ _{k-1 of the first k-1} signals and sending them to the decision module, and then receiving the cumulative frequency offsets θ _k of the first k signals transmitted by the frequency offset estimation module. 2. The joint equalization and frequency offset estimation device according to claim 1, wherein the equalization module is an adaptive equalizer based on the LMS algorithm, comprising a series of FIR filters. 3. The combined equalization and frequency offset estimation device according to claim 1, wherein the frequency offset estimation module includes a memory for storing the cumulative phase error estimate _{z_tmpk} , and _{z_tmpk-1} is taken out earlier for Delay one beat to get z_tmp_delay _k-1 to synchronize with z_tmp _k , and then write z_tmp _k signal. 4. The joint equalization and frequency offset estimation device according to claim 1, characterized in that, said equalization tap weight coefficient C _k-1 , error signal ε _k-1 , cumulative phase error estimate z_tmp _k-1 , The initial value of the cumulative frequency offset θ _k-1 is obtained by initializing the training sequence, and the initialization process is: The length of the training sequence pilot _k is L. L is set by the user and must be greater than the length l of the sliding window of the filter in the frequency offset estimation module. Considering the requirement for the number of divergent symbols of the burst frequency offset, L must be long enough to ensure Make the equalization module converge; the training sequence pilot _k obtains the signal sequence R _k after being processed by analog interference; The equalization module receives the signal sequence R _k , and the equalization tap weight coefficient is always C _k =1; the signal R ₁ generates the first equalization value through the equalization module Signal R ₂ passes through the equalization module to generate the second equalized value Delay z_tmp ₁ by one beat to generate z_tmp_delay ₁ , phase error estimate α ₂ = z_tmp ₂ ·(z_tmp_delay ₁ ) ^* , take an angle to α ₂ to get the first angle value β ₂ ; repeat the signal sequence R _k with the second The signals are processed in the same way until the signal R _l+1 , the angle value is accumulated to l, and the l angle values enter the filter to suppress the noise phase together to obtain the first frequency offset estimation value θ _l+1 , and the error signal is obtained The equalization tap weight coefficient starts to be iteratively updated C _l+2 ＝C _l+1 -Δε _l+1 R _l+1 ^* ＝1-Δε _l+1 R _l+1 ^* , the signal R _l+2 is equalized through the equalization module value Perform frequency offset estimation to obtain the first cumulative frequency offset θ _l+2 ; then use the iteratively updated equalization tap weight coefficient to perform joint equalization and frequency offset estimation on the signal sequence R _k , where C _k =C _k-1 -Δε _{k -1} R _k-1 ^* , The initialization of the equalization tap weight coefficient is completed until the end of the training sequence. At this time, the equalization module converges, and the current C _L , ε _L , z_tmp _L , and θ _L are used as the equalization tap weight coefficient C _k-1 and the error signal ε _{k- 1.} Initial values of the cumulative phase error estimate z_tmp _k-1 and the cumulative frequency offset θ _k-1 .