Method for estimating iq skew of transmitter based on special sequence in multi-subcarrier system

Abstract

The application provides a transmitter IQ skew estimation method based on a special sequence in a multi-subcarrier system. The configuration method of a training sequence of each subcarrier of a transmitting end is as follows: a random training sequence is generated for a subcarrier which is symmetric with the subcarrier in a frequency domain; the first half of the training sequence of the symmetric subcarrier which is symmetric with the subcarrier in the frequency domain is negative conjugate with the first half of the training sequence of the subcarrier; the second half of the training sequence of the symmetric subcarrier which is symmetric with the subcarrier in the frequency domain is conjugate with the second half of the training sequence of the subcarrier; further, the training sequences of the symmetric subcarriers are the same. The receiving end directly extracts a delay value to complete the estimation of the IQ skew of the transmitting end by performing phase noise elimination on the training sequence in the received signal after Fourier transformation and then dividing each two symmetric subcarrier sequences to obtain a phase containing a delay, and the estimation process will not interrupt normal communication and will not increase additional overhead.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of optical fiber communication, and in particular to an ultra-wideband coherent optical communication system and a multi-subcarrier transmission technology. BACKGROUND

[0002] In recent years, with the rapid development of cloud computing, cloud storage, Internet of Things, ultra-high-definition video and mobile data communication technologies, the global network data traffic has shown explosive growth. This trend has prompted optical fiber communication systems to continue to develop towards greater capacity and higher transmission rates. In order to cope with the growing demand for data transmission, on the one hand, the spectral efficiency of existing systems needs to be improved; on the other hand, the symbol rate of the transmission signal needs to be continuously increased. High-speed high-modulation format signals not only put higher requirements on the response linearity of optoelectronic devices, but also have poorer tolerance to channel noise and fiber nonlinear effects, thereby limiting the further improvement of transmission capacity and distance. Compared with traditional single-carrier systems, the multi-subcarrier multiplexing (DSCM) scheme has received more attention because of its outstanding advantages, such as flexible data rate, reduced transceiver complexity, and better robustness to various impairments, such as equalization-enhanced phase noise caused by dispersion compensation, frequency-dependent polarization mode dispersion, self-phase modulation, etc. Despite these advantages, DSCM systems also face some challenges, including frequency offset recovery, carrier phase recovery, and IQ skew compensation. IQ skew is the relative time delay between the I- and Q-signals of each polarization state on the transmitter, and is one of the main challenges faced by DSCM signals, because IQ skew will cause the subcarriers to be interfered by their "symmetric" subcarriers. This interference is frequency-dependent and increases with the increase of the frequency of each subcarrier. Therefore, it is crucial to monitor and compensate for the IQ skew, especially for the application of high-order modulation format signals.

[0003] For coherent optical fiber communication systems, a variety of IQ skew estimation and compensation methods have been proposed. In general, the IQ skew impairment can be reduced by adaptive equalization or calibration. Several adaptive equalization schemes have been proven to be effective in mitigating dynamic transmit (Tx) or receive (Rx) IQ skew. The equalizer can completely compensate for the Rx IQ skew, but the Tx IQ skew is more difficult to handle because it is mixed with channel impairments such as polarization rotation and phase noise. Therefore, the equalization capability of the Tx IQ skew is insufficient when the signal is transmitted with a large IQ skew. In addition, there have been many calibration methods proposed to address the Tx IQ skew. According to the requirements of the transmitted signal or whether the IQ skew monitoring can be provided during data transmission, higher estimation accuracy can be achieved, but they are used to calibrate the transmission system in the initial stage. When the Tx IQ skew changes slowly over time, because these methods require the transmission of a specially designed signal different from the transmitted signal, the above-mentioned methods cannot be used to track the change of the Tx IQ skew, which will interrupt normal communication. In addition, a multiple-input multiple-output equalizer based on the minimum mean square error criterion can be used at the receiving end to estimate the Tx IQ skew according to the characteristics of the equalizer tap coefficient matrix, however, due to the conjugate interference of the "symmetric" subcarriers, these methods are not effective in digital subcarrier modulation (DSCM) transmission. In a DSCM transmission system, due to the IQ skew, there is crosstalk between the symmetric subcarriers, and at this time the information of the IQ skew is covered by the information of the symmetric subcarriers, as shown in Figure 1 Figure 1, which includes four subcarriers, Sub-1, Sub-2, Sub-3, and Sub-4. Sub-1 and Sub-4 are a pair of symmetric subcarriers, and Sub-2 and Sub-3 are a pair of symmetric subcarriers. Sub-1 and Sub-2 are same-side subcarriers, and Sub-3 and Sub-4 are same-side subcarriers. SUMMARY

[0004] The technical problem to be solved by the present application is to provide a method for eliminating the interference caused by symmetric subcarriers by designing a special training sequence, and then extracting the skew value from the received frequency domain signal phase to complete the estimation of the transmitted signal.

[0005] The technical solution adopted by the present application to solve the above technical problem is a transmitter IQ skew estimation method based on a special sequence in a multi-subcarrier system. When the skew value is extracted from the received frequency domain signal phase by a special training sequence, it is considered that the extraction of the IQ skew will also be interfered by the phase noise. Therefore, the present application also estimates the Tx IQ skew based on the pair of subcarriers by using the correlation of the interference of the phase noise between the subcarriers, including the steps of:

[0006] The transmitting end generates the DSCM signal with multiple sub-carriers offline, generates independent pseudo-random bit sequences for each sub-carrier and maps into M-QAM payload symbols to complete QAM modulation to form a QAM modulated transmission sequence; a training sequence for avoiding interference of symmetric sub-carriers is added before each transmission sequence; the training sequence of each sub-carrier is generated under the limited condition of two parts of the training sequence based on symmetric sub-carriers and the training sequence based on the same side sub-carrier; the QAM modulated transmission sequence and the training sequence constitute a sub-carrier sequence

[0007] The specific method for generating the training sequence of each sub-carrier is as follows:

[0008] A random training sequence is generated for a symmetric sub-carrier in the frequency domain; the first half of the training sequence of the symmetric sub-carrier symmetric to the sub-carrier in the frequency domain is negatively conjugated with the first half of the training sequence of the sub-carrier for calculating Q path delay; and the second half of the training sequence of the symmetric sub-carrier symmetric to the sub-carrier in the frequency domain is conjugated with the second half of the training sequence of the sub-carrier for calculating I path delay;

[0009] The transmitting end performs pulse shaping on the sub-carrier sequence by using a root raised cosine (RRC) filter and then performs sub-carrier multiplexing in the frequency domain to obtain a transmitting signal;

[0010] The receiving end performs phase noise elimination by dividing each pair of two symmetric sub-carrier sequences after Fourier transformation on the training sequence in the receiving signal by using an offline digital signal processor (DSP) to obtain I path and Q path phases θ I and θ Q , and then extracts the delay values of the I path and the Q path from θ I and θ Q to complete estimation of Tx IQ skew.

[0011] The configuration method of the training sequence of each sub-carrier is further improved, and the training sequences of the sub-carriers on the same side in the frequency domain are all the same.

[0012] The present application proposes a Tx IQ skew estimation scheme based on a special sequence designed, and the special training sequence avoids interference of symmetric sub-carriers, so that the phase information contained in the frequency domain signal can be extracted more accurately.

[0013] The present application has the beneficial effects that: based on the specially designed training sequence, the interference of IQ skew estimation caused by the aliasing between subcarriers due to IQ skew can be effectively avoided. On this basis, the estimation scheme of the paired subcarriers can effectively eliminate the interference of phase noise on IQ skew estimation. Therefore, the extracted IQ skew value can maintain high accuracy. And compared with other traditional schemes, the scheme can be used for real-time monitoring of DSCM transmission system without interrupting communication and using other special signals to estimate IQ skew again. And the scheme can be found from the formula derivation that it eliminates the interference of Tx IQ amplitude and phase imbalance, so the algorithm has high robustness. In addition, it can also show good tolerance to line width because it better eliminates phase noise. BRIEF DESCRIPTION OF DRAWINGS

[0014] Figure 1 IQ skew effect diagram in a plurality of subcarriers;

[0015] Figure 2 Tx IQ skew estimation scheme flow diagram in a plurality of subcarriers;

[0016] Figure 3 The present application is a plurality of subcarrier coherent optical transmission system based on a specially designed training sequence. DETAILED DESCRIPTION

[0017] First, a transmitter model containing IQ impairment is built, and based on the model, the information contained in each subcarrier is derived. Assuming that the DSCM signal x(t) is: Where N is the number of subcarriers, n is the subcarrier sequence variable, t is the time domain variable, f n is the carrier frequency, x n (t) is the nth subcarrier signal, which is transmitted through the following IQ impairment transmission matrix:

[0018]

[0019] φ represents the IQ phase imbalance angle, α I represents the amplitude amplification multiple of the I signal, α Q represents the amplitude amplification multiple of the Q signal, δ represents the impulse response function, τ I represents the I relative delay, τ Q represents the Q relative delay;

[0020] Considering that the received time domain signal y(t) contains phase noise, it can be represented as:

[0021]

[0022] k1, k2 represent time-domain coefficients of sub-carrier signals and their conjugate components respectively, and θ represents phase noise, * represents conjugate, represents convolution;

[0023] After the Fourier transform operation, it can be represented in the frequency domain as:

[0024]

[0025] w represents a frequency domain variable, K1, K2 represent frequency domain coefficients of sub-carrier signals and their conjugate components respectively, · represents multiplication, X n represents the frequency domain signal of the nth sub-carrier;

[0026] After down-conversion of the received signal, the received signal of each sub-carrier can be represented as:

[0027]

[0028] where the sub-carrier number Sub-m is the symmetric sub-carrier of Sub-n, the sub-carrier number satisfies the relationship m+n=N+1, and N is the number of sub-carriers. It can be intuitively found that the most direct way to extract IQ skew is to extract τ I and τ Q from K1 and K2. However, K2 is the coefficient carried by the symmetric sub-carrier signal, and the two interfere with each other. We need to design a scheme to eliminate the interference of the symmetric sub-carrier, so we assume and two conditions, under which the corresponding τ I and τ Q can be extracted. Because this information is contained in the phase, the interference of the phase noise should be eliminated before estimating the Tx IQ skew. Considering that for the DSCM system, in the case of optical back-to-back, each sub-carrier is affected by the same phase noise. Therefore, we use the method of paired sub-carriers to eliminate the interference of the phase noise, which is defined as paired sub-carrier estimation. Then we use the innermost pair of sub-carriers as the reference sub-carrier to eliminate the phase noise, and other pairs of sub-carriers that satisfy the corresponding relationship can also be used as the reference sub-carrier for estimation.

[0029]

[0030] However, the above formula assumes that X nI (w) and e jθ(w) satisfy the commutative law, which is not actually satisfied, resulting in residual phase noise. Further, we can assume that the same training sequence is sent on the same side of the sub-carrier pair, and then the phase noise elimination can be simplified as:

[0031]

[0032] Based on the two assumptions, by using inverse Fourier transform on the two assumptions, we can get its relationship in time domain as follows:

[0033]

[0034] Therefore, we can eliminate the sub-carrier interference by designing a special training sequence to estimate the IQ skew of the transmitting end. The estimation process is shown in Figure 2 The special training sequence is designed by two parts: the first half of the training sequence is a negative conjugate relationship, which is used to calculate the Q path delay, and the second half is a conjugate relationship, which is used to calculate the I path delay. In order to overcome the influence of phase noise, we use two kinds of paired sub-carriers: one is that the training sequence on the symmetric sub-carrier of each pair satisfies the conjugate relationship, and the other is that the training sequence of the same side sub-carrier is designed as the same digital sequence, so as to better eliminate the phase noise. Then, by calculating the Fourier transform of the signal on the paired sub-carrier, the I path and Q path delays are calculated by using formula (5.a) and (5.b) (for conjugate sub-carrier pair), formula (6.a) and (6.b) (for adjacent sub-carrier pair on the same side), so as to calculate the relative delay of the two paths. Therefore, after the phase noise is eliminated, the delay value contained in the phase can be directly extracted, and the size of the Tx IQ skew existing in the system can be accurately estimated. And from the formula, it can be seen that the present scheme is independent of the amplitude and phase imbalance between the I path and the Q path in the IQ skew estimation, which improves the robustness of the estimation.

[0035] The present application is based on the specific working principle of a coherent optical fiber communication system carrying a multi-subcarrier signal with a specially designed training sequence, as shown in Figure 3 , which includes information processing at the transmitting end, a coherent optical transceiver, and signal processing at the receiving end. Taking a DSCM signal with four subcarriers as an example:

[0036] Tx information processing at the transmitting end: the transmitter generates a DSCM signal with four subcarriers offline, generates an independent pseudo-random bit sequence for each subcarrier and maps it to M-QAM valid payload symbols to complete QAM modulation. In order to avoid the interference of the signal under the influence of Tx IQ skew by the symmetric subcarriers, we design a special training sequence before the valid payload symbol. According to the estimation process of the present application as shown in Figure 2As shown, first, a random training sequence on Sub-1 is generated and divided into two parts, one based on the training sequence on symmetric subcarriers and the other based on the training sequence on same-side subcarriers. The training sequence on symmetric subcarriers should satisfy the relationship of the first half being negative conjugate and the second half being conjugate with the training sequence in Sub-1. The same relationship is also satisfied for other pairs of symmetric subcarriers, and the training sequence on same-side subcarriers is the same. Then, pulse shaping is performed using a root raised cosine (RRC) filter. The shaped signal is up-converted to different frequencies and then subcarrier multiplexing is performed in the frequency domain. Various impairments of the transmitter are simulated digitally in the Tx DSP. In the offline DSP at the receiving end, the Tx IQ skew is estimated using the proposed scheme.

[0037] Based on the special sequence designed, two Tx IQ skew estimation schemes are proposed. Since the special training sequence avoids the interference of symmetric subcarriers, the phase information contained in the frequency domain signal can be extracted more accurately.

[0038] The first scheme directly uses a pair of symmetric subcarriers. The frequency domain signals corresponding to the two subcarriers are extracted by performing Fourier transform on the received signals of the two subcarriers. Since the training sequence is known, the Fourier transform of the corresponding training symbol can be directly removed at the receiving end. By calculating the designed front and rear training sequences, the phase θ I and θ Q without the symbol information of the training sequence itself can be obtained.

[0039] The other scheme takes a pair of same-side subcarriers in the frequency domain, such as Sub-1 and Sub-2. Based on the training sequence on symmetric subcarriers, the training sequences on Sub-1 and Sub-2 are further set to the same symbol. Therefore, the operation of removing the symbol information by performing Fourier transform on the training sequence and then dividing can be avoided. Instead, the frequency domain signals of the two subcarriers are directly divided to obtain the corresponding phase θ I and θ Q containing the delay. The phase noise is completely eliminated in this scheme, and the phase information is more accurate. Further, by using the relationship between the product of the subcarrier frequency interval and the delay and the phase, the delays of the I and Q channels are calculated using the phases obtained by the front and rear sequences, respectively. Finally, the accurate Tx IQ skew estimation value is obtained by subtracting the two delays.

[0040] The present application can be used in optical communication system using multi-subcarrier transmission, but the optical communication system can be used for high-precision estimation of IQ skew under multi-subcarrier transmission. Since no additional pilot is needed, the estimation is realized only through a special training sequence, so the normal communication is not interrupted, and the corresponding training sequence can still be used for channel estimation and equalization, etc., without increasing additional overhead. The present application can be used in various communication systems using multi-subcarrier mode to realize high-precision estimation of IQ skew.

Claims

1. A transmitter IQ skew estimation method based on a special sequence in a multi-subcarrier system, characterized in that: include: The transmitter generates a DSCM signal with multiple subcarriers offline, generates an independent pseudo-random bit sequence for each subcarrier, and maps it into an M-QAM payload symbol to perform QAM modulation to form a QAM modulated transmission sequence. A training sequence is added before each transmission sequence to avoid interference between symmetric subcarriers. The training sequence for each subcarrier is generated under the constraints of two parts: a training sequence based on the symmetric subcarriers and a training sequence based on the same-side subcarriers. The QAM modulated transmission sequence and the training sequence form a subcarrier sequence. The specific method for generating the training sequence for each subcarrier is: A pair of symmetric subcarriers that are symmetric in the frequency domain generates a random training sequence for one subcarrier; the first half of the training sequence of the symmetric subcarrier that is symmetric in the frequency domain is negatively conjugated with the first half of the training sequence of the subcarrier to calculate the Q path delay; The second half of the training sequence of the symmetric subcarrier that is symmetric with the subcarrier in the frequency domain is conjugated with the second half of the training sequence of the subcarrier to calculate the I-path delay; The transmitter uses a root raised cosine RRC filter to pulse-shape the subcarrier sequence, and then performs subcarrier multiplexing in the frequency domain to obtain the transmission signal; The receiving end uses an offline digital signal processor DSP to perform Fourier transformation on the training sequence in the received signal, and then divides the two symmetrical subcarrier sequences to eliminate the phase noise to obtain the I and Q phases with delay θ I and θ Q , then, from θ I and θ Q The IQ skew estimation at the transmitter is completed by directly extracting the I and Q delay values.

2. The method according to claim 1, wherein: Get the I and Q phases θ including the delay I and θ Q Specifically: Among them, angle is the phase angle function, Y nI (w) is the frequency domain received signal corresponding to the second half of the training sequence of the nth subcarrier, Y mI (w) is the frequency domain received signal corresponding to the second half of the training sequence of the mth subcarrier, X nI (w) is the frequency domain transmitted signal corresponding to the second half of the training sequence of the nth subcarrier, X mI (w) is the frequency domain transmitted signal corresponding to the second half of the training sequence of the mth subcarrier, Y nQ (w) is the frequency domain received signal corresponding to the first half of the training sequence of the nth subcarrier, Y mQ (w) is the frequency domain received signal corresponding to the first half of the training sequence of the mth subcarrier, X nQ (w) is the frequency domain transmitted signal corresponding to the first half of the training sequence of the nth subcarrier, X mQ (w) is the frequency domain transmitted signal corresponding to the first half of the training sequence of the m-th subcarrier.

3. The method according to claim 1, wherein: From θ I and θ Q Directly extract the delay value τ of I path I and the delay value τ of the Q path Q Specifically: I =θ I / [2π(f m -f n )]; τ Q =θ Q / [2π(f m -f n )]; Among them, f n is the frequency of the nth subcarrier, f m is the frequency of the mth subcarrier symmetrical to the nth subcarrier.

4. The method according to claim 1, wherein: In the method for configuring the training sequence of each subcarrier, the training sequences of the subcarriers on the same side in the frequency domain are all the same.

5. The method according to claim 4, wherein: Get the I and Q phases θ including the delay I and θ Q Specifically: Among them, angle is the phase angle function, Y 1I (w) is the frequency domain received signal of the I-channel subcarrier on the frequency domain side, Y 2I (w) is the I-channel frequency domain received signal of the subcarrier on the other side of the frequency domain; Y 1Q (w) is the Q-path frequency domain received signal of the subcarrier on one side of the frequency domain, Y 2Q (w) is the Q-path frequency domain received signal of the subcarrier on the other side of the frequency domain.

6. The method according to claim 4, wherein: From θ I and θ Q Directly extract the delay value τ of I path I and the delay value τ of the Q path Q Specifically: I =θ I / [2π(f2-f1)]; τ Q =θ Q / [2π(f2-f1)]; Among them, f1 is the frequency of the subcarrier on one side of the frequency domain, and f2 is the frequency of the subcarrier on the other side of the frequency domain.

Images