CN115242584A - A method and device for optimizing the complexity of MLSE algorithm based on lookup table - Google Patents

Abstract

The present application discloses a method and a device for optimizing the complexity of an MLSE algorithm based on a lookup table, and relates to the technical field of signal processing. The method includes: receiving a coherent optical signal in a coherent optical communication system in which a signal at a transmitting end is filtered by a narrow band; Generate an N-symbol distorted signal LUT according to the coherent optical signal; calculate a branch metric value according to the received signal, the N-symbol distorted signal LUT and the calculation scheme of optimizing complexity; The MLSE algorithm recovers the original data in the coherent optical signal. It solves the problem of high computational complexity in the existing MLSE technology, and achieves the operation of replacing the calculation of Euclidean distance with the operation of calculating the difference between the received signal and the N-symbol LUT and taking the absolute value, reducing the required amount in the process of digital signal processing. Computational complexity, thereby reducing the complexity and power consumption of the corresponding optical communication integrated circuit chip.

Description

A method and device for optimizing the complexity of MLSE algorithm based on lookup table

technical field

The invention relates to a method and a device for optimizing the complexity of an MLSE algorithm based on a look-up table, and belongs to the technical field of signal processing.

Background technique

Due to the advantages of high performance and high spectral efficiency of optical communication systems that combine pre-filtering and sequence detection, the method of sequence detection at the receiver has been widely used in the research of pre-filtered optical communication systems. Among them, the sequence detection algorithm is a method to effectively equalize the inter symbol interference (ISI). Therefore, in the coherent optical communication system with narrowband filtering, the severe ISI introduced by the narrowband filtering can be eliminated by the sequence detection algorithm. effectively eliminated. The Maximum Likelihood Sequence Estimation (MLSE) algorithm is a commonly used sequence detection algorithm. The current MLSE algorithm is implemented based on the Viterbi algorithm. At the receiving end, channel estimation is first required to obtain channel pulses. The channel response is used as the weight value for calculating the branch metric value in the grid graph, which is used to estimate the signal expected to be received. Finally, the Viterbi algorithm is used to restore the original data in turn.

列计算欧式距离，因此，可以省去传统MLSE算法中，采用N个抽头系数估计理想输出符号所需的大量乘法和加法计算过程。因此，相比于传统的MLSE算法，现行的LUT-MLSE方案可以有效降低计算复杂度。但是，该现行方案的技术复杂度仍然以M为底数，N为幂的指数增长，当M和N较大时，仍需要较大的计算复杂度。In the existing solution, by using the N-symbol distorted signal LUT to record the multi-symbol correlation features introduced by narrowband filtering and guiding the MLSE process at the receiving end, the ISI introduced by pre-filtering can be effectively equalized. The specific scheme is as follows: LUT-MLSE system, only with the first LUT

The column calculates the Euclidean distance, and therefore, a large number of multiplication and addition calculations required in the traditional MLSE algorithm to estimate the ideal output symbol using N tap coefficients can be omitted. Therefore, compared with the traditional MLSE algorithm, the current LUT-MLSE scheme can effectively reduce the computational complexity. However, the technical complexity of the current solution still grows exponentially with M as the base and N as the power. When M and N are large, a large computational complexity is still required.

SUMMARY OF THE INVENTION

The purpose of the present invention is to provide a method and an apparatus for optimizing the complexity of the MLSE algorithm based on a lookup table, so as to solve the problems existing in the prior art.

To achieve the above object, the present invention provides the following technical solutions:

According to a first aspect, an embodiment of the present invention provides a method for optimizing the complexity of a lookup table-based MLSE algorithm, the method comprising:

In a coherent optical communication system in which the signal at the transmitting end is filtered by a narrow band, the coherent optical signal is received;

generating an N-symbol distorted signal LUT according to the coherent optical signal;

Calculate the branch metric value according to the received signal, the N-symbol distorted signal LUT and the calculation scheme for optimizing the complexity;

The original data in the coherent optical signal is restored according to the calculated branch metric values.

Optionally, calculating the branch metric value according to the received signal, the N-symbol distorted signal LUT and the calculation scheme of the optimization complexity includes:

Obtain the sampling value of the middle column in the N-symbol distortion signal LUT;

Calculate the difference between the real part of the received signal and the real part of the acquired sample value and take the absolute value;

Calculate the difference between the imaginary part of the received signal and the imaginary part of the obtained sampling value and take the absolute value;

The sum of the two calculated absolute values is used as the branch metric value.

Optionally, the sum of the two absolute values obtained by the calculation is used as the branch metric value, including:

The branch metrics are:

均为复数，y_k为所述接收信号，

为所述采样值，real(y_k)为所述接收信号的实部，imag(y_k)为所述接收信号的虚部，

为所述采样值的实部，

为所述采样值的虚部。where y _k and

are complex numbers, y _k is the received signal,

is the sampled value, real(y _k ) is the real part of the received signal, imag(y _k ) is the imaginary part of the received signal,

is the real part of the sampled value,

is the imaginary part of the sampled value.

Optionally, calculating the branch metric value according to the received signal, the N-symbol distorted signal LUT and the calculation scheme of the optimization complexity includes:

transforming the N-symbol distorted signal LUT into a real LUT;

Obtain the real sampling value of the middle column in the real LUT;

separating the received signal into a real part signal and an imaginary part signal;

For each signal obtained by separation, calculate the absolute value of the difference between the real part in each signal and the real sample value;

The absolute value of the calculated difference is determined as the branch metric value.

Optionally, determining the absolute value of the calculated difference as the branch metric value, including:

The branch metrics are:

都是实数信号，y_k为分离得到的每路信号中的实部，

为获取得到的实数采样值。where y _k and

are real signals, y _k is the real part of each signal obtained by separation,

to obtain the real sample value obtained.

Optionally, the generating an N-symbol distorted signal LUT according to the coherent optical signal includes:

performing analog-to-digital conversion on the coherent optical signal to obtain a digital signal;

performing digital signal processing on the digital signal;

Separating the training sequence in the processed digital signal;

The N-symbol distortion signal LUT is generated according to the training sequence and a LUT training generator.

Optionally, the DSP processing includes one or more of dispersion compensation, clock recovery, polarization demultiplexing, polarization mode dispersion compensation, frequency offset compensation, and phase offset compensation.

In a second aspect, an apparatus for optimizing the complexity of an MLSE algorithm based on a lookup table is provided, the apparatus includes a memory and a processor, the memory stores at least one program instruction, and the processor loads and executes the At least one program instruction to implement the method of the first aspect.

By receiving a coherent optical signal in a coherent optical communication system; generating an N-symbol distorted signal LUT according to the coherent optical signal; calculating a branch metric value according to the received signal, the N-symbol distorted signal LUT and a calculation scheme for optimizing complexity; The original data in the coherent optical signal is restored according to the calculated branch metric values. The problem of high computational complexity in the prior art is solved, and the calculation of Euclidean distance is replaced by the operation of calculating the absolute value of the difference between the real and imaginary parts of the complex signal and the real LUT, which reduces the need for digital signal processing. Computational complexity, thereby reducing the complexity and power consumption of the corresponding optical communication integrated circuit chip.

The above description is only an overview of the technical solution of the present invention. In order to understand the technical means of the present invention more clearly, and implement it according to the content of the description, the preferred embodiments of the present invention are described in detail below with the accompanying drawings.

Description of drawings

1 is a method flowchart of a method for optimizing the complexity of an MLSE algorithm based on a lookup table provided by an embodiment of the present invention;

2 is a possible schematic diagram of generating an N-symbol distorted signal LUT after the transmitting end sends a signal to the receiving end receiving the signal provided by an embodiment of the present invention;

3 is a schematic diagram of calculating branch metric values in a first solution provided by an embodiment of the present invention;

4 is a schematic diagram of the entire process of signal processing by a receiving end provided by an embodiment of the present invention;

FIG. 5 is a possible schematic diagram of state transition provided by an embodiment of the present invention;

6 is a state transition diagram from time T1 to time T2 provided by an embodiment of the present invention;

7 is a schematic diagram of a possible logic gate circuit of a full adder provided by an embodiment of the present invention;

8 is a schematic diagram of a possible logic gate circuit of an array multiplier provided by an embodiment of the present invention;

FIG. 9 is a schematic diagram illustrating a comparison of system performance between an existing solution provided by an embodiment of the present invention and the first solution in the present application;

FIG. 10 is a schematic diagram showing a comparison of system performance between the existing solution provided by an embodiment of the present invention and the second solution in this application.

Detailed ways

The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are a part of the embodiments of the present invention, but not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

In the description of the present invention, it should be noted that the terms "center", "upper", "lower", "left", "right", "vertical", "horizontal", "inner", "outer", etc. The indicated orientation or positional relationship is based on the orientation or positional relationship shown in the accompanying drawings, which is only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the indicated device or element must have a specific orientation or a specific orientation. construction and operation, and therefore should not be construed as limiting the invention. Furthermore, the terms "first", "second", and "third" are used for descriptive purposes only and should not be construed to indicate or imply relative importance.

In the description of the present invention, it should be noted that the terms "installed", "connected" and "connected" should be understood in a broad sense, unless otherwise expressly specified and limited, for example, it may be a fixed connection or a detachable connection Connection, or integral connection; can be mechanical connection, can also be electrical connection; can be directly connected, can also be indirectly connected through an intermediate medium, can be internal communication between two elements. For those of ordinary skill in the art, the specific meanings of the above terms in the present invention can be understood in specific situations.

In addition, the technical features involved in the different embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other.

Please refer to FIG. 1, which shows a method flowchart of a method for optimizing the complexity of an MLSE algorithm based on a lookup table provided by an embodiment of the present application. As shown in FIG. 1, the method includes:

Step 101: Receive a coherent optical signal in a coherent optical communication system in which the signal at the transmitting end is filtered by a narrow band;

The method of the present application is used in a coherent optical communication system, and the present application mainly introduces the signal processing method of the receiving end. Unless otherwise specified below, this method is used in the receiving end of a coherent optical communication system based on QPSK (Quadrature Phase Shift Keying, quadrature phase shift keying) modulation.

Optionally, the original binary bit stream is mapped into QPSK symbols at the transmitting end, and a filtered QPSK signal is obtained through a narrowband filter. After that, a digital signal to be sent is obtained through Nyquist sampling, an analog signal is obtained by converting the digital signal through a digital-to-analog converter, and optical modulation is performed to obtain an optical domain sending signal. The transmitted signal in the optical domain reaches the receiving end through the optical fiber, and realizes coherent optical reception through the coherent receiver, that is, the receiving end can correspondingly receive the coherent optical signal sent by the transmitting end.

Step 102, generating an N-symbol distortion signal LUT according to the coherent optical signal;

Optionally, this step includes:

First, performing analog-to-digital conversion on the coherent optical signal to obtain a digital signal;

Second, perform digital signal DSP (Digital Signal Processing, digital signal processing) processing on the digital signal;

The DSP processing includes one or more of dispersion compensation, clock recovery, polarization demultiplexing, polarization mode dispersion compensation, frequency offset compensation, and phase offset compensation.

Third, separate the training sequence in the processed digital signal;

Fourth, the N-symbol distortion signal LUT is generated according to the training sequence and the LUT training generator.

Optionally, the training sequence is input into the LUT training generator, and the N-symbol distortion signal LUT is generated by the LUT training generator.

For example, please refer to FIG. 2 , which shows a possible schematic diagram of generating an N-symbol distorted signal LUT after the transmitting end sends a signal to the receiving end receiving the signal.

In this application, the N-symbol distorted signal LUT is generated by using the training sequence at the receiving end, so that the narrowband filtering at the transmitting end, and the influence of other integrated channel responses and nonlinear impairments on the characteristics of the original transmitted signal can be recorded, and the effect of equalization on various types of signals can be more effectively realized. Universal equalization of signal impairments.

Step 103, according to the received signal, the N-symbol distortion signal LUT and the calculation scheme of the optimization complexity, calculate the branch metric value;

Optionally, as a possible implementation, this step includes:

First, obtain the sampling value of the middle column in the N-symbol distortion signal LUT;

The sampling value obtained in the middle column is:

Second, calculate the difference between the real part of the received signal and the real part of the obtained sampling value and take the absolute value;

其中，y_k和

均为复数，y_k为所述接收信号，

为所述采样值，real(y_k)为所述接收信号的实部，

为所述采样值的实部。The first absolute value is:

where y _k and

are complex numbers, y _k is the received signal,

is the sampled value, real(y _k ) is the real part of the received signal,

is the real part of the sampled value.

thirdly, calculating the difference between the imaginary part of the received signal and the imaginary part of the obtained sampling value and taking the second absolute value;

和

均为复数，y_k为所述接收信号，

为所述采样值，imag(y_k)为所述接收信号的虚部，

为所述采样值的虚部。The second absolute value is:

and

are complex numbers, y _k is the received signal,

is the sampled value, imag(y _k ) is the imaginary part of the received signal,

is the imaginary part of the sampled value.

Fourth, the sum of the two calculated absolute values is used as the branch metric value.

That is, in the first possible implementation, the branch metric is:

To sum up, in the first possible implementation, please refer to Figure 3, use the real part and the imaginary part to calculate the difference respectively, then take the absolute value to calculate the distance between the two right-angled sides in the figure, and calculate the two complex points by approximation. The difference between the multipliers can be eliminated and the number of adders can be reduced. Since in the hardware circuit system, the complexity and power consumption of the multipliers are significantly higher than that of the adders. Therefore, the present application calculates the branch metric value by adopting the above method. Reduced computational complexity.

In a second possible implementation manner of this step, this step includes:

First, the N-symbol distortion signal LUT is optimized into a real LUT;

Different from the first possible implementation manner, in the second possible implementation manner, the complex-numbered N-symbol distorted signal LUT is replaced with a real-numbered LUT.

Second, obtain the real sampling value of the middle column in the real LUT;

For example, the obtained real sample value is:

thirdly, separating the received signal into a real part signal and an imaginary part signal;

The received signal is a complex signal, and the present application separates the received signal into a real part signal and an imaginary part signal. For example, if the received signal is a-bi, the separated real part signal is a, and the imaginary part signal is -b.

Fourth, for each signal obtained by separation, calculate the absolute value of the difference between the real part in each signal and the real sample value;

In this step, the absolute value of the difference between the real part in each signal and the real sampled value is calculated.

其中，y_k和

都是实数信号，y_k为分离得到的每路信号中的实部，

为获取得到的实数采样值。Specifically, for one signal, the calculated absolute value is:

where y _k and

are real signals, y _k is the real part of each signal obtained by separation,

to obtain the real sample value obtained.

Fifth, the absolute value of the calculated difference is determined as the branch metric value.

The calculated score measure is:

Step 104: According to the calculated branch metric values, use an N-symbol LUT-based MLSE algorithm to restore the original data in the coherent optical signal.

After the branch metric value is calculated, the original data in the coherent optical signal can be restored according to the calculated branch metric value. Specifically, please refer to FIG. 4 , which shows a complete flowchart of the method described in this application. As shown in FIG. 4 , the establishment of transition state and transition output grid, the accumulation of branch metric values + comparison + Selection, survival path storage and traceback output. This is similar to the recovery method in the existing solution, and details are not described here.

Taking the second possible implementation as an example, assuming that the modulation format is QPSK and the sequence detection length is 3, after the real and imaginary parts of the received signal are separated, for the LUT-MLSE technical scheme with sequence detection length 3, the real part and the imaginary part are The total number of states using the MLSE algorithm in each part is ²² .

As shown in Figure 5, when the input symbol at time T1 is "-1", the current four states will have state transitions due to the input of "-1", as shown by the hollow pointed arrows between time T1 and time T2. When the input symbol at time T1 is "1", the current four states will all undergo state transition due to the input of "1", as shown by the solid pointed arrows between time T1 and time T2.

For a LUT whose modulation format is QPSK, N=3 is shown in Table 1, where ε is the pre-filtered symbol amplitude reduction factor, and Δ is the ISI received by the symbol from adjacent symbols. As shown in FIG. 6, it shows a state transition diagram from time T1 to time T2. Different states, according to different input symbols, can be calculated

Table 1

列计算欧式距离，因此，可以省去传统MLSE算法中，采用N个抽头系数估计理想输出符号所需的大量乘法和加法计算过程。因此，相比于传统的MLSE算法，现行的LUT-MLSE方案可以有效降低计算复杂度。但是，该现行方案仍然以M为底数，N为幂的指数增长，当M和N较大时，仍需要较大的计算复杂度。而在本申请中，将原方案中计算复数接收信号与复数LUT之间欧氏距离的操作，替换为对实数差值取绝对值的操作，省去了原本需要的实数乘法器，同时大大减少了实数加法器的数量，也因此降低集成电路芯片实现的复杂度以及系统运行功耗。Different branch metrics, when the state is {-1, -1} and the input symbol is "-1", correspond to the first row in Table 1, and similarly, when the state is {1, 1}, The input symbol is "1", which corresponds to the eighth row in Table 1. Therefore, the present application uses the sampled values recorded in the lookup table as an estimate of the received signal, and calculates the branch metric value. Also, because the ISI introduced by narrowband filtering comes from the overlapping of adjacent pulses, the middle column in the lookup table effectively records the correlation between the symbol at the current moment and the adjacent symbols. In the existing scheme, only the first

The column calculates the Euclidean distance, and therefore, a large number of multiplication and addition calculations required in the traditional MLSE algorithm to estimate the ideal output symbol using N tap coefficients can be omitted. Therefore, compared with the traditional MLSE algorithm, the current LUT-MLSE scheme can effectively reduce the computational complexity. However, the current scheme still uses M as the base and N as the power of exponential growth, and when M and N are large, a large computational complexity is still required. In the present application, the operation of calculating the Euclidean distance between the complex received signal and the complex LUT in the original scheme is replaced by the operation of taking the absolute value of the real difference, which omits the originally required real multiplier and greatly reduces the The number of real number adders is reduced, and the complexity of integrated circuit chip implementation and the power consumption of system operation are also reduced.

At time T2, each state has two branches pointing to this state. According to the Viterbi algorithm, the larger branch is removed from the two branch values pointing to this state, and the other smaller branch is reserved. For example, from time T1 to time T2, states S0 to S1 are retained, and the dotted line from S2 to S1 is discarded. By analogy, at each moment, each state discards the branch with a larger branch metric value, and accumulates the branch metric value of the reserved branch. According to statistics and research, when the accumulated length reaches 5 times the sequence detection length N, there is almost no loss in decoding performance. Therefore, in the example of the present application, at time T16, the path with the smallest accumulated value is determined, and the input value at time T1 is decoded retrospectively.

The expression of the computational complexity of the branch metric in the existing LUT-MLSE system is:

The number of real multipliers required for each bit is: N _rm1 =2×(M) ^N /log ₂ M;

The number of real adders required for each bit is: N _ra1 =3×(M) ^N /log ₂ M.

And the branch metric computational complexity of the first technical solution in this application can be expressed as:

The number of real multipliers required for each bit is: N _rm2 =0;

The number of real adders required for each bit is: N _ra2 =3×(M) ^N /log ₂ M.

The branch metric computational complexity of the second technical solution of the present application can be expressed as:

The number of real multipliers required for each bit is: N _rm3 =0;

The number of real adders required per bit is:

Among them, in each of the above expressions, N is the sequence detection length, and M is the modulation order.

In a possible implementation, the obtained complexity of real multiplication and real addition is further refined to the digital hardware circuit level, which is specific to the number of PMOS tubes and NMOS tubes, and binary adders and array multipliers are used respectively. Implements real addition and real multiplication. The present application adopts the full adder of the serial binary two-input terminal to realize the addition operation, and its logical function expression is:

表示异或运算。如图7所示，一个全加器会用到2个异或门和3个逻辑与非门，而在逻辑门电路的实现中，一个异或门采用2个传输门和2个逻辑非门实现，一个传输门用到一个NMOS管和一个PMOS管，一个逻辑非门会用到一个NMOS管和一个PMOS管；一个逻辑与非门会用到2个NMOS管和2个PMOS，因此综上考虑一个全加器会用到14个NMOS管和14个PMOS管。一个基本的S比特二进制串行进位加法器可由S个全加器构成，将会用到(14×S)个NMOS管，(14×S)个PMOS管。Where A and B are the inputs of the adder, C _in is the low-order carry number, Sum is the sum of the bits, C _out is the carry-to-high-order bit, and the sign

Represents an exclusive OR operation. As shown in Figure 7, a full adder will use 2 XOR gates and 3 logical NAND gates, while in the implementation of the logic gate circuit, an XOR gate uses 2 transmission gates and 2 logical NOT gates To achieve, a transmission gate uses an NMOS tube and a PMOS tube, a logic NOT gate uses an NMOS tube and a PMOS tube; a logic NAND gate uses 2 NMOS tubes and 2 PMOS tubes, so in summary Consider a full adder that uses 14 NMOS transistors and 14 PMOS transistors. A basic S-bit binary serial carry adder can be composed of S full adders, which will use (14×S) NMOS tubes and (14×S) PMOS tubes.

For multipliers, there are many types of multipliers in the implementation process of digital circuits, and each has its own advantages and disadvantages. This application selects a P×P array multiplier, as shown in FIG. 8 is the implementation process of a 4×4 array multiplier, Y ₀ is the lowest digit of the multiplier, which is "anded" with the four digits of the multiplicand to obtain a partial product, which is then input into the adder as the summand of each digit, and is carried out with the partial product obtained by the next stage. In the addition operation, the adder will input the carry of the previous bit, and then output the carry of the current stage to the adder of the next bit. Through a reasonably arranged adder array, the process of the multiplication principle can be simulated, and the multiplication result Z can be output. Combined with Figure 8, it can be found that the overall circuit structure uses logic AND gates, full adders and half adders. A half adder is implemented with a logic AND gate and XOR gate. The implementation of the logic AND gate will use 3 NMOS transistors and 3 PMOS tubes, combined with the full adder logic implementation structure mentioned above, so the realization of a P×P array multiplier needs to use 17P ² -21P NMOS tubes and 17P ² -21P PMOS tubes.

With reference to the above description, the number of NMOS and PMOS corresponding to the computational complexity of the branch metric value in the prior art is shown in Table 2, the number of NMOS and PMOS corresponding to the first solution of the present application is shown in Table 3, and the second solution of the present application is as shown in Table 3. The corresponding numbers of NMOS and PMOS are shown in Table 4.

Table 2

table 3

Table 4

In the above distance, when the modulation format is QPSK, the modulation order is M=4, the sequence detection length is N=5, the number of bits of the real multiplier is P=16, and the number of bits of the real adder is S=12. At this time, the existing scheme and the computational complexity of the technical solution of this proposal is:

1) Computational complexity of the current LUT-MLSE system scheme:

The number of real multipliers required for each bit is: N _rm1 =1024

The number of real adders required for each bit is: N _ra1 = 1536

2) Computational complexity of the first technical solution in this application:

The number of real multipliers required for each bit is: N _rm2 =0

The number of real adders required for each bit is: N _ra2 = 1536

3) Computational complexity of the second technical solution in this application:

The number of real multipliers required for each bit is: N _rm3 =0

The number of real adders required for each bit is: N _ra3 =32

The number of NMOS and PMOS corresponding to the computational complexity of the branch metric value in the prior art is shown in Table 5, the number of NMOS and PMOS corresponding to the first solution of the present application is shown in Table 6, and the corresponding NMOS and PMOS of the second solution of the present application. The numbers are shown in Table 7.

table 5

Table 6

Table 7

To sum up, in this application, the branch metric is calculated by separating the real and imaginary parts of the received signal to generate the corresponding real LUTs, and then separately calculating the absolute value of the difference between the real and imaginary parts of the complex signal and the real LUTs. The computational complexity and the number of NMOS and PMOS required to be used are greatly reduced.

In addition, as shown in Figure 9, under the LUT-MLSE coherent optical communication system architecture, we simulated and demonstrated the relationship between the fiber input power and the Q factor of a 32GBd dual-polarization QPSK signal transmitted through a 4GHz narrowband filter for 800km. It can be seen from the experimental results that the operation of calculating the Euclidean distance between complex numbers is replaced by the operation of taking absolute values of the real part and the imaginary part and then adding them together (that is, the first solution in this application), so that System performance has barely changed.

As shown in Figure 10, we experimentally verified the 400km fiber transmission of the 32GBd dual-polarization QPSK signal through a 4GHz narrowband filter. Figure 10 shows that the operation of calculating the Euclidean distance between complex numbers and complex numbers in the current LUT-MLSE technical solution is replaced by a simple operation of taking the absolute value of real numbers (that is, the second solution in this application), so that the Q factor The performance is almost not degraded, and the real multiplier can be omitted, and the number of real adders can be greatly reduced, thereby effectively reducing the complexity and power consumption of the corresponding optical communication integrated circuit chip.

From the comparison between Fig. 9 and Fig. 10, it can be found that the first technical solution proposed in this proposal and its further optimized second technical solution have almost no deterioration in system performance during the experimental verification, that is, the solution provided by this application. The complexity is reduced under the premise that the system is almost unchanged.

To sum up, by receiving a coherent optical signal in a coherent optical communication system; generating an N-symbol distorted signal LUT according to the coherent optical signal; Calculate the branch metric value; and use the N-symbol-based MLSE algorithm to restore the original data in the coherent optical signal according to the calculated branch metric value. The problem of high computational complexity in the prior art is solved, the calculation of the Euclidean distance between complex numbers and the complex numbers is replaced by the operation of calculating the absolute value of the difference between the real numbers and the real numbers, and the required calculation of branch metric values is reduced. Computational complexity, thereby effectively reducing the complexity and power consumption of the corresponding optical communication integrated circuit chip.

The present application also provides an apparatus for optimizing the complexity of an MLSE algorithm based on a lookup table, the apparatus includes a memory and a processor, the memory stores at least one program instruction, and the processor loads and executes the at least one program instruction by loading and executing the at least one program instruction. A program instruction to implement the method described above.

The technical features of the above-described embodiments can be combined arbitrarily. For the sake of brevity, all possible combinations of the technical features in the above-described embodiments are not described. However, as long as there is no contradiction between the combinations of these technical features, All should be regarded as the scope described in this specification.

The above-mentioned embodiments only represent several embodiments of the present invention, and the descriptions thereof are specific and detailed, but should not be construed as a limitation on the scope of the invention patent. It should be pointed out that for those of ordinary skill in the art, without departing from the concept of the present invention, several modifications and improvements can also be made, which all belong to the protection scope of the present invention. Therefore, the protection scope of the patent of the present invention should be subject to the appended claims.

Claims (8)

1. a method for optimizing the complexity of the MLSE algorithm based on look-up table, is characterized in that, described method comprises: In a coherent optical communication system in which the signal at the transmitting end is filtered by a narrow band, the coherent optical signal is received; generating an N-symbol distorted signal LUT according to the coherent optical signal; Calculate the branch metric value according to the received signal, the N-symbol distorted signal LUT and the calculation scheme for optimizing the complexity; According to the calculated branch metric values, the MLSE algorithm based on the N-symbol LUT is used to restore the original data in the coherent optical signal. 2. The method according to claim 1, wherein calculating the branch metric value according to the received signal, the N-symbol distorted signal LUT and the calculation scheme of the optimization complexity, comprises: Obtain the sampling value of the middle column in the N-symbol distortion signal LUT; Calculate the difference between the real part of the received signal and the real part of the acquired sample value and take the absolute value; Calculate the difference between the imaginary part of the received signal and the imaginary part of the obtained sampling value and take the absolute value; The sum of the two calculated absolute values is used as the branch metric value. 3. The method according to claim 2, wherein the sum of the two absolute values obtained by calculation is used as the branch metric value, comprising: The branch metrics are:

where y _k and

are complex numbers, y _k is the received signal,

is the sampled value, real(y _k ) is the real part of the received signal, imag(y _k ) is the imaginary part of the received signal,

is the real part of the sampled value,

is the imaginary part of the sampled value. 4. The method according to claim 1, wherein calculating the branch metric value according to the received signal, the N-symbol distorted signal LUT and the calculation scheme of the optimization complexity comprises: Optimizing the N-symbol distorted signal LUT into a real LUT; Obtain the real sampling value of the middle column in the real LUT; separating the received signal into a real part signal and an imaginary part signal; For each signal obtained by separation, calculate the difference between the real part in each signal and the real sampled value, and calculate the absolute value; The absolute value of the calculated difference is determined as the branch metric value. 5. The method according to claim 4, wherein, determining the absolute value of the calculated difference as the branch metric value, comprising: The branch metrics are:

where y _k and

are real signals, y _k is the real part of each signal obtained by separation,

to obtain the real sample value obtained. 6. The method according to claim 1, wherein the generating an N-symbol distorted signal LUT according to the coherent optical signal comprises: performing analog-to-digital conversion on the coherent optical signal to obtain a digital signal; performing digital signal processing on the digital signal; Separating the training sequence in the processed digital signal; The N-symbol distortion signal LUT is generated according to the training sequence and a LUT training generator. 7. The method according to claim 6, wherein the DSP processing comprises one or more of dispersion compensation, clock recovery, polarization demultiplexing, polarization mode dispersion compensation, frequency offset compensation and phase offset compensation . 8. A device for optimizing the complexity of an MLSE algorithm based on a lookup table, characterized in that the device comprises a memory and a processor, wherein at least one program instruction is stored in the memory, and the processor loads and executes the At least one program instruction to implement a method as claimed in any one of claims 1 to 7.

Images