Low complexity probability shaping 16qam-fso coherent detection dsp method

Abstract

The present application relates to a kind of low complexity probability shaping 16QAM-FSO coherent detection DSP method, belong to optical communication system technical field, including the following steps: S1: by regionalization control matrix, adjust the strategy of bit interleaved coding shaping, after optimization, constellation point is converted into expected distribution;S2: in receiving end, two-dimensional multiple DBSCAN clustering algorithm and Viterbi-Viterbi algorithm are used to carrier phase recovery processing is implemented to shaping signal, reach adaptive phase recovery effect.The present application avoids error diffusion, weakens the performance damage caused by the expansion of constellation along radius after propagation by turbulence;The present application realizes adaptive signal phase recovery effect, effectively improves system performance.The present application has the advantages of low complexity, simple execution, adaptive, etc., can be applied to free space optical communication, fiber communication system or fiber wireless communication system.

Description

Technical Field

[0001] This invention belongs to the field of optical communication system technology and relates to a low-complexity probabilistic shaping 16QAM-FSO coherent detection DSP method. Background Technology

[0002] With the emergence of increasingly more mobile broadband-intensive services and various network applications, such as digital video, high-definition smart TVs, cloud computing, virtual reality, and artificial intelligence, countries around the world are placing higher demands on communication capacity. Radio frequency (RF)-based communication is limited by spectrum resource congestion in wireless networks. Free-space optical communication uses lasers as carriers, without any wired channels as transmission media. Therefore, space optical communication systems have become a research hotspot in the past 15 years due to their advantages such as low power consumption, no need for spectrum authentication, high security, high speed, and rapid link deployment. However, atmospheric turbulence, caused by temperature, pressure, humidity, and other factors, results in irregular flow, leading to refractive index disturbances during optical transmission and altering the phase and amplitude of the light wave. Especially for high-power constellation points of high-order QAM signals, severe interference occurs after passing through the FSO channel, increasing the system's bit error rate and thus degrading transmission performance.

[0003] To overcome this problem, probabilistic shaping techniques have been applied to optical communication systems. Probabilistic shaping can reshape the probability distribution of high-power constellations, thereby significantly reducing the overall average transmit power and mitigating light intensity fluctuations caused by atmospheric turbulence. Currently, the classic constant component distribution matching scheme is commonly used for implementing probabilistic shaping. This scheme relies on forward error correction codes during transmission to eliminate error propagation, resulting in excessively high computational complexity in practical applications. Summary of the Invention

[0004] In view of this, the purpose of this invention is to provide a low-complexity probabilistic shaping 16QAM-FSO coherent detection DSP method. By adding bits 0 to the regionalized data in the interleaver and feeding back to optimize the regional data, a dynamic shaping effect can be achieved. Compared with traditional probabilistic shaping schemes, this method achieves low-complexity, dynamically adjustable shaping.

[0005] To achieve the above objectives, the present invention provides the following technical solution:

[0006] A low-complexity probabilistic shaping 16QAM-FSO coherent detection DSP method includes the following steps:

[0007] S1: By adjusting the bit interleaving coding strategy through the regional control matrix, the constellation points are transformed into the desired distribution after optimization;

[0008] S2: At the receiving end, a two-dimensional multi-DBSCAN clustering algorithm and a Viterbi-Viterbi algorithm are used to perform carrier phase recovery processing on the shaped signal to achieve adaptive phase recovery effect.

[0009] Furthermore, in step S1, dynamic shaping is achieved by adding bits 0 to the regionalized data in the interleaver and feeding back to optimize the regional data; for a 16QAM signal, four bits are mapped to one symbol; then, according to the 16QAM prefix codes 00, 01, 10 and 11, the constellation dot map is divided into three power levels S1, S2 and S3; the pseudo-random binary sequence of serial length M is converted into parallel and input into the interleaver to obtain a 4×N parallel sequence; the first two columns of the parallel sequence are manipulated using a control matrix.

[0010] Furthermore, the set of control matrices is {B1, B2, ..., B}. n}, where B1~B n The dimensions are the same, where B i The resulting control matrix is ​​as follows:

[0011]

[0012] Where N = n(k + m), b i {i=1,…,k} represents Figure 1 The first part of the input information sequence at positions k1 and k2 needs to be processed, i.e., the information data in matrix P, a i {i=k+1,…,k+m} represents the information data in matrix Q. The values ​​of k and m are adjustable. The information at point P is processed by comparing the total number of bits 1 in the upper and lower rows and setting it to 0. The information set to 0 is retained. The information sequence at point Q remains unchanged. The matrix pairs of prefix codes 00, 01, 10, 11 follow a Bernoulli distribution, and their distribution matrix is ​​D=[0.25,0.25,0.25,0.25].

[0013] Furthermore, in step S1, after optimization using the P matrix, let... The minimum contribution matrix for the probabilities of prefix codes 00, 01, 10, and 11:

[0014]

[0015] Maximum contribution matrix:

[0016]

[0017] The dynamic distribution of each prefix code after probabilistic shaping is obtained from D' = D × H:

[0018] D′ min=[0.45,0.1785,0.1785,0.0284]~D′ max =[0.6146,0.25,0.25,0.05]

[0019] exist Given a fixed ratio, we optimize the value of n, where the target probability distribution to be optimized is... The subset of constellations is X = {x1, x2, ..., x} M}, M is the modulation order of the QAM signal, v is the probability distribution factor, and the optimal value of n that approaches the target shaping factor is found through feedback.

[0020] Furthermore, in step S2, the QPSK segmentation algorithm requires obtaining the inner and outer constellation points of 16QAM. The phase estimate is obtained using the Viterbi-Viterbi algorithm, and the estimation formula is as follows:

[0021]

[0022] Where y s1 and y s3 These represent the constellation points of the inner and outer rings of the 16QAM, respectively, and L is the length of the filter for removing additive white Gaussian noise.

[0023] Furthermore, the improved DBSCAN unsupervised clustering algorithm includes the following steps:

[0024] S21: At the receiving end, the 16QAM complex signal without phase estimation is used as a dataset for the DBSCAN clustering algorithm; after DBSCAN clustering, multiple signal star families with different Euclidean distances are obtained, which are used as complex sample datasets S = {S1, S2, ..., S...} N}, where N is the number of cluster races;

[0025] S22: Take the modulus of the obtained complex sample dataset to obtain the racial constellation modulus sample dataset Y = {Y1, Y2, ..., Y}. N In Y, only the amplitude information mod|S| is retained;

[0026] S23: Re-cluster the modulo-based dataset using the DBSCAN clustering algorithm, repeating step S22, to obtain a new population S' = {S'1, S'2, ..., S'...} M The data in the original population S are merged according to the population merging results in S'. This continues until the final complex sample dataset contains only three populations, which are the required constellation points S1, S2, and S3.

[0027] S24: Extract all constellation points from races S1 and S3, and then perform phase estimation on the constellation points using the QPSK algorithm.

[0028] The beneficial effects of this invention are as follows: Based on an optimized scheme for a bit interleaving shaping strategy using a control matrix adjustment, this invention can transform constellation points from a uniform distribution to a desired distribution, avoiding error propagation and mitigating the performance degradation caused by the expansion of the constellation along its radius after turbulent propagation. At the receiving end, constellation points are obtained using an unsupervised clustering algorithm based on DBSCAN, achieving adaptive signal phase recovery and effectively improving system performance. This invention has advantages such as low complexity, simple execution, and adaptability, and can be applied to free-space optical communication, fiber optic communication systems, or fiber optic wireless communication systems.

[0029] Other advantages, objectives, and features of the invention will be set forth in part in the description which follows, and in part will be apparent to those skilled in the art from the following examination, or may be learned from practice of the invention. The objectives and other advantages of the invention can be realized and obtained through the following description. Attached Figure Description

[0030] To make the objectives, technical solutions, and advantages of the present invention clearer, the preferred embodiments of the present invention will be described in detail below with reference to the accompanying drawings, wherein:

[0031] Figure 1 The structure diagram of the low-complexity probabilistic integer shaping algorithm of this invention is shown below;

[0032] Figure 2 The image shows the 16QAM clustering result obtained using the DBSCIN clustering algorithm;

[0033] Figure 3 (a) shows the standard 16QAM signal at the signal receiver, and (b) shows the shaped 16QAM signal.

[0034] Figure 4 This is a schematic diagram of the signal transmitting end structure of the present invention;

[0035] Figure 5 This is a schematic diagram of the signal receiving end structure of the present invention;

[0036] Figure reference numerals: 1-Laser, 2-I / Q modulator, 3-Offline probabilistic shaping (PS) digital signal processing module, 4-Erbium-doped fiber amplifier (EDFA), 5-Local laser, 6-90° optical mixer, 7-First balanced detector, 8-Second balanced detector, 9-Receiver line digital signal processing module. Detailed Implementation

[0037] The following specific examples illustrate the implementation of the present invention. Those skilled in the art can easily understand other advantages and effects of the present invention from the content disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and various details in this specification can be modified or changed based on different viewpoints and applications without departing from the spirit of the present invention. It should be noted that the illustrations provided in the following embodiments are only schematic representations of the basic concept of the present invention. Unless otherwise specified, the following embodiments and features can be combined with each other.

[0038] The accompanying drawings are for illustrative purposes only and are schematic diagrams, not actual pictures. They should not be construed as limiting the invention. To better illustrate the embodiments of the invention, some parts in the drawings may be omitted, enlarged, or reduced, and do not represent the actual product dimensions. It is understandable to those skilled in the art that some well-known structures and their descriptions may be omitted in the drawings.

[0039] In the accompanying drawings of the embodiments of the present invention, the same or similar reference numerals correspond to the same or similar components. In the description of the present invention, it should be understood that if terms such as "upper," "lower," "left," "right," "front," and "rear" indicate the orientation or positional relationship based on the orientation or positional relationship shown in the drawings, they are only for the convenience of describing the present invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, the terms used to describe positional relationships in the drawings are only for illustrative purposes and should not be construed as limiting the present invention. For those skilled in the art, the specific meaning of the above terms can be understood according to the specific circumstances.

[0040] To combat atmospheric turbulence and simplify computational complexity, this invention provides a low-complexity probabilistic shaping coherent reception scheme. Probabilistic shaping, through a regionalized control matrix and optimized bit interleaving coding strategy, can transform constellation points from a uniform distribution to a desired distribution, avoiding error propagation and mitigating performance degradation caused by constellation radius expansion after turbulence propagation. Furthermore, it achieves shaping gain with low computational complexity, realizing a dynamic probabilistic shaping distribution. At the receiver, a two-dimensional multi-DBSCAN clustering algorithm and the Viterbi-Viterbi algorithm are used to perform carrier phase recovery processing on the shaped signal, achieving adaptive phase recovery.

[0041] This invention achieves dynamic shaping by adding bits 0 to the regionalized data in the interleaver and feeding back to optimize the regional data. Compared with traditional probabilistic shaping schemes, it realizes low-complexity dynamic adjustable shaping.

[0042] For a 16QAM signal, four bits are mapped to one symbol. Then, based on the 16QAM prefix codes 00, 01, 10, and 11, the constellation dot pattern is divided into three power levels, S1, S2, and S3. A pseudo-random binary sequence of serial length M is converted to parallel and input into an interleaver to obtain a 4×N parallel sequence. The control matrix manipulates the first two columns of the parallel sequence.

[0043] The control matrix set is {B1, B2, ..., B} n}, where B i (i = 1, 2, ..., n) have the same dimension, B i The resulting control matrix is ​​as follows:

[0044]

[0045] Where N = n(k + m), b i {i=1,…,k} represents Figure 1 The first part of the input information sequence at positions k1 and k2 needs to be processed, i.e., the information data in matrix P, a i {i=k+1,…,k+m} represents the information data in the Q matrix, where the values ​​of k and m are adjustable. Information at point P is processed by comparing the total number of 1 bits in the upper and lower rows and setting them to 0; information with 0 bits is retained. The information sequence at Q remains unchanged. The matrix pairs 00, 01, 10, 11 follow a Bernoulli distribution with a distribution matrix D=[0.25,0.25,0.25,0.25]. After optimization of the P matrix, considering that constellation diagrams have different probability distributions, let… The minimum contribution matrix for the probabilities of prefix codes 00, 01, 10, and 11:

[0046]

[0047] Maximum contribution matrix:

[0048]

[0049] From D' = D × H, we can obtain the dynamic distribution of each prefix code after probability shaping:

[0050] D′ min =[0.45,0.1785,0.1785,0.0284]~D′ max =[0.6146,0.25,0.25,0.05]

[0051] exist Given a fixed ratio, optimize the value of n. Figure 1 This is an example diagram of a probabilistic integer optimization algorithm, where the target probability distribution to be optimized is... The subset of constellations is X = {x1, x2, ..., x} M}, where M is the modulation order of the QAM signal, v is the probability distribution factor, and A v Ensure the assigned probability P X (x i The sum of ) is 1, and the optimal value of n that approaches the target shaping factor is found through feedback.

[0052] After probabilistic shaping, the 16QAM signal will be amplified to a certain extent. The traditional Viterbi-Viterbi algorithm obtains the segmentation amplitude threshold of the 16QAM signal based on the maximum a posteriori algorithm. Since the phase estimation algorithm depends on the position of the constellation points, the decision threshold varies with the degree of shaping. Therefore, this invention estimates the phase noise by combining the two-dimensional multi-DBSCAN clustering algorithm and the Viterbi-Viterbi algorithm. The advantage of this method is that it does not require the calculation of the decision amplitude value and can automatically track and divide the regional signal constellation points, achieving an adaptive effect.

[0053] The QPSK segmentation algorithm requires the inner and outer constellation points of 16QAM. The phase estimate is obtained through the Viterbi-Viterbi algorithm, and the estimation formula is as follows:

[0054]

[0055] in and These represent the constellation points of the inner and outer rings of the 16QAM, respectively, and L is the length of the filter for removing additive white Gaussian noise.

[0056] The advantage of the DBSCAN clustering algorithm over the K-means clustering algorithm is that it can cluster dense datasets of arbitrary shapes and is applicable to both convex and non-convex sample sets.

[0057] In carrier phase recovery processing, the received signals are used to form a dataset with neighborhood parameters {ε, MinPts}, where ε describes the neighborhood distance threshold of a sample and MinPts describes the number of samples in the neighborhood of a sample at a distance of ε.

[0058] like Figure 4 As shown, at the transmitting end, the offline PS digital signal processing module 3 generates the PS-16QAM baseband signal as described below. As the electrical drive signal for the in-phase / quadrature (I / Q) modulator; a(t) and These are the amplitude and phase information of the 16QAM signal, respectively; A is the receiver responsivity, and laser 1 is in f... c A continuous light wave is emitted at a center frequency of 1550 nm, where Δf = f c -flo f lo The frequency of the local oscillator is 1550.001 nm; P represents the total phase noise of the laser. s and P l0 These represent the optical powers of the emitting laser and the local oscillator, respectively. After passing through the balanced detector 7, the resulting signal is:

[0059]

[0060] like Figure 5 As shown, the receiver consists of a local laser 5, a 90° optical mixer 6, a first balanced detector 7, and a second balanced detector 8 detection circuit. After processing by the receiver line digital signal processing module 9, the 16QAM amplitude and phase information is recovered. The DSP processing flow is resampling, clock recovery, frequency offset estimation, and an improved carrier phase estimation algorithm.

[0061] In carrier phase recovery processing, the received signals are constructed into a dataset with neighborhood parameters {ε, MinPts}, where ε describes the neighborhood distance threshold between 16QAM constellation points in the constellation samples, and MinPts describes the number of samples in the neighborhood of a given sample at a distance of ε. Figure 2 The following are the specific implementation steps for using the clustering results obtained from DBSCAN clustering:

[0062] Before the carrier recovery algorithm, 16QAM is divided into three circles based on power levels. The received signals form a dataset, and the DBSCAN algorithm is used to cluster the three-circle dataset. The specific clustering process is to first set the neighborhood parameter ε and the value of MinPts, where ε describes the neighborhood distance threshold between 16QAM constellation points in the constellation samples, and MinPts describes the number of samples in the neighborhood with a distance of ε. The clustering process is divided into two stages. The first stage is to set ε and MinPts for the obtained constellation points. In this process, the neighborhood ε is made as small as possible and the number of samples in the neighborhood MinPts is made as large as possible. The reason for this is that DBSCAN clusters based on the density of data points, and the initial processing yields more cluster families S = {S1, S2, ..., S...}. NWhen channel noise is high, the middle and outer ring data of 16QAM are severely affected by noise and overlap. In this case, this initial processing method can obtain more constellation families, allowing for the differentiation of characteristics between each family and reducing the interference of constellation overlap on constellation points of different amplitudes. The second stage involves taking the modulus (mod|S|) of the N family signals from the 16QAM signals partitioned in the first stage. Then, the mean (mod|S|) is calculated, resulting in N signal points. This dataset (i.e., the N signal points) is then re-clustered using the DBSCAN clustering algorithm. The resulting clusters are S' = {S'1, S'2, ..., S'...}. M}, where M < N, and S' population is the population formed by merging populations with similar modulus characteristics from the original S population. Next, we need to merge the data in the original population S according to the result of merging S'. This continues until three populations are obtained in the S population, namely S = {S1, S2, S3}. At this point, S1, S2, and S3 represent the three power levels of the 16QAM constellation points. The flowchart of the DBSCAN clustering algorithm is shown in Table 1.

[0063] Table 1

[0064]

[0065]

[0066] The output population contains three clusters S = {S1, S2, S3}, thus the clustering is complete; the clustering results are as follows. Figure 2 As shown.

[0067] Substituting the constellation points of S1 and S3 after clustering into the formula Phase estimation is performed, and then the receiver demodulates the original information sequence. For example... Figure 3 As shown, (a) is the standard 16QAM signal at the signal receiver, and (b) is the shaped 16QAM signal diagram.

[0068] Those skilled in the art will understand that all or part of the steps in the methods of the above embodiments can be implemented by a program instructing related hardware. The program can be stored in a computer-readable storage medium. When the program is executed, it can implement the steps of the method. The storage medium may be, for example, ROM / RAM, magnetic disk, optical disk, etc.

[0069] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A low-complexity probabilistic shaping 16QAM-FSO coherent detection DSP method, characterized in that: Includes the following steps: S1: By adjusting the bit interleaving coding strategy through the regional control matrix, the constellation points are transformed into the desired distribution after optimization; S2: At the receiving end, the two-dimensional multi-DBSCAN clustering algorithm and the Viterbi-Viterbi algorithm are used to perform carrier phase recovery processing on the shaped signal to achieve adaptive phase recovery effect; In step S1, by adding bits 0 to the regionalized data in the interleaver, the regional data is optimized to achieve dynamic shaping. For a 16QAM signal, four bits are mapped to one symbol. Then, based on the 16QAM prefix codes 00, 01, 10, and 11, the constellation dot map is divided into three power levels. ; Serial length is The pseudo-random binary sequence is converted into parallel data and input into an interleaver to obtain... Parallel sequences; manipulating the first two columns of a parallel sequence using a control matrix; The control matrix set is ,in B 1~ B n The dimensions are the same, among which The resulting control matrix is ​​as follows: in , express Information data in the matrix, express Information data in the matrix, and The value is adjustable. The information is processed by comparing the total number of 1 bits in the upper and lower rows and setting it to 0. The information that is set to 0 is retained. The information sequence remains unchanged; the prefix codes 00, 01, 10, and 11 in the matrix follow a Bernoulli distribution, and their distribution matrix is... ; In step S1, through After matrix optimization, let The minimum contribution matrix for the probabilities of prefix codes 00, 01, 10, and 11: Maximum contribution matrix: Depend on The dynamic distribution of each prefix code after probabilistic shaping is obtained: exist Given a fixed ratio, for The value is optimized, where the target probability distribution to be optimized is... The constellation subset is , The modulation order of the QAM signal. For probability distribution factor, Ensure the assigned probability The sum is 1, and the optimal value that approaches the target shaping factor is found through feedback. value.

2. The low-complexity probabilistic shaping 16QAM-FSO coherent detection DSP method according to claim 1, characterized in that: In step S2, the QPSK segmentation algorithm requires obtaining the inner and outer constellation points of 16QAM. The phase estimate is obtained using the Viterbi-Viterbi algorithm, and the estimation formula is as follows: in and These represent the constellation points of the inner and outer rings of the 16QAM, respectively. The length of the filter is to remove additive white Gaussian noise.

3. The low-complexity probabilistic shaping 16QAM-FSO coherent detection DSP method according to claim 2, characterized in that: Carrier phase recovery based on the improved DBSCAN algorithm includes the following steps: S21: At the receiving end, the 16QAM complex signal without phase estimation is used as a dataset for the DBSCAN clustering algorithm; After DBSCAN clustering, multiple signal star populations with different Euclidean distances were obtained, which served as a complex sample dataset. ,in The number of races in the cluster; S22: Take the modulus of the obtained complex number sample dataset to obtain the racial constellation modulus sample dataset. , Only amplitude information is retained. ; S23: Re-cluster the modulo-based dataset using the DBSCAN clustering algorithm, repeating step S22, to obtain a new population. , the original population The data in The results of population merging are used to perform race merging; this continues until only three races remain in the final complex sample dataset, which is the desired result. , and Constellation points; S24: Remove and All constellation points in the race are then used to estimate the phase of the constellation points using the QPSK algorithm.

Images