Transmission digital signal generating circuit and optical transmitter

Abstract

An optical transmitter according to the technology of the present disclosure converts the probability of generation of a symbol (X) into a histogram, and determines the allocation of a DAC (121) on the basis of the cumulative frequencies of the histogram. However, the class value of the histogram is the value of the symbol (X), the frequencies are nonzero in any class, and the aggregate of frequencies is the quantization value of the DAC (121).

Description

[Technical field]

[0001] The present disclosure relates to a transmission digital signal generating circuit and an optical transmitter. [Background technology]

[0002] The transmission digital signal generating circuit according to the present disclosure and an optical transmitter having the same as a component thereof are applied to, for example, optical fiber communication in a backbone network (hereinafter referred to as "backbone large-capacity optical fiber communication"). One of the pioneering signal processing technologies in the technical field of backbone large-capacity optical fiber communication is "compression shaping," developed by the applicant.

[0003] Simply put, "compression shaping" is a technology that combines data compression, which is source coding, and Probabilistic Constellation Shaping (hereinafter simply referred to as "PS"), which is channel coding. The theory of compression shaping is described in, for example, Non-Patent Document 1. [Prior art documents] [Non-patent literature]

[0004] [Non-Patent Document 1] Tsuyoshi Yoshida and Koji Igarashi, "Probability Distribution Shaping and Pseudo Data Compression in Optical Fiber Communications," IEICE Transactions on Information and Communication Engineers, Vol. J103-B, No. 9, pp. 361-371, 2020 Summary of the Invention [Problem to be solved by the invention]

[0005] As mentioned above, since compression shaping is a recent pioneering technology, the implementation or application of compression shaping has not been discussed much yet. For example, when compression shaping is applied, data compression is performed when communication traffic decreases, and the average symbol energy of the transmitted data symbol changes. At this time, the output signal quality may deteriorate or the downstream system may become unstable.

[0006] The disclosed technique proposes compression shaping that deals with changes in average symbol energy, and aims to provide a transmission digital signal generation circuit and an optical transmitter that implement more practical compression shaping. [Means for solving the problem]

[0007] The optical transmitter according to the present disclosure comprises: Converted into a digital signal for transmission and input to the DAC The probability of occurrence of symbol (X) is plotted as a histogram, and based on the cumulative frequency of the histogram, Assign a range of DAC output values ​​to a symbol (X) The allocation of the DAC is determined, where the bin values ​​of the histogram are the values ​​of the symbols (X), the frequency in every bin is non-zero, and the total frequency is the quantization number of the DAC. Effect of the Invention

[0008] Since the transmission digital signal generation circuit according to the disclosed technique has the above configuration, it is possible to perform compression shaping that corresponds to changes in average symbol energy. [Brief description of the drawings]

[0009] [Figure 1] FIG. 1 is a system model showing an optical fiber communication system. [Diagram 2] FIG. 2 is a diagram for explaining compression shaping, which is an example of a functional block configuration. [Diagram 3] FIG. 3 is a second diagram illustrating compression shaping, showing an example of symbol probability distribution. [Figure 4]FIG. 4 is a block diagram of a functional configuration of the optical transmitter according to the first embodiment. [Diagram 5] FIG. 5 is a diagram showing an example of numerical values ​​of digital amplitude adjustment in an optical transmitter according to the conventional technology. [Figure 6] FIG. 6 is a diagram showing a first numerical example of digital amplitude adjustment in an optical transmitter with compressive shaping. [Figure 7] FIG. 7 is a second diagram showing a numerical example of digital amplitude adjustment in an optical transmitter with compressive shaping. [Figure 8] FIG. 8 is a diagram illustrating an example in which undesirable clipping occurs as a result of digital amplitude adjustment. [Figure 9] FIG. 9 is a diagram for explaining the allocation of DACs and ADCs to a symbol (X) taking into account the occurrence probability. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0010] (acronym) The major alphabetic acronyms used in this specification are as set forth in the table below. TIFF0007682403000001.tif63166

[0011] (Introduction 1, Compression Shaping) Generally, in backbone networks, source coding, which has a data compression effect, is not performed; instead, 0 / 1 is made uniform by bit scrambling, and only channel coding such as forward error correction (hereafter referred to as "FEC") or probability distribution shaping (hereafter referred to as "PS") is performed, assuming a uniformly distributed input. However, since the effective amount of communication traffic fluctuates, it would be of great benefit if source coding capable of real-time data compression could be realized. The basic concept of compression shaping is to combine source coding and channel coding, in other words to realize joint source-channel coding.

[0012] Figure 1 is a system model showing an optical fiber communication system. S in Figure 1 is an information bit, expressed in binary. As shown in Figure 1, an information bit (S) is divided into a code information bit (S s ) and the amplitude information bit (S a ) and the amplitude information bit (S a ) are PS-encoded in k [bits] chunks. The encoded symbol is represented as A. The number of A's obtained by PS encoding is n [symbols]. For simplicity's sake, the symbol (X) sent to the communication channel is assumed to be a one-dimensional real number, two of which are bundled together to form a complex symbol, and two complex symbols are further bundled together to form a polarization multiplexed complex symbol. The number of bits that make up the symbol (X) sent to the communication channel is assumed to be m. Also, A is assumed to be equal to the absolute value of X. For example, in the case of polarization multiplexed 16-QAM, m=2. A is the bit string (B ps , m-1-dimensional binary). The sign information bit (S s ) and PS encoded bit string (B ps )(total k c k [bits] are systematically FEC coded as information bits. c For [bit] FEC information bits, (n c -k c ) [bits] of parity bits are added. A symbol (X) is generated from the FEC-encoded bit string (B, m-dimensional binary). Of the bit string (B), the bit (B sign ) is the sign information bit (S s ) and FEC parity bits. The bit string that controls the absolute value of the symbol (X), A, is B ps It consists only of. We may assume that the channel can be approximated by discrete memoryless AWGN. Transmitted symbols (X) are transformed into received symbols (Y).

[0013] FIG. 2 is the first diagram explaining compression shaping, and is an example of a functional block configuration. The acronyms that appear in FIG. 2 can be referred to in Table 1 (table for explanation of acronyms) for their full names. S in FIG. 2 represents an information bit. S, which is a part of S, S is bit-scrambled and assigned to the sign bit. The sign bit is used to determine the coordinates of the QAM symbol. The probability distribution of the FEC parity bits cannot be controlled, and generally "0" and "1" are distributed uniformly. Therefore, the FEC parity bits are assigned to the sign bit. The remaining information bits, S a As pre-processing, the following conditional branching processing is performed. Condition: There are many "1"s in a given unit length. YES: All input bits are inverted and parity "1" is added before output. NO: Add parity "0" to output This corresponds to all inputs to this function becoming "1" when an alarm signal is sent.

[0014] The block shown as "Hierarchical DM" in Fig. 2 performs PS by implementing hierarchical distribution matching (hereinafter referred to as "hierarchical distribution matching"). As shown in Fig. 2, the hierarchical distribution matching is composed of a large number of small-scale LUTs. The input bit strings and output bit strings stored in each LUT are sorted, for example, as shown in the following table. TIFF0007682403000002.tif55166, where P S (1) is the probability that S gets “1”, also called the mark rate. Table 2 shows the case where the mark rate is 0.3. As shown in Table 2, the input bit string is sorted so that the number of "1"s (Hamming weight = number of non-zero components) contained therein is in ascending order. Also, the output statistics are sorted so that, for example, the expected value of the average symbol energy (E) is in ascending order. This sorting is performed before the signal is communicated. The main signal is communicated after writing to the LUT is complete. The average symbol energy (E) is defined as the expected value of A squared. In other words, the average symbol energy (E) is calculated as the average of the squared distances from the origin. For example, in the second example of Table 2, A={1,1,1,3}, the average symbol energy (E) is “(1 2 +1 2 +1 2 +3 2 )÷4=3”. Pr in Table 2 is P S Represents the probability of each case calculated based on (1).

[0015] Figure 3 is a second diagram for explaining compression shaping, which is an example of a symbol probability distribution. More specifically, Figure 3 is an example of a PS-128-QAM symbol probability distribution based on a bit string related to the modulation symbol amplitude (A) of the hierarchical distribution matching output. The left side of Figure 3 shows the source mark rate, P S (1) shows the case where the information entropy (H(S)) is 50%. The right side of Figure 3 shows the source mark rate, P S (1) is 20%, i.e., the information entropy (H(S)) is 0.72. When the information source is relatively sparse like this, the probability distribution is more concentrated in the center. The example on the right side of Figure 3 has a larger kurtosis than the example on the left side. As shown in FIG. 3, in compression shaping, as the information entropy (H(S)) decreases, the symbol entropy (H(A)) and average symbol energy (E) also decrease, and the SNR required to obtain communication quality that meets the specifications can be reduced. The term "compression" in compression shaping does not mean reducing the number of bit slots. In compression shaping, the symbol entropy and average symbol energy can be reduced as if the number of bit slots expressing the information source had been reduced (compressed), hence the terms "compression" or "pseudo-compression". In addition, compression shaping is sometimes referred to as a technology that performs "instantaneous (real-time) compression" because it does not require a large-scale storage device or cause processing delays.

[0016] (Introduction 2, Pilot Symbol) In the art of optical modulation, a symbol refers to a designation associated with a "state" of a light wave (e.g., X, Y, etc., as seen in Introduction Part 1). A commonly known pilot symbol is a known symbol that is prepared in advance and is used for detecting cycle slips (for example, International Publication No. WO 2010 / 138198). On the other hand, although known symbols are also used in the disclosed technology (see, for example, known symbol generation unit 112 and known symbol insertion unit 113 described below), the known symbols in the disclosed technology are used in relation to the average symbol energy, which is the central theme of the disclosed technology. Note that the known symbols in the disclosed technology are essentially the same as pilot symbols, and therefore the known symbols can be used as general pilot symbols in addition to the uses specific to the disclosed technology.

[0017] Embodiment 1 Fig. 4 is a block diagram showing a functional configuration of the optical transmitter 100 according to the first embodiment. As shown in Fig. 4, the optical transmitter 100 according to the first embodiment includes a transmission digital signal generating circuit 110, a DAC 121, a light source 122, an optical modulation unit 123, and an optical amplification unit 124. The transmission digital signal generating circuit 110 included in the optical transmitter 100 according to the first embodiment is made up of an encoding section 111 , a known symbol generating section 112 , a known symbol inserting section 113 , an amplitude adjusting section 114 , and a gain control section 115 . The functional components of the optical transmitter 100 according to the first embodiment are connected as shown in FIG. It can be said that a special technical feature of the optical transmitter 100 according to the disclosed technique is that it includes a gain control section 115 .

[0018] <<Encoding unit 111 constituting transmission digital signal generating circuit 110>> The encoding unit 111 constituting the transmission digital signal generation circuit 110 is a component that generates transmission data symbols based on external information. As shown in Fig. 4, the encoding unit 111 outputs "statistical information" to the gain control unit 115. The statistical information output by the encoding unit 111 is statistical information related to the transmission data symbols that change every moment. Examples of statistical information intended by the present disclosure include average mark rate, entropy, probability distribution, and modulation format information. Details of the statistical information will become clear from the explanation based on the numerical examples described below.

[0019] <<Known Symbol Generator 112 Constituting Transmission Digital Signal Generator 110>> The known symbol generator 112 constituting the transmission digital signal generator 110 is a component that generates known symbols to be periodically inserted between two adjacent transmission data symbols. As described in Introduction 2, the known symbols generated by the known symbol generator 112 are essentially the same as pilot symbols.

[0020] <<Known symbol insertion unit 113 constituting transmission digital signal generation circuit 110>> Known symbol insertion section 113 constituting transmission digital signal generation circuit 110 is a component that periodically inserts a known symbol between two adjacent transmission data symbols. The known symbol inserting unit 113 inserts known symbols at a pre-designed cycle. The frequency at which the known symbols appear may be lower than the frequency at which the transmission data symbols appear. Also, the frequency at which the known symbols appear may be increased in order to increase the probability that the receiver can correctly detect the signal. In this specification, the transmission data symbols and known symbols combined in known symbol insertion section 113 are collectively referred to as "transmission symbols."

[0021] <<Amplitude adjustment unit 114 constituting transmission digital signal generation circuit 110>> The amplitude adjustment unit 114 constituting the transmission digital signal generation circuit 110 is a component that converts a transmission symbol into a transmission digital signal so that the transmission symbol is converted into an analog electrical signal of an appropriate amplitude in the subsequent DAC 121. In this specification, the process performed by the amplitude adjustment unit 114 is referred to as "digital amplitude adjustment." The amplitude adjustment unit 114 performs digital amplitude adjustment based on gain information obtained by a gain control unit 115, which will be described later. Details of the digital amplitude adjustment performed by the amplitude adjustment unit 114 will also become clear from the explanation based on numerical examples, which will be described later. In conventional trunk-system large-capacity optical fiber communications, a fixed gain coefficient is used for digital amplitude adjustment, and the gain does not change dynamically. On the other hand, the transmission digital signal generation circuit 110 and the optical transmitter 100 according to the disclosed technology perform digital amplitude adjustment dynamically. It can be said that a special technical feature of the optical transmitter 100 according to the disclosed technology is that it performs digital amplitude adjustment dynamically.

[0022] <Gain control unit 115 constituting transmission digital signal generating circuit 110> The gain control section 115 constituting the transmission digital signal generating circuit 110 is a component that dynamically generates gain information based on statistical information that changes from moment to moment. Since the gain information generated by the gain control section 115 changes dynamically in this manner, the process performed by the gain control section 115 is referred to as "amplification gain control" or simply "gain control." The gain control performed by the gain control section 115 enables dynamic digital amplitude adjustment.

[0023] 《DAC121》 The DAC 121 is a component that converts a transmission digital signal into an analog electrical signal. The DAC 121 is a digital-to-analog converter.

[0024] 《Light source 122》 The light source 122 is a component that emits coherent continuous light to realize optical fiber communication. The light source 122 is typically a light-emitting device such as a light-emitting diode (LED), a semiconductor laser (LD), or a quantum well (QW) laser.

[0025] <<Light Modulation Section 123>> The optical modulation unit 123 is a component that performs optical modulation on the continuous light emitted by the light source 122 based on the analog electrical signal output by the DAC 121 . In the backbone large-capacity optical fiber communication, for Beyond 5G, multi-values ​​such as, for example, polarized MIMO using 256QAM are expected. Therefore, the optical modulation unit 123 according to the present disclosure is designed to be fully compatible with this multi-value. The signal obtained by the optical modulation in the optical modulation section 123 is called an optical signal. The optical signal is sent to the optical amplification section .

[0026] Optical amplifier 124 The optical amplifier 124 is a component that optically amplifies an optical signal. The optical amplifier 124 is an optical amplifier. The optical signal optically amplified in the optical amplifier 124 is output to a transmission path. In the case of optical fiber communication, the transmission path is an optical fiber.

[0027] (Numerical example, Figure 5) Fig. 5 is a diagram showing a numerical example of digital amplitude adjustment in an optical transmitter according to the prior art. In the numerical example shown in Fig. 5, the first column from the left represents a symbol (X). This numerical example assumes 8PAM with symbols {-7, -5, -3, -1, 1, 3, 5, 7}.

[0028] 5, the second column from the left represents the probability of occurrence of the symbol (X). In the conventional optical transmitter, the probability of occurrence of the symbol (X) has been assumed to be uniform. 5, the first row from the top shows the case where the symbol (X) is "-7". Taking the case where the symbol (X) is "-7" as an example, the occurrence probability is given by the following formula: TIFF0007682403000003.tif9166In this way, the letter P is used to represent functions of probability.

[0029] In the numerical example shown in FIG. 5, the third column from the left shows the amplitude before amplitude adjustment (A 0 ) before amplitude adjustment (A 0 ) is expressed by the following formula: TIFF0007682403000004.tif14166However, k 0 represents the default coefficient, and q represents half the quantization number of the DAC. In the numerical example shown in FIG. 5, the default coefficient (k 0 ) is 14, the DAC is 8 [bit], and q = 256 ÷ 2. When the symbol (X) is “-7”, the amplitude (A 0 Specifically, the calculation is as follows: TIFF0007682403000005.tif29166Note, A 0 The subscript "0" used in 0 The "0" in the lower right subscript used in the above is also a sequential number to distinguish it from other situations that will appear later, and has no other meaning. The simplest definition of amplitude is that it is proportional to the symbol (X), as seen in Equation (2) and Equation (3). A transformation given only in proportion in this way is a linear transformation. Although it differs from the numerical example in Equation (3), the simplest way to think of it is to perform a linear transformation in which the amplitude corresponding to the most positive symbol (X=7) is set to 1, and the amplitude corresponding to the opposite negative symbol (X=-7) is set to -1.

[0030] In the numerical example shown in FIG. 5, the fourth column from the left is the power before amplitude adjustment (E 0 ) Strictly speaking, it is not possible to measure the power of a digital symbol (X). Power here means the value obtained by squaring the amplitude, just like the average symbol energy. Incidentally, this definition comes from the fact that the average power of a sinusoidal signal is proportional to the square of the signal's amplitude. The power before amplitude adjustment (E 0 ) is given by the following formula: TIFF0007682403000006.tif9166 When the symbol (X) is "-7", the power before amplitude adjustment (E 0 Specifically, the calculation is as follows: TIFF0007682403000007.tif15166

[0031] Now, the expected value of the power before the amplitude adjustment (hereinafter referred to as the "average power of the input signal") is given by the following formula. TIFF0007682403000008.tif16166However, the E in script typeface represents the expected value. In the numerical example shown in FIG. 5, the average power of the input signal is: TIFF0007682403000009.tif10166

[0032] In the numerical example shown in FIG. 5, the fifth column from the left shows the amplitude after amplitude adjustment (A 1 ), and the sixth column from the left is the amplitude-adjusted power (E1 ), respectively. The digital amplitude adjustment is done by the default coefficient (k 0 Considering the analogy with equation (2), the amplitude after amplitude adjustment (A 1 ) is given by the following formula: TIFF0007682403000010.tif15166A 1 The subscript "1" used in 1 The "1" in the lower right subscript used in the above is also a sequential number to distinguish it from other situations that will appear later, and has no other meaning. k 1 The method for determining will become clear from the explanation given below.

[0033] Power before amplitude adjustment (E 1 ) is given by the following formula, similar to formula (4). TIFF0007682403000011.tif10166

[0034] The expected value of the power after the amplitude adjustment (hereinafter referred to as the "average power after the amplitude adjustment") is given by the following formula. TIFF0007682403000012.tif15166

[0035] In the numerical example shown in FIG. 5, digital amplitude adjustment is performed so that the average power after the amplitude adjustment is 0.25. In this case, k 1 The value can be determined as follows: TIFF0007682403000013.tif36166

[0036] In the numerical example shown in FIG. 5, the seventh column from the left shows the amplitude before quantization (B 1 ) represents the amplitude before quantization (B 1 ) is given by the following formula: TIFF0007682403000014.tif10166Equation (12) is the amplitude after amplitude adjustment (A 1 ) is considered to be full scale from -1.0 to +1.0 and is assigned to the conversion of DAC 121. This can be understood as, for example, the output voltage range of DAC 121 specifications being from -1.0 [V] to +1.0 [V]. In the case where the symbol (X) is “-7”, the amplitude before quantization (B 1 Specifically, the calculation is as follows: TIFF0007682403000015.tif24166

[0037] In the numerical example shown in FIG. 5, the 12th column from the left shows the amplitude after quantization (B 2 ) represents the amplitude after quantization (B 2 ) is given by the following formula: TIFF0007682403000016.tif10166Here, int() on the right side of formula (14) is a function that rounds down to obtain an integer, and sign() is a sign function that returns +1, -1, or 0 depending on the sign of the argument. In the numerical example shown in FIG. 5, the eighth to eleventh columns from the left simply perform the transformation shown in equation (14) in stages.

[0038] In the numerical example shown in FIG. 5, the 13th column from the left shows the amplitude after quantization (B 2 ) does not exceed the quantization number of the DAC. In the example shown in FIG. 5, q=128, so if the amplitude after quantization (B 2 If the amplitude (B) exceeds 127.5, it is set to 127.5. 2 The same process is performed when ) is negative. In other words, the 13th column from the left simulates saturation in the DAC. In the numerical example shown in FIG. 5, none of the amplitudes exceed the quantization number of the DAC. Although the values ​​are the same, the amplitudes after quantization (B 2 ) to distinguish it from the amplitude after saturation judgment, B 3 The amplitude after saturation judgment (B 3 ) is given by the following formula: TIFF0007682403000017.tif19166Here, the sign() appearing in formula (15) is the sign function, just like the one appearing in formula (14).

[0039] In Figure 5, there is a section that displays "Correlation: 0.999993." The correlation displayed here is the amplitude after amplitude adjustment (A 1 ) and the amplitude after quantization (B 3 ) is the correlation with the amplitude after amplitude adjustment (A 1 ) and the amplitude after quantization (B 3 ) is given by the following formula: TIFF0007682403000018.tif18166Note that the accent symbol in the horizontal bar in formula (16) represents the sample average. Correl() on the left side of formula (16) is a function that calculates correlation. The value obtained by the calculation shown on the right side of formula (16) is sometimes called the correlation coefficient. The closer the correlation or correlation coefficient is to +1 or -1, the greater the correlation between the arrays (in this case, A 1 and B 3 It indicates a positive (+1) or negative (-1) correlation between the

[0040] The optical signal is transmitted through a transmission line and sampled by an ADC (Analog Digital Converter) at the receiver. The receiver finally obtains an estimate of the symbol (X) based on the digital signal obtained by sampling with the ADC. The transmitter and the receiver perform optical modulation and demodulation based on a common rule, for example, the content defined by various standards. If the receiver side can correctly estimate the symbol (X) based on this common rule, the symbols arranged in the constellation map do not necessarily have to be arranged at equal intervals.

[0041] (Numerical example, FIG. 6) FIG. 6 is a first diagram showing a numerical example of digital amplitude adjustment in the optical transmitter 100 that performs compression shaping. As can be seen by comparing with FIG. 5, in the numerical example of FIG. 6, the appearance probability of the symbol (X) is not uniform. The appearance probability of the symbol (X) shown in FIG. 6 is specifically given by the following mathematical formula. TIFF0007682403000019.tif15166Here, the numerator of the mathematical formula (17) represents a Gaussian distribution. 0.02 appearing in the mathematical formula (17) is a parameter related to the variance of the Gaussian distribution. In the case where the symbol (X) is "-7", the appearance probability is specifically calculated as follows. TIFF0007682403000020.tif22166

[0042] Also in the numerical example shown in FIG. 6, the average power of the input signal is given by the mathematical formula (6). In the numerical example shown in FIG. 6, the average power of the input signal is the following value. TIFF0007682403000021.tif10166

[0043] Digital amplitude adjustment is performed by adjusting the aforementioned default coefficient (k 0 ). Similar to the mathematical formula (8), the amplitude (A 4 ) after amplitude adjustment is given by the following mathematical formula. TIFF0007682403000022.tif14166A 4 The subscript "4" used for 4The digit "4" in the lower right subscript used in this example is a sequential number to distinguish it from other situations that will appear later, and has no other meaning. k 4 The method for determining will become clear from the explanation given below.

[0044] In the numerical example of FIG. 6, the average power after the amplitude adjustment is given by the following formula, similar to formula (10). TIFF0007682403000023.tif16166

[0045] In the numerical example shown in FIG. 6, digital amplitude adjustment is performed so that the average power after the amplitude adjustment is 0.25. In this case, k 4 The value can be determined as follows: TIFF0007682403000024.tif36166

[0046] In the numerical example in Fig. 6, the seventh column from the left shows the amplitude before quantization (B 4 ) represents the amplitude before quantization (B 4 ) is given by the following formula, similar to formula (12). TIFF0007682403000025.tif9166 When the symbol (X) is "-7", the amplitude before quantization (B 4 Specifically, the calculation is as follows: TIFF0007682403000026.tif23166

[0047] In the numerical example shown in FIG. 6, the 12th column from the left shows the amplitude after quantization (B 5 ) represents the amplitude after quantization (B 5 ) is given by the following formula: TIFF0007682403000027.tif10166 In the numerical example shown in FIG. 6, the eighth to eleventh columns from the left simply perform the transformation shown in equation (25) in stages.

[0048] In the numerical example shown in FIG. 6, the amplitude after saturation determination (B 6 ) is given by the following formula, similar to formula (15). TIFF0007682403000028.tif17166In the numerical example shown in Figure 6, the quantized amplitude (B 5 ) is not saturated.

[0049] In the numerical example shown in FIG. 4 and B 6 The correlation with was 1 (calculated using formula (16)).

[0050] (Numerical example, Figure 7) Fig. 7 is a second diagram showing a numerical example of digital amplitude adjustment in the optical transmitter 100 that performs compression shaping. The numerical example shown in Fig. 7 simulates a case where the occurrence probability of symbols far from the origin becomes low, resulting in an extreme drop in the average power of the input signal. Specifically, the occurrence probability of the symbol (X) shown in FIG. TIFF0007682403000029.tif15166As shown in equation (27), in the numerical example of FIG. 7, the parameter related to the variance of the Gaussian distribution is 0.2, which is 10 times larger than 0.02 in equation (17). In the case where the symbol (X) is "-7", the occurrence probability is specifically calculated as follows. 7, the probability that the symbol (X) is -7 is extremely low. The probability that the symbol (X) is 7 is also low.

[0051] In the numerical example shown in Fig. 7, the average power of the input signal is also given by Equation (6). In the numerical example shown in Fig. 7, the average power of the input signal is the following value. TIFF0007682403000031.tif10166

[0052] The digital amplitude adjustment is done by the default coefficient (k 0 ) is adjusted. As in equation (8), the amplitude after amplitude adjustment (A 7 ) is given by the following formula: TIFF0007682403000032.tif14166

[0053] Now, if we want to perform digital amplitude adjustment so that the average power after amplitude adjustment is 0.25, then k 7 is given as follows: TIFF0007682403000033.tif37166k given by equation (31) 7 is larger than the aforementioned k1 (=13.96594) and k4 (=16.50674).

[0054] In fact, in the case where the symbol (X) is -7, 7 Calculating the value gives the following: TIFF0007682403000034.tif29166

[0055] Furthermore, here, as shown in formula (12), the amplitude after amplitude adjustment (A 7 ) assuming that -1.0 to +1.0 is the full scale and assigned to the DAC121 conversion, naturally some undesirable phenomena will occur. As shown in Figure 7, the amplitude after amplitude adjustment (A 7 ) is greater than 1, specifically, in the four cases of X={-7,-5,5,7}, all of the values ​​(B9) after saturation determination are saturated. This undesirable saturation phenomenon is called "(digital signal) clipping" in the technical field of signal processing.

[0056] FIG. 8 is a diagram illustrating an example in which undesirable clipping occurs as a result of digital amplitude adjustment. FIG. 8A is a graph before digital amplitude adjustment, and FIG. 0 The probability distribution of and A related to Fig. 7 0 The probability distribution of and two distributions of are plotted. FIG. 8B is a graph after digital amplitude adjustment, and FIG. 4 The probability distribution of and A related to Fig. 7 7 As shown in FIG. 8B, in the numerical example shown in FIG. 7, the probability distribution of A 7 Four of these points are included in the clipping region.

[0057] Figure 8B suggests that there is no need to allocate the conversion of DAC121 for X={-7,-5,5,7}, which rarely occurs. In other words, if X={-7,-5,5,7} does not occur, 8PAM is excessive in the first place, and 4PAM of X={-3,-1,1,3} is sufficient. In this way, observing the probability distribution is important in determining the specifications of optical modulation. Whether or not optical communication equipment employs digital amplitude adjustment based on average power can be determined according to specifications such as the purpose of use of optical communication and the environment in which it is used.

[0058] (Definition of amplitude and DAC allocation considering occurrence probability) The optical transmitter 100 also operates by performing a linear transformation of a fixed coefficient on the symbol (X) to determine the amplitude (see Equations (2) and (3)). However, if we dispel the fixed idea that linear conversion is essential and define the amplitude and DAC allocation taking into account the occurrence probability of the symbol (X), it is possible to provide a larger margin for the symbol (X) that occurs frequently.

[0059] The DAC allocation method that takes into account the occurrence probability can be understood by first creating a histogram of the occurrence probability. FIG. 9 is a diagram for explaining the allocation of DAC and ADC to a symbol (X) in consideration of occurrence probability. FIG. 9A explains the allocation of DAC and ADC to a symbol (X) in a table format. FIG. 9B explains the allocation of DAC and ADC to a symbol (X) in a graph format. Both FIG. 9A and FIG. 9B use the numerical example shown in FIG. 7. Specifically, in the table shown in FIG. 9A, what is shown in the second column from the left is the probability (P) shown in the second column from the left in FIG. 7.

[0060] 9A, the third column from the left is a histogram of the occurrence probability of the symbol (X), subject to the following conditions: Condition 1: The class values ​​of the histogram are the values ​​of the symbols (X). Condition 2: The frequency is non-zero in every class. Condition 3: The total number of degrees is the quantized number of the DAC 121. In the table shown in FIG. 9A, the total number of values ​​in the third column from the left, that is, the total number of degrees, is the quantization number of the DAC 121, which is 256 in this numerical example.

[0061] 9A, the fourth column from the left indicates cumulative frequencies, i.e., the values ​​shown in the fourth column are obtained by accumulating the values ​​in the third column.

[0062] In the table shown in Fig. 9A, the fifth column from the left indicates the range assigned to each symbol (X) in the ADC of the receiver. For example, if the value sampled in the ADC of the receiver is "5 out of 256 levels", since 5 is included in "3 to 23" written in the third row, the corresponding symbol (X) is determined to be "-3". The width of the range assigned to each symbol (X) is the size of the margin. In this manner, the disclosed technique can provide a larger margin for symbols (X) that occur more frequently.

[0063] In the table shown in Fig. 9A, the sixth column from the left is a representative value of the "range" shown in the fifth column. More specifically, the sixth column from the left is a median value of the "range" shown in the fifth column. This representative value is an output value that the DAC 121 of the optical transmitter 100 outputs for a given symbol (X). To be precise, the DAC 121 outputs an analog electrical signal with an amplitude corresponding to the representative value among 256 steps.

[0064] In the graph shown in Fig. 9B, the horizontal axis represents the symbol (X) and the vertical axis represents the value of the DAC or ADC. In the graph shown in Fig. 9B, the dashed curve corresponds to the cumulative distribution function of the symbol (X). If a conventional linear transformation is performed, this dashed line becomes a straight line.

[0065] The technical feature of the transmission digital signal generation circuit 110 and the optical transmitter 100 according to the disclosed technology is that the amplitude and DAC allocation is performed taking into consideration the occurrence probability of the symbol (X). Information on how the DAC is allocated to this symbol (X) is shared with the receiver side of the optical communication.

[0066] The transmission digital signal generation circuit 110 and the optical transmitter 100 according to the disclosed technique have the above-mentioned technical features, and therefore have the effect of being able to provide a larger margin for the frequently occurring symbol (X).

[0067] Embodiment 2 The transmission digital signal generation circuit 110 and the optical transmitter 100 according to the second embodiment are modified examples of the transmission digital signal generation circuit 110 and the optical transmitter 100 according to the present disclosure. In the second embodiment, unless otherwise specified, the same reference symbols as those used in the first embodiment are used. Furthermore, in the second embodiment, descriptions that overlap with those in the first embodiment are omitted as appropriate.

[0068] In the first embodiment, the subject is a trunk-system large-capacity optical fiber communication in which compression shaping is performed, but the present disclosure is not limited thereto. The present disclosure may be applied to an optical transmitter 100 that performs extremely short burst generation for the purpose of improving reception sensitivity.

[0069] The extremely short burst generation of optical signals is a technique proposed by the applicant and described in, for example, the following reference documents: Reference: Shota Koshikawa, Keisuke Matsuda, Masashi Binami, Tsuyoshi Yoshida, and Kyosei Suzuki, "Improvement of receiver sensitivity by making 10Gb / s QPSK signals extremely short bursts for free-space optical communications," 2019 Institute of Electronics, Information and Communication Engineers General Conference, B-10-12, pp201.

[0070] The extremely short burst optical signal is a technology that assumes the application of free space optical communications to communications in the atmosphere, and addresses the issue of line disconnection due to increased transmission loss, particularly in bad weather. The extremely short burst optical signal concentrates power only in a short section of the optical signal, resulting in bursts that can withstand increased transmission loss at the expense of reduced communication capacity. With this technology for extremely short burst optical signals, when extremely short bursts are used in bad weather, the average output power is maintained at the same value as when bursts are not used in fine weather.

[0071] The optical transmitter 100, in which an optical signal is converted into extremely short bursts during bad weather, should also apply amplitude and DAC allocation that takes into account the occurrence probability of the symbol (X).

[0072] As described above, the disclosed technology can be applied to optical communications in which optical signals are converted into extremely short bursts, and has the effect of providing such optical communications systems with a larger margin for frequently occurring symbols (X). [Industrial Applicability]

[0073] The disclosed technology can be applied to, for example, Beyond 5G backbone high-capacity optical fiber communications, and has industrial applicability. [Explanation of symbols]

[0074] 100 optical transmitter, 110 transmission digital signal generation circuit, 111 encoding unit, 112 known symbol generation unit, 113 known symbol insertion unit, 114 amplitude adjustment unit, 115 gain control unit, 121 DAC, 122 light source, 123 optical modulation unit, 124 optical amplification unit.

Claims

1. A histogram is created of the occurrence probability of a symbol (X) that is converted into a transmission digital signal and input to a DAC, determining an allocation of the DAC for allocating a range of output values ​​of the DAC to the symbol (X) based on the cumulative frequency of the histogram; where the class values ​​of the histogram are the values ​​of the symbol (X), In every class, the frequency is non-zero, The total number of degrees is the quantization number of the DAC. Optical transmitter.

2. In trunk-system large-capacity optical fiber communications, compression shaping is performed on a transmitted digital signal.

2. The optical transmitter according to claim 1.

3. In optical space communications applied to communications in the atmosphere, power is concentrated only in a short section of the optical signal to be transmitted, thereby forming a burst that can withstand increased transmission loss, or forming an ultra-short burst.

2. The optical transmitter according to claim 1.

4. A transmission digital signal generation circuit constituting an optical transmitter, A histogram is created of the occurrence probability of a symbol (X) that is converted into a transmission digital signal and input to a DAC. determining an allocation of the DAC for allocating a range of output values ​​of the DAC to the symbol (X) based on the cumulative frequency of the histogram; where the class values ​​of the histogram are the values ​​of the symbol (X), In every class, the frequency is non-zero, The total number of degrees is the quantization number of the DAC. Transmit digital signal generation circuit.

5. In a trunk-system large-capacity optical fiber communication, compression shaping is performed on a transmission digital signal.

5. The transmission digital signal generating circuit according to claim 4.

6. In optical space communications applied to communications in the atmosphere, a burst is generated that can withstand increased transmission loss by concentrating power only in a short section of the optical signal to be transmitted, and an ultra-short burst is generated.

5. The transmission digital signal generating circuit according to claim 4.

Images