Communication method and communication system

Abstract

The present invention makes it possible to adaptively optimize the mapping shape of a non-uniform constellation (NUC) according to various transmission paths.　In an initialization sequence, a transmission-side device performs transmission processing while changing the constellation shape. In a reception-side device, a mutual information amount calculation unit 18 calculates a mutual information amount on the basis of an LLR obtained with respect to a reception signal from the transmission-side device, a mutual information amount analysis unit 19 analyzes the mutual information amount in a predetermined period, and an NUC optimization unit 20 optimizes an error rate and a constellation shape related to error correction encoding on the basis of an analysis result of the mutual information amount. Thereafter, in a normal sequence, the transmission-side device performs transmission processing by applying the error rate and the constellation shape optimized by the initialization sequence.

Description

Communication method and communication system

[0001] The present invention relates to a communication system in which a transmitting device encodes transmission information into an error correction code, maps the encoded information into a constellation in the complex domain, and transmits the resulting data to a receiving device.

[0002] Conventionally, information communication systems have been used to transmit information to remote locations via wireless or wired transmission paths. One information transmission method involves mapping the information to be transmitted on the transmitting side into a complex number domain called In-Phase (I) or Quadrature-Phase (Q) and transmitting it, and restoring the transmitted information from the received mapped signal on the receiving side. In this mapping method, when transmitting n-bit information, 2 n A QAM (Quadrature Amplitude Modulation) scheme using mapping points of the above is often used. As a mapping arrangement for QAM, a UC (Uniform Constellation) scheme is often used, in which the mapping points are arranged at equal intervals for both I and Q, as shown in FIG.

[0003] Furthermore, information transmission using mapping is generally used in conjunction with error correction, where information that has been error-corrected and coded on the transmitting side is transmitted using mapping, and the receiving side performs the reverse process to achieve information communication. In error correction, redundant bits for error correction are added to the information to be transmitted, and the more redundant bits there are, the greater the error correction capability. The ratio of these redundant bits can be expressed as the coding rate R. The coding rate R is defined as the result of dividing the number of information bits by the sum of the number of information bits and the number of redundant bits.

[0004] Regarding the above mapping, when the number of bits n to be transmitted is large, it is known that using UC mapping causes degradation in transmission performance, called shaping loss, in which the error rate after error correction coding increases. This degradation is particularly pronounced when n≧8 (256QAM). A known method for reducing shaping loss is to use NUC (Non Uniform Constellation), which uses mapping points at irregular intervals. Patent Document 1 discloses a measurement device that can suppress degradation in the measurement accuracy of the modulation error ratio even when NUC is used.

[0005] Japanese Patent Application Laid-Open No. 2017-183921

[0006] When the above-mentioned NUC is used, the receiving process becomes more complicated compared to UC, but there are effects such as improving the required SNR (Signal-to-Noise Ratio). However, when NUC is used, in addition to the increased complexity of the receiving process, there are also issues such as (1) low flexibility of the communication system, (2) when waveform distortion occurs due to nonlinear characteristics, transmission performance deteriorates, and (3) it may not be optimal when the mixed noise is not Gaussian distributed.

[0007] The problem (1), low flexibility of the communication system, will now be described. In a typical NUC, the mapping shape is optimized according to the SNR to be applied. That is, in current communication systems, the SNR to be operated (required SNR) is set in advance, and an NUC mapping shape corresponding to that SNR is used. Since the modulation multi-level number and coding rate R of the mapping are determined from the required SNR, the NUC is optimized for the required SNR. In this way, since it is necessary to set the required SNR in advance and then optimize the NUC shape, if the SNR of the assumed transmission path differs from the actual SNR, the NUC shape will not be optimal, resulting in low flexibility as a communication system.

[0008] This section explains issue (2), which concerns degradation of characteristics due to nonlinear transmission paths. While communication systems are based on the premise that they are linear, nonlinearity exists in wired systems such as power amplifiers in the transmitting section and optical fibers, and this nonlinearity degrades transmission performance. Generally, the degradation of transmission performance due to nonlinear characteristics is mitigated by compensating for nonlinear characteristics to make them linear. However, degradation of transmission performance can occur if such compensation is not performed, if nonlinear characteristics remain due to insufficient compensation even when a compensation function is installed, or if nonlinear characteristics occur in the receiving section, which are difficult to compensate for.

[0009] Finally, we will explain problem (3), which is the case where the mixed noise is not Gaussian distributed. The optimization is performed on the assumption that the noise considered when optimizing the NUC is noise exhibiting a normal distribution, so-called Gaussian noise. Therefore, if the mixed noise is not Gaussian noise, the NUC will not be optimal. Specific examples of noise other than Gaussian noise include multiplexing methods called LDM (Layered Division Multiplexing) or NOMA (Non Orthogonal Multiple Access).

[0010] In LDM, 2 n -QAM and 2 m -QAM is multiplexed by providing a power difference in the mapping area, n -QAM is used as a high-power UL (Upper Layer), 2 m 1-QAM is called the low-power LL (Lower Layer). The LDM operation method takes advantage of the difference in required SNR between short-distance users and long-distance users, transmitting m bits to short-distance users via the LL and n bits to long-distance users via the UL. Long-distance users are targeted for demodulation via the UL, but the LL mapping signal behaves as noise in the UL. Therefore, the noise contained in the UL signal is a mixture of the LL signal and Gaussian noise, and the amplitude distribution of this noise differs from a normal distribution.

[0011] The present invention has been made in view of the above-described conventional circumstances, and has an object to make it possible to adaptively optimize the mapping shape of an Unequally Spaced Constellation (NUC) in accordance with various transmission paths.

[0012] In order to achieve the above object, a communication method according to one aspect of the present invention is configured as follows: That is, the communication method performs a transmission process in which a transmitting device performs error correction coding on transmission information, maps the encoded information onto a complex domain constellation, and transmits the resulting information to a receiving device, the method comprising an initialization sequence and a normal sequence, in which, in the initialization sequence, the transmitting device performs the transmission process while changing a constellation shape, the receiving device optimizes the constellation shape by repeating a process of calculating and analyzing mutual information based on LLRs obtained for received signals from the transmitting device, and in the normal sequence, the transmitting device performs the transmission process using the constellation shape optimized by the initialization sequence.

[0013] In the above communication method, an allowable error rate for error correction coding may be set in advance, and in the initialization sequence, the receiving device may estimate the error rate for error correction coding based on a histogram of mutual information over a predetermined period, calculate a coding rate at which the error rate becomes equal to the allowable value, transmit the calculated coding rate to the transmitting device for application to subsequent error correction coding processing, and then proceed to an operation of optimizing the constellation shape.

[0014] Alternatively, in the above communication method, a required SNR may be set in advance, and in the initialization sequence, the receiving side device may estimate the SNR of the received signal, add noise equivalent to a margin of the estimated SNR with respect to the required SNR to the equalization result of the received signal, calculate LLRs for the signal after the addition of noise, and analyze the mutual information based on the calculated LLRs, thereby optimizing the constellation shape.

[0015] A communication system according to another aspect of the present invention is configured as follows: That is, a communication system in which a transmitting-side device performs a transmission process in which transmission information is error-correction-encoded and mapped to a complex-domain constellation, and then transmits the resulting constellation to a receiving-side device, includes an initialization sequence and a normal sequence, in which, in the initialization sequence, the transmitting-side device performs the transmission process while changing a constellation shape, the receiving-side device optimizes a coding rate for the error-correction coding and a constellation shape while repeating a process of calculating and analyzing mutual information based on LLRs obtained for received signals from the transmitting-side device, and in the normal sequence, the transmitting-side device performs the transmission process by applying the coding rate and constellation shape optimized by the initialization sequence.

[0016] According to yet another aspect of the present invention, there is provided a communication system configured as follows: A communication system in which a transmitting device performs a transmission process in which transmission information is error-correction-encoded and mapped to a complex-domain constellation, and then transmits the resulting constellation to a receiving device, the communication system having an initialization sequence and a normal sequence, and a required SNR is set in advance, in which in the initialization sequence the transmitting device performs the transmission process while changing a constellation shape, in which the receiving device estimates the SNR of the received signal, adds noise corresponding to a margin of the estimated SNR with respect to the required SNR, to an equalization result of the received signal, and calculates and analyzes mutual information based on an LLR obtained for the signal after the noise has been added, thereby optimizing the constellation shape, and in which in the normal sequence the transmitting device performs the transmission process using the constellation shape optimized by the initialization sequence.

[0017] According to the present invention, it is possible to adaptively optimize the mapping shape of a non-uniformly spaced constellation (NUC) in accordance with various transmission paths, thereby providing a communication system with improved transmission rates and received power margins.

[0018] 1 is a diagram illustrating a configuration example of a communication system according to a first embodiment of the present invention; FIG. 2 is a diagram illustrating a configuration example of a communication system according to a second embodiment of the present invention; FIG. 3 is a diagram illustrating an example of UC mapping; 8 1 is a diagram showing an example of I-axis / Q-axis mapping points for QAM; n 1 is a diagram showing an example of NUC mapping after convergence when SNR=10 dB in -QAM; FIG. 2 is a diagram showing an example of the convergence process of mutual information using the Nelder-Mead method; FIG. 3 is a diagram showing an example of an improvement in a histogram of mutual information; FIG. 4 is a diagram showing an example of UC mapping when passing through a nonlinear element; FIG. 5 is a diagram showing an example of NUC mapping suitable for a nonlinear element;

[0019] Several embodiments of the present invention will be described with reference to the drawings. Fig. 1 shows an example of the configuration of a communication system according to a first embodiment of the present invention. The communication system of Fig. 1 includes an error correction coding unit 11, an NUC mapping unit 12, a modulation unit 13, a transmission path 14, an equalization unit 15, an LLR calculation unit 16, an error correction decoding unit 17, a mutual information calculation unit 18, a mutual information analysis unit 19, an NUC optimization unit 20, and a feedback transmission unit 21. For example, a transmitting device includes the error correction coding unit 11, the NUC mapping unit 12, and the modulation unit 13, and a receiving device includes the equalization unit 15, the LLR calculation unit 16, the error correction decoding unit 17, the mutual information calculation unit 18, the mutual information analysis unit 19, and the NUC optimization unit 20.

[0020] Information to be transmitted is input to the error correction coding unit 11. The error correction coding unit 11 performs coding processing for error correction on the input information. Commonly used error correction methods include LDPC (Low Density Parity Check) and turbo codes because of their high error correction capabilities. As mentioned above, the coding rate R is an index that determines the error correction capability. The smaller the coding rate R, the higher the correction capability; however, the greater the number of redundant bits for error correction coding, which reduces the rate at which the number of information bits can be transmitted. Conversely, as the coding rate R increases, the correction capability decreases but the transmission rate increases, creating a trade-off.

[0021] The NUC mapping unit 12 receives the result of the error correction coding process by the error correction coding unit 11 and converts it into n Here, n is the number of bits that can be transmitted in one mapping, and the modulation multi-level number is 2 n The NUC mapping unit 12 also receives information from the feedback transmission unit 19, and determines the mapping shape based on the information from the feedback transmission unit 19. Details of the mapping shape will be described later.

[0022] The modulation unit 13 converts the 2 n A 1-QAM mapped signal is input, and a modulation process such as OFDM (Orthogonal Frequency Division Multiplexing) or SC (Single Carrier) is performed on the mapped signal to generate a modulated signal. The appropriate modulation method varies depending on the transmission method, such as wireless transmission or wired transmission. In the case of wireless transmission, the bandwidth is often limited, so a modulation method with high frequency utilization efficiency, such as OFDM, is often adopted. On the other hand, in the case of wired transmission, there are no such constraints, so an SC method that does not impose band limitations is often used. Furthermore, in the case of wireless transmission, a process of frequency conversion is also performed to convert the center frequency to a wireless carrier frequency, and in the case of wired transmission, optical modulation is also performed when, for example, optical fiber is used as the transmission path.

[0023] The modulated signal obtained by the modulator 13 is transmitted from the transmitting side to the receiving side via the transmission path 14 and input to the equalizer 15. The transmission path 14 is air in the case of wireless transmission, and a metal wire or optical fiber in the case of wired transmission. Furthermore, in both wireless and wired transmission, as the transmission distance increases, the signal power also attenuates accordingly. Therefore, the transmission device has a low-noise amplifier at the receiving end to amplify the attenuated signal power. However, the low-noise amplifier generates thermal noise of power depending on the temperature and amplifier characteristics, and this thermal noise becomes the dominant noise that determines the reception SNR.

[0024] The equalization unit 15 performs equalization processing to restore the received mapped signal whose amplitude and phase have fluctuated, and outputs the resulting equalized signal. The equalization method for demodulation processing varies depending on the modulation method, such as OFDM or SC, but since the demodulation processing itself is not the essence of the present invention, a detailed description thereof will be omitted here.

[0025] The equalized signal obtained by the equalization unit 15 is input to the LLR calculation unit 16. The LLR calculation unit 16 has a function of calculating an LLR (Log Likelihood Ratio) based on the equalized signal. The LLR represents the reliability of each of the n bits calculated from the received mapping point. For example, if a certain bit receives a '0' with high reliability, the LLR will be a large negative value, if a '1' is received with high reliability, the LLR will be a large positive value, and if it is completely impossible to distinguish between a '0' and a '1', the LLR will be 0.

[0026] The error correction decoding unit 17 receives the LLRs calculated by the LLR calculation unit 16 and performs error correction based on the rules used when encoding was performed by the error correction encoding unit 11. Performing error correction processing from the input LLRs is not the essence of the present invention, and a detailed explanation will be left to specialized books. Note that as the SNR decreases, the absolute value of the LLRs decreases, and the reliability indicated by the LLRs is low. However, even in such a situation, correct decoding is possible by setting the encoding rate R to a small value.

[0027] The LLRs calculated by the LLR calculation unit 16 are also input to a mutual information calculation unit 18. The mutual information calculation unit 18 converts the LLRs into mutual information (MI). Since mutual information is an important concept in the present invention, it will be described in detail below.

[0028] Mutual information is expressed as an index ranging from 0 to 1, with the closer the value is to 1, the higher the reliability of the demodulated data; 1 indicates no transmission errors, and 0 indicates a bit error rate of 0.5. Mutual information is also closely related to the error correction limit; for example, if the LDPC coding rate is R, this means that LDPC can be decoded without error when the following equation (1) is satisfied: where f(·) is a monotonically increasing function and is determined based on the block length and configuration of the LDPC code.

[0029]

[0030] For example, if the required mutual information f(R) is 0.3 and MI is 0.8, the difference is 0.5, which is defined as the mutual information margin. A large mutual information margin indicates (1) that there is a sufficient SNR margin, and (2) that there is a possibility that the coding rate R can be increased to increase the transmission rate.

[0031] The process of calculating the mutual information from the LLR begins by first calculating the bit probability p for bit '0' from the LLR as shown in the following equation (2). (0) m,i , bit probability p for bit '1' (1) m,i Calculate.

[0032]

[0033] Here, i indicates the number of the received mapping signal, and m indicates the m-th bit. By substituting the bit probability calculated by equation (2) into the following equation (3), the instantaneous mutual information MI is calculated. m,i can be calculated.

[0034]

[0035] Next, as shown in the following equation (4), the instantaneous mutual information MI m,i The mutual information MI is calculated by averaging m and i. Here, the averaging parameter In is set to a value that approximates the error correction capability of LDPC or the like, so it is desirable that it be the average value within the error correction block of LDPC or the like, or approximately the same.

[0036] The mutual information MI calculated by the mutual information calculation unit 18 is input to the mutual information analysis unit 19. The mutual information analysis unit 19 assigns an index k to the order of the input mutual information MI and redefines it as MI(k). For example, in the case of mobile wireless transmission, received power fluctuations called fading occur, in which the characteristics of the transmission path 14 fluctuate from moment to moment, and MI(k) also fluctuates in accordance with the received power fluctuations. The mutual information analysis unit 19 analyzes the average value, maximum value, minimum value, histogram, etc. of MI(k) within a predetermined period. For example, by calculating a histogram and calculating the probability that MI(k)<f(R), the block error rate of the LDPC block can be estimated. Furthermore, in wireless fixed transmission and wired transmission, the propagation path conditions are often constant, so the average value, minimum value, and maximum value of MI(k) are approximately the same.

[0037] The NUC optimization unit 20 optimizes the mapping shape of the NUC (unequally spaced constellation) and the error correction coding rate R based on the analysis result of MI(k) by the mutual information analysis unit 19. This optimization is the main focus of the present invention, and will be described in detail later.

[0038] The optimization result by the NUC optimization unit 20 is transmitted to the NUC mapping unit 12 and the error correction coding unit 11 via the feedback transmission unit 21. The optimized coding rate R is also transmitted to the error correction decoding unit 17. As a specific implementation method of the feedback transmission unit 21, in the case of wireless transmission, techniques such as time division multiplexing communication and frequency division multiplexing communication are used, but because the amount of information to be fed back is small, different propagation paths such as a narrowband control channel or wired communication may also be used. In addition, in the case of wired transmission, the same applies as in wireless, but full duplex communication is often used in the case of wired transmission. In any case, the feedback transmission method is not essential to the present invention, so details will be omitted.

[0039] The above explanation has clarified the outline of the series of processes in the communication system according to the first embodiment of the present invention shown in Fig. 1. Next, a specific sequence for optimizing the NUC and coding rate R of the communication system will be explained in detail.

[0040] When the communication system is started, it starts with an initialization sequence and then transitions to a normal sequence. If the communication system experiences a major failure or a continuous outage, it will return from the normal sequence to the initialization sequence and then transition back to the normal sequence.

[0041] In the initialization sequence, first n The transmitting device transmits a UC (uniform constellation) of QAM to the receiving device. The receiving device receives the UC mapping and performs the process described above to analyze MI(k) from the received UC mapping.

[0042] The NUC optimization unit 20 optimizes the NUC mapping shape and also optimizes the coding rate R based on the strategy shown below. There are roughly two approaches to this optimization strategy.

[0043] The first strategy tolerates a certain degree of LDPC block error rate β and optimizes the transmission rate to maximize the rate. Transmission errors occur with a probability equal to the block error rate β, but if an error occurs, retransmission can be performed to ensure smooth communication. While retransmission processing has the disadvantage of complicating the communication system and increasing transmission delays, no significant transmission delays occur as long as the block error rate β is below a predetermined probability.

[0044] The second strategy is applied to communication systems that cannot tolerate the increased complexity and transmission delay of the communication system that are the retransmission problems of the first strategy. For example, the second strategy includes wireless transmission devices such as FPUs (Field Pickup Units) used for live broadcasting that cannot tolerate retransmission, and 5G wireless fronthauls that require extremely low latency.

[0045] First, the first strategy will be described. From the histogram of MI(k) analyzed by the mutual information analysis unit 19, the coding rate R=R is determined so that the probability that MI(k)<f(R) is β. 0 The index 0 indicates the initial value. 0 is transmitted to the error correction coding unit 11 and the error correction decoding unit 17, where processing according to the coding rate R0 is performed.

[0046] Next, the optimized coding rate R 0 After confirming that the block error rate is β, the process moves to optimizing the NUC. 0 It is not required that the block error rate at β exactly matches β, but rather that they substantially match (that is, they should match within a practically effective range). The basic principle of NUC optimization is to optimize the NUC mapping shape so as to maximize the mutual information MI.

[0047] The algorithm used to optimize the NUC mapping shape is not particularly limited, but one example is the Nelder-Mead method. The Nelder-Mead method has the advantage of not requiring the derivative of the optimization function. Mutual information depends on the mapping shape, noise mixed in at the receiving unit, characteristics of the transmission path, etc., and is a complex function of multiple variables, so it is generally not possible to calculate the derivative. For this reason, the Nelder-Mead method is suitable for optimizing the NUC mapping shape.

[0048] There are two types of optimization for the NUC mapping shape: optimization in a one-dimensional domain and optimization in a two-dimensional domain. In the one-dimensional domain, the mapping points are separated into I / Q, and the mapping positions on the I axis and Q axis are equal. In the two-dimensional domain, optimization is performed on the IQ plane, so although the degree of MI improvement by optimization is large, it has the disadvantages of taking a long time to perform optimization and of making LLR calculation on the receiving side complicated.

[0049] Here, we will explain the NUC optimization method in a one-dimensional domain using the Nelder-Mead method. 8 In -QAM, as shown in Figure 4, there are 16 mapping points on both the I axis and the Q axis. The mapping points are symmetrical around the origin. In the Nelder-Mead method, as shown in Figure 4, the mapping positions are arranged in ascending order from the closest to the origin. (j) = {b (j) 1 , b (j) 2 , ..., b (j) 8 } is defined as

[0050] Also, as shown in FIG. 5, a vector set called a simplex {a (1) , a (2) , ..., a (N) } (where N=√2 n / 2+1) is used to iteratively optimize the worst value of the objective function g(·) in the simplex. For example, when n=8, {a (1) , a (2) , ..., a (9) } is used as a simplex. In the present invention, the objective function g(a) is the mutual information MI when the actual propagation path is passed through the mapping point a.

[0051] In the present invention, this simplex is transmitted to the NUC mapping unit 12 via the feedback transmission unit 20. The NUC mapping unit 12 transmits a (1) Mapping is generated by (1) The mapping signal is transmitted to the receiving device via the transmission path 14, and the amount of mutual information is measured at the receiving device. (1) When the mapping is made, the mutual information MI is g(a (1) ) is measured. (2) From a (9) are transmitted for a predetermined period, and their mutual information g(a (2) ) to g(a (9) ) is measured.

[0052] As mentioned above, in the Nelder-Mead method, the worst mapping position a is selected to improve the worst value of the objective function (MI) in the simplex. W The mapping position is corrected by moving the mapping points little by little by trial and error using multiple methods such as reflection, expansion, contraction, and reduction, measuring the mutual information for each method, and adopting the correction method with the largest mutual information. Specifically, the corrected mapping positions by reflection, expansion, contraction, and reduction are respectively W ref , a W exp , a W cont , a W shrk For example, if g(a W cont ) was the best, W ←a W cont and update the worst mapping position. This update updates the worst mapping position in the simplex and also increases the corresponding mutual information.

[0053] In the second iteration, the updated mapping position a W Among the simplices including , the mapping position with the smallest mutual information is selected as the target for correction. As with the first time, the mutual information is measured using each correction method: reflection, expansion, contraction, and reduction, and the best correction result is adopted to correct the mapping position.

[0054] By repeating this iterative process multiple times, all the simplices will have almost the same value, which means that the NUC mapping position will converge. 8 The figure shows the converged NUC mapping for -QAM with SNR = 15 dB. Compared to the UC mapping shown in Figure 3, it can be seen that the mapping points are unevenly spaced. Similarly, as shown in Figure 7, the mutual information also converges to a constant value, improving from 0.38 for UC to 0.41 for NUC.

[0055] Optimizing the NUC improves the average value of the mutual information in the mutual information analysis unit 19, and as shown in FIG. 9, the distribution of the histogram also shifts in the direction of increasing the mutual information. 0 The new coding rate R 1 (R 1 >R0 ) can be updated to the new coding rate R 1 The criteria for updating to MI(k)<f(R 1 ) is β 1 To reiterate, by increasing the error correction coding rate, it is possible to improve the transmission rate.

[0056] The series of optimization processes described above makes it possible to allow a block error rate up to β, optimize the NUC mapping shape and coding rate under that constraint, and maximize the transmission rate. In conventional NUC, a required SNR is set in advance and a fixed NUC mapping based on the required SNR is used, which reduces the degree of freedom of the communication system. However, according to the present invention, the NUC mapping shape and coding rate are adaptively optimized in accordance with the propagation path, thereby significantly improving the degree of freedom of the communication system.

[0057] When the transmission path does not fluctuate, operation can be performed using the NUC obtained by the above optimization sequence, but when the transmission path characteristics fluctuate, it is possible to follow the transmission path fluctuations by periodically executing the above optimization sequence. When performing periodic optimization, the optimization communication can be allocated by time or by frequency.

[0058] Next, we will refer to an example of a nonlinear transmission path mentioned in the above-mentioned problem (2). FIG. 9 shows the mapping shape when UC mapping passes through a nonlinear element. As can be seen from the figure, the amplitude decreases as one moves toward the periphery of the mapping region. Such transmission causes a problem in that the error rate increases at mapping points located on the outer periphery. In response to this problem, by applying the first embodiment described above, it is possible to optimize the NUC mapping shape to adapt to nonlinear characteristics, as shown in FIG. 10. Examining the shape in the figure reveals that the spacing on the outer periphery side is wider, while the spacing on the inner periphery side is narrower.

[0059] Similarly, a transmission path with a lot of phase noise will be described. When a wireless transmission path is used, the baseband signal needs to be frequency-shifted to the carrier frequency band. This process involves multiplying the baseband signal by a sine wave called a local signal. However, when the sine wave contains a lot of phase noise, a rotation component occurs in the mapping, and fluctuations due to rotation increase, especially on the outer periphery. As with a nonlinear transmission path, this results in a higher error rate on the outer periphery. Even in such cases, optimizing the NUC mapping shape makes it possible to provide an optimal mapping shape in an environment where phase noise is present.

[0060] As described above, in the communication system according to the first embodiment, in the initialization sequence, the transmitting device performs transmission processing while changing the constellation shape. In the receiving device, the mutual information calculation unit 18 calculates mutual information based on the LLRs obtained for the received signal from the transmitting device, the mutual information analysis unit 19 analyzes the mutual information over a predetermined period, and the NUC optimization unit 20 optimizes the error rate and constellation shape related to error correction coding based on the results of the mutual information analysis. Thereafter, in the normal sequence, the transmitting device performs transmission processing by applying the error rate and constellation shape optimized in the initialization sequence. This allows the NUC mapping shape to be adaptively optimized according to various transmission paths.

[0061] Next, a communication system according to a second embodiment of the present invention will be described in detail with reference to FIG. 2 . The second embodiment is configured to implement the second strategy described above. The second strategy is applied when low-latency transmission is required, retransmission is not permitted (or a small number of retransmissions is permitted), and highly reliable information communication is required. Therefore, while the first strategy allows β as a block error rate, the second strategy assumes operation with a predetermined transmission margin so that β approaches 0. Therefore, while the first embodiment does not require a required SNR for operation, the second embodiment sets a required SNR and then optimizes the NUC mapping shape to maximize the transmission margin.

[0062] The communication system in Fig. 2 includes an error correction coding unit 11, an NUC mapping unit 12, a modulation unit 13, a transmission path 14, an equalization unit 15, an LLR calculation unit 16, an error correction decoding unit 17, a mutual information calculation unit 18, a mutual information analysis unit 19, an NUC optimization unit 20, a feedback transmission unit 21, an SNR estimator 22, a noise addition unit 23, and an LLR calculation unit 24. That is, the communication system in Fig. 2 has a configuration in which the SNR estimator 22, the noise addition unit 23, and the LLR calculation unit 24 are added to the configuration shown in Fig. 1.

[0063] The error correction coding unit 11, NUC mapping unit 12, modulation unit 13, transmission path 14, equalization unit 15, LLR calculation unit 16, error correction decoding unit 17, mutual information calculation unit 18, mutual information analysis unit 19, NUC optimization unit 20, and feedback transmission unit 21 in the communication system of Fig. 2 are the same as those shown in Fig. 1, and detailed description thereof will be omitted. However, the coding rate R is a value set according to the desired transmission rate, and the coding rate R is not adaptively controlled.

[0064] The SNR estimation unit 22 receives a signal received via the transmission path 14 and estimates the SNR of the received signal. Examples of SNR estimation methods include a method in which a transmitting device transmits a known signal and a receiving device receives the signal and estimates the noise power contained in the known signal, and a method in which the transmitting device provides an empty space in the time or frequency domain and the receiving device estimates the noise power contained in the empty space.

[0065] The SNR estimation result by the SNR estimation unit 22 is input to the noise addition unit 23. The noise addition unit 23 also receives the output signal from the equalization unit 15. In the second embodiment, a required SNR is set, and the noise addition unit 23 determines the difference between the required SNR and the estimated SNR; for example, if the required SNR is 25 dB and the estimated SNR is 28 dB, there is a transmission margin of 3 dB. The second embodiment aims to improve this transmission margin to 3 dB or more by optimizing the NUC mapping shape.

[0066] The noise adding unit 23 adds noise of the transmission margin to the signal resulting from equalization by the equalization unit 15. As a result, the SNR after adding noise becomes the required SNR. Specifically, the required SNR is expressed as Γ req [dB], and the estimated SNR is Γ est [dB], the added noise power P N is expressed by the following equation (5): s denotes the signal power after equalization.

[0067]

[0068] The noise added by the noise addition unit 23 is noise of statistically the same nature as the noise generated in the low-noise amplifier. This noise is generally Gaussian noise. The output of the noise addition unit 23 is input to the LLR calculation unit 24. The processing by this LLR calculation unit 24 is equivalent to the processing by the LLR calculation unit 16. As in the first embodiment, the NUC mapping shape is optimized by the mutual information calculation unit 18, mutual information analysis unit 19, NUC optimization unit 20, and feedback transmission unit 21 using the LLRs calculated by the LLR calculation unit 24.

[0069] As described above, in the second embodiment, new noise is added to the equalization result. The signal after the noise addition simulates noise that passes through an actual transmission path and whose SNR has decreased to the required SNR. By optimizing the NUC mapping shape for the signal that simulates the required SNR, it becomes possible to maximize the amount of mutual information in the required SNR environment, thereby maximizing the transmission margin. Therefore, as described above, by setting the required SNR and using the NUC mapping shape that is optimal for the required SNR, it becomes possible to achieve stable information communication.

[0070] As described above, in the communication system according to the second embodiment, in the initialization sequence, the transmitting device performs transmission processing while changing the constellation shape. In the receiving device, the SNR estimation unit 22 estimates the SNR of the received signal, the noise addition unit 23 adds noise corresponding to a margin of the estimated SNR relative to the required SNR to the equalization result of the received signal, the LLR calculation unit 24 calculates the LLR for the signal after the noise addition, the mutual information calculation unit 18 calculates the mutual information based on the LLR, the mutual information analysis unit 19 analyzes the mutual information over a predetermined period, and the NUC optimization unit 20 optimizes the constellation shape based on the analysis result of the mutual information. Thereafter, in the normal sequence, the transmitting device performs transmission processing using the constellation shape optimized by the initialization sequence. This allows the NUC mapping shape to be adaptively optimized according to various transmission paths.

[0071] Although the embodiments of the present invention have been described above, these embodiments are merely illustrative and do not limit the technical scope of the present invention. The present invention can take on various other embodiments, and various modifications such as omissions and substitutions can be made without departing from the spirit of the present invention. These embodiments and modifications thereof are included in the scope and spirit of the invention described in this specification, etc., and are included in the invention described in the claims and their equivalents.

[0072] Furthermore, the present invention can be provided not only as devices such as those described above or as systems composed of these devices, but also as methods executed by these devices, programs for realizing the functions of these devices using a processor, and storage media for storing such programs in a computer-readable manner.

[0073] The present invention can be used in a communication system in which a transmitting device encodes transmission information into an error correction code, maps the encoded information into a constellation in the complex domain, and transmits the resulting data to a receiving device.

[0074] 11: Error correction coding unit, 12: NUC mapping unit, 13: Modulation unit, 14: Transmission path, 15: Equalization unit, 16: LLR calculation unit, 17: Error correction decoding unit, 18: Mutual information calculation unit, 19: Mutual information analysis unit, 20: NUC optimization unit, 21: Feedback transmission unit, 22: SNR estimator, 23: Noise addition unit, 24: LLR calculation unit

Claims

1. A communication method in which a transmitting device performs a transmission process in which transmission information is error-correction coded and mapped to a complex domain constellation, and then transmitted to a receiving device, the method comprising an initialization sequence and a normal sequence, wherein in the initialization sequence, the transmitting device performs the transmission process while changing the constellation shape, and the receiving device optimizes the constellation shape by repeating a process of calculating and analyzing mutual information based on LLRs obtained for the received signal from the transmitting device, and in the normal sequence, the transmitting device performs the transmission process by applying the constellation shape optimized by the initialization sequence.

2. A communication method according to claim 1, wherein an allowable error rate for error correction coding is set in advance, and in the initialization sequence, the receiving device estimates the error rate for error correction coding based on a histogram of mutual information over a predetermined period, calculates a coding rate at which the error rate becomes equal to the allowable value, transmits the calculated coding rate to the transmitting device for application to subsequent error correction coding processing, and then proceeds to an operation of optimizing the constellation shape.

3. A communication method according to claim 1, wherein a required SNR is set in advance, and in the initialization sequence, the receiving device estimates the SNR of the received signal, adds noise equivalent to a margin of the estimated SNR relative to the required SNR to the equalization result of the received signal, calculates LLRs related to the signal after the addition of noise, and analyzes the amount of mutual information based on the calculated LLRs, thereby optimizing the constellation shape.

4. A communication system in which a transmitting device performs a transmission process in which transmission information is error-correction coded and mapped to a complex domain constellation, and then transmitted to a receiving device, the communication system having an initialization sequence and a normal sequence, wherein in the initialization sequence, the transmitting device performs the transmission process while changing the constellation shape, and the receiving device optimizes the coding rate for the error-correction coding and the constellation shape while repeating a process of calculating and analyzing mutual information based on LLRs obtained for the received signal from the transmitting device, and in the normal sequence, the transmitting device performs the transmission process by applying the coding rate and constellation shape optimized by the initialization sequence.

5. A communication system in which a transmitting device performs a transmission process in which transmission information is error-correction coded, mapped to a complex domain constellation, and transmitted to a receiving device, the system having an initialization sequence and a normal sequence, and a required SNR is set in advance, wherein in the initialization sequence, the transmitting device performs the transmission process while changing the constellation shape, and the receiving device estimates the SNR of the received signal, adds noise equivalent to a margin of the estimated SNR with respect to the required SNR, to the equalization result of the received signal, calculates mutual information based on the LLR obtained for the signal after the noise has been added, and optimizes the constellation shape while repeating this analysis process, and in the normal sequence, the transmitting device performs the transmission process by applying the constellation shape optimized by the initialization sequence.

Images