HARQ apparatus, communication device, control program, and conversion method in HARQ apparatus

Abstract

To propose a new method for compressing LLR data to be temporarily stored in a memory in HARQ.SOLUTION: A conversion unit (32) of a HARQ device (22) obtains a positive representative value that is a value that represents a positive LLR in LLR data including a plurality of LLRs for bits, and a negative representative value that is a value that represents a negative LLR in the LLR data, and for each LLR included in the LLR data, if the LLR is positive, calculates an LLR difference from the positive representative value, while if the LLR is negative, calculates an LLR difference from the negative representative value, and stores data including a sign indicating whether the LLR is positive or negative, the LLR difference, the positive representative value, and the negative representative value for each LLR in a memory (35) as conversion data.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The present invention relates to an HARQ device for performing HARQ in a communication device, a communication device, a control program, and a conversion method in the HARQ device. [Background technology]

[0002] HARQ (Hybrid Automatic Repeat reQuest) is known as a technique for achieving highly reliable wireless communication. In HARQ, a communication device (receiving device) performs error correction on LLR (Log Likelihood Ratio) data received and demodulated from another communication device (transmitting device). If the error correction fails, the receiving device requests retransmission from the transmitting device, combines the LLR data received and demodulated from the transmitting device with the LLR data for which the error correction failed, and performs error correction again. For this reason, the receiving device is provided with a memory for temporarily storing the LLR data for which the error correction failed. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Japanese Patent Application Laid-Open No. 2013-535912 Summary of the Invention [Problem to be solved by the invention]

[0004] With the advancement of wireless communication technology, the amount of data communication per unit time is increasing. Accordingly, it is necessary to increase the capacity of memory for temporarily storing LLR data. To address this problem, Patent Document 1 discloses a method for compressing LLR data stored in memory using a lookup table, μ-Law compression, etc.

[0005] An object of one aspect of the present disclosure is to propose a new method for compressing LLR data temporarily stored in memory in HARQ. [Means for solving the problem]

[0006] In order to solve the above problem, an HARQ device according to one embodiment of the present invention includes an acquisition unit that acquires LLR data including a plurality of LLRs of 1 bit, and a conversion unit that converts the LLR data to create converted data and stores the converted data in a memory. The conversion unit obtains a positive representative value that is a value that represents a positive LLR in the LLR data and a negative representative value that is a value that represents a negative LLR in the LLR data, and, for each LLR included in the LLR data, if the LLR is positive, calculates a difference of the LLR from the positive representative value, while, if the LLR is negative, calculates a difference of the LLR from the negative representative value, and stores data in the memory as the converted data including a sign indicating whether the LLR is positive or negative, the LLR difference, the positive representative value, and the negative representative value.

[0007] A conversion method in an HARQ device according to another aspect of the present invention is a conversion method for creating conversion data by converting LLR data including a plurality of LLRs of 1 bit, and includes the steps of: obtaining a positive representative value that is a value that represents positive LLRs in the LLR data; and a negative representative value that is a value that represents negative LLRs in the LLR data; calculating, for each LLR included in the LLR data, if the LLR is positive, a difference of the LLR from the positive representative value; and, for each LLR included in the LLR data, calculating a difference of the LLR from the negative representative value if the LLR is negative; and storing, in a memory as the conversion data, data including a sign indicating whether the LLR is positive or negative, the difference of the LLR, the positive representative value, and the negative representative value. [Effects of the Invention]

[0008] According to one aspect of the present invention, it is possible to compress LLR data temporarily stored in memory using HARQ. [Brief explanation of the drawings]

[0009] [Figure 1] 1 is a block diagram showing a schematic configuration of an HARQ device in a communication system according to one embodiment of the present invention. [Figure 2] FIG. 2 is a block diagram illustrating an overview of the communication system. [Figure 3] 10 is a graph showing a probability distribution of LLR included in LLR data acquired by a conversion unit of the HARQ device. [Figure 4] 10 is a flowchart showing a flow of conversion processing in the conversion unit. [Figure 5] 10 is a flowchart showing the flow of a restoration process in the restoration unit of the HARQ device. [Figure 6] 10 is a flowchart showing the flow of conversion processing in a conversion unit of an HARQ device in a communication system according to another embodiment of the present invention. [Figure 7] This is a diagram of a signal point arrangement in 16QAM. [Figure 8] 10 is a graph showing an example of a probability distribution of LLRs included in LLR data acquired by an HARQ device in a communication system according to yet another embodiment of the present invention. [Figure 9] 10 is a graph showing another example of the probability distribution of the LLR. DETAILED DESCRIPTION OF THE INVENTION

[0010] Hereinafter, embodiments of the present disclosure will be described in detail. For the sake of convenience, components having the same functions as those in the embodiments will be denoted by the same reference numerals, and descriptions thereof will be omitted as appropriate.

[0011] [Embodiment 1] An embodiment of the present disclosure will be described with reference to FIGS.

[0012] (Communication Systems) 2 is a block diagram showing an outline of a communication system according to this embodiment. The communication system 1 of this embodiment employs error correction codes and HARQ to achieve high-quality and highly reliable communication.

[0013] As shown in Fig. 2, in communication system 1 of this embodiment, communication device 2 that transmits a message and communication device 3 that receives the message can communicate via communication path 4. Hereinafter, communication device 2 that transmits a message will be referred to as "transmitting device 2," and communication device 3 that receives the message will be referred to as "receiving device 3." Transmitting device 2 includes an encoder 11 and a modulator 12. Receiving device 3 includes a demodulator 21, a device for HARQ 22, and a decoder 23.

[0014] In the transmitting device 2, the encoder 11 acquires the message and performs encoding on the acquired message for error correction at the receiving device 3. For the encoding, a coding method such as a low density parity check (LDPC) code, a turbo code, or a polar code is used. The encoding generates coded data including an information bit string corresponding to the message and a redundant bit string for error correction. The encoder 11 sends a portion of the generated coded data to the modulator 12 as transmission data.

[0015] The modulator 12 performs digital modulation on the transmission data from the encoder 11 in order to transmit information over the communication channel 4. For the digital modulation, a modulation method such as QPSK (Quadrature Phase Shift Keying) or multilevel QAM (Quadrature Amplitude Modulation) is used. A modulated signal corresponding to the transmission data is generated by the digital modulation. The modulator 12 transmits the generated modulated signal to the receiving device 3 over the communication channel 4.

[0016] In the receiving device 3, the demodulator 21 demodulates the modulated signal using a demodulation method corresponding to the modulation method, and generates LLR data corresponding to the transmission data. The demodulator 21 sends the generated LLR data to the HARQ device 22.

[0017] The LLR data includes a plurality of LLRs corresponding to the plurality of bits in the transmission data. The LLRs are expressed by the following equation (1). LLR = log{(probability that the corresponding bit is 0) / (probability that the corresponding bit is 1)} (1).

[0018] As can be seen from the above formula (1), the greater the LLR is from 0, the higher the probability that the corresponding bit is 0, and the smaller the LLR is from 0, the higher the probability that the corresponding bit is 1. Therefore, as the LLR approaches 0, it becomes unclear whether the corresponding bit is 0 or 1, and when the LLR is 0, it is completely unclear whether the corresponding bit is 0 or 1.

[0019] The HARQ device 22 receives and temporarily stores the LLR data from the demodulator 21. This LLR data corresponds to a part of the encoded data. The HARQ device 22 also combines one or more pieces of temporarily stored LLR data with the LLR data from the demodulator 21, and sends the combined data to the decoder 23 as LLR data for decoding.

[0020] Decoder 23 performs error correction and decoding on the LLR data to be decoded from HARQ device 22 using a decoding method corresponding to the encoding method. If the error correction is successful, decoder 23 outputs the decoded message, and receiver 3 transmits an ACK (acknowledgement of successful reception) to transmitter 2. In this case, transmitter 2 repeats the above operation to transmit a new message to receiver 3. On the other hand, if the error correction is unsuccessful, receiver 3 transmits a NACK to transmitter 2. In this case, encoder 11 of transmitter 2 sends the untransmitted portion or a part of the encoded data to modulator 12 as transmission data.

[0021] (Example of HARQ operation) An example of the operation of HARQ in the communication system 1 having the above configuration will be described. Note that in the following example, the coding rate R is 1 / 3, but the same applies when the coding rate R is other than 1 / 3.

[0022] In the transmitting device 2, the encoded data created by the encoder 11 has three times the bit length of the information bit string corresponding to the message. The encoder 11 first sends the first partial data, which includes the information bit string and a part of the redundant bit string, of the encoded data to the modulator 12 as transmission data.

[0023] In this case, the LLR data that the HARQ device 22 of the receiving device 3 receives from the demodulator 21 is first LLR data corresponding to the first partial data. The HARQ device 22 temporarily stores the first LLR data and sends the first LLR data to the decoder 23 as LLR data to be decoded. If the decoder 23 has successfully corrected the error in the LLR data to be decoded, it outputs the decoded message.

[0024] On the other hand, if the decoder 23 of the receiving device 3 fails to correct the error in the decoding LLR data, the encoder 11 of the transmitting device 2 sends second partial data, which includes a part of the encoded data other than the first partial data, to the modulator 12 as transmission data. Therefore, the second partial data does not include an information bit string, but includes a part of a redundant bit string.

[0025] In this case, the LLR data that the HARQ device 22 of the receiving device 3 receives from the demodulator 21 is second LLR data corresponding to the second partial data. The HARQ device 22 temporarily stores the second LLR data from the demodulator 21, and combines the first LLR data for which the error correction has failed with the second LLR data, and sends the combined data to the decoder 23 as LLR data for decoding. If the decoder 23 succeeds in error correction of the LLR data for decoding, it outputs the decoded message.

[0026] On the other hand, if the decoder 23 of the receiving device 3 fails to correct the error in the decoding LLR data, the encoder 11 of the transmitting device 2 sends the final partial data, which includes all of the coded data except for the first and second partial data, to the modulator 12 as transmission data. Therefore, the final partial data does not include the information bit string, but includes the remaining redundant bit string.

[0027] In this case, the LLR data that the HARQ device 22 of the receiving device 3 receives from the demodulator 21 is final LLR data corresponding to the final partial data. The HARQ device 22 combines the first LLR data and the second LLR data for which the error correction has failed with the final LLR data, and sends the combined data to the decoder 23 as LLR data to be decoded. Note that the HARQ device 22 does not need to temporarily store the final LLR data. If the decoder 23 has successfully corrected the error in the LLR data to be decoded, it outputs the decoded message.

[0028] On the other hand, if the decoder 23 fails to perform the error correction again, the transmitting device 2 and the receiving device 3 determine that the message transmission has failed and discard the coded data held by the encoder 11 and the first LLR data and the second LLR data temporarily stored in the HARQ device 22.

[0029] (HARQ equipment) Fig. 1 is a block diagram showing a schematic configuration of the HARQ device 22. As shown in Fig. 1, the HARQ device 22 includes an acquisition unit 31, a conversion unit 32, a bus matrix 33, a memory controller 34, a memory 35, a restoration unit 36, and a combination unit 37.

[0030] The acquiring unit 31 acquires the LLR data from the demodulator 21. The acquiring unit 31 sends the LLR data to the converting unit 32 and the combining unit 37. Note that, when the LLR data corresponds to the final partial data in the encoded data, the acquiring unit 31 does not need to send the LLR data to the converting unit 32. In this case, the operations of the converting unit 32, the bus matrix 33, the memory controller 34, and the memory 35 related to the LLR data corresponding to the final partial data are omitted.

[0031] The conversion unit 32 converts the LLR data from the acquisition unit 31 to generate converted data. The conversion unit 32 sends the converted data to the memory controller 34 via the bus matrix 33. Details of the conversion will be described later.

[0032] The conversion unit 32 and the restoration unit 36 ​​are connected to the memory controller 34 via a bus. The bus matrix 33 switches the device occupying the bus between the conversion unit 32 and the restoration unit 36. The bus matrix 33 prevents the conversion unit 32 and the restoration unit 36 ​​from accessing the memory controller 34 at the same time.

[0033] When the memory controller 34 receives the converted data from the conversion unit 32 via the bus matrix 33, it instructs the memory 35 to write the converted data. In addition, the memory controller 34 reads the converted data from the memory 35 in accordance with an instruction from the restoration unit 36 ​​via the bus matrix 33, and sends the converted data to the restoration unit 36 ​​via the bus matrix 33.

[0034] It is desirable that the memory controller 34 and memory 35 be high-speed controllers and memories that comply with standards such as DDR SDRAM (Double-Data-Rate Synchronous Dynamic Random Access Memory). The memory controller 34 may also control the memory 35 to overwrite conversion data related to a previous message with conversion data related to the current message. Alternatively, when the success or failure of message transmission is determined, the memory controller 34 may control the memory 35 to erase the conversion data related to the message.

[0035] The restoration unit 36 ​​instructs the memory controller 34 via the bus matrix 33 to acquire the transformed data read from the memory 35, and restores the original LLR data from the acquired transformed data. The restoration unit 36 ​​sends the restored LLR data to the combining unit 37. Details of the restoration will be described later.

[0036] The combining unit 37 combines the LLR data from the reconstruction unit 36 ​​with the LLR data from the acquisition unit 31. The combining unit 37 sends the combined LLR data to the decoder 23 as LLR data for decoding. Note that when the LLR data from the acquisition unit 31 corresponds to the first partial data in the encoded data, the combining unit 37 sends the LLR data from the acquisition unit 31 to the decoder 23 as LLR data for decoding because there is no LLR data from the reconstruction unit 36.

[0037] (Details of the conversion and restoration parts) Fig. 3 is a graph showing the probability distribution of LLR included in the LLR data acquired by the conversion unit 32. The upper part of Fig. 3 shows a case where the SNR (Signal-to-Noise Ratio) is high, and the lower part of Fig. 3 shows a case where the SNR is low. As shown in Fig. 3, it is known that the probability distribution when the LLR is positive and the probability distribution when the LLR is negative are approximated to Gaussian distributions.

[0038] Therefore, the conversion unit 32 first calculates a positive average value μ(+) which is the average value of positive LLRs in the LLR data, and a negative average value μ(-) which is the average value of negative LLRs in the LLR data. Next, for each LLR included in the LLR data, if the LLR is positive, the conversion unit 32 calculates a difference D of the LLR from the positive average value μ(+), and if the LLR is negative, the conversion unit 32 calculates a difference D of the LLR from the negative average value μ(-).

[0039] The conversion unit 32 then generates, as the conversion data, data that includes, for each of the LLRs, a sign indicating whether the LLR is positive or negative, a difference D between the LLRs, and also includes the magnitude of the positive average value μ(+) and the magnitude of the negative average value μ(−). The conversion unit 32 instructs the memory controller 34 via the bus matrix 33 to store the conversion data in the memory 35.

[0040] Meanwhile, the restoration unit 36 ​​instructs the memory controller 34 via the bus matrix 33 to acquire the transformation data stored in the memory 35. Next, for each of the signs indicating positive or negative LLRs and the LLR difference D included in the transformation data, if the signs indicate positive, the restoration unit 36 ​​adds the LLR difference D to the positive average value included in the transformation data, whereas if the signs indicate negative, the restoration unit 36 ​​adds the LLR difference D to the negative average value included in the transformation data to restore the LLRs. Then, the restoration unit 36 ​​creates LLR data including the restored LLRs and sends the LLR data to the combining unit 37.

[0041] For example, when the LLR is in the range of -1024 to 1023, the LLR can be expressed in 11 bits. On the other hand, for positive LLR, suppose the average value μ(+) is 300 and the standard deviation σ(+) is 60. Also, for negative LLR, suppose the average value μ(-) is -300 and the standard deviation σ(-) is 60.

[0042] In this case, the sign indicating whether the LLR is positive or negative can be expressed in 1 bit. If the difference D of the positive LLR falls within the range of ±3σ(+), it is in the range of -180 to 180 and can be expressed in 9 bits. If the difference D of the negative LLR falls within the range of ±3σ(-), it is in the range of -180 to 180 and can be expressed in 9 bits. The magnitudes (absolute values) of the positive average value μ(+) and the negative average value μ(-) can be expressed in 9 bits.

[0043] Therefore, if the number of LLRs included in the LLR data is n (n is an integer equal to or greater than 1), the bit length of the LLR data is 11×n (bits). Also, the bit length of the converted data is (1+9)×n+9+9=10×n+18 (bits).

[0044] Therefore, the condition under which the bit length of the transformed data is smaller than the bit length of the LLR data, i.e., the condition under which the transformed data is more compressed than the LLR data, is n>18. Furthermore, as n increases, the value (difference) obtained by subtracting the bit length of the transformed data from the bit length of the LLR data increases, i.e., the degree of compression increases. Recently, for example, the number of bits included in a bit string coded by an LDPC code or the like, i.e., the number n of LLRs included in the LLR data corresponding to the bit string, reaches approximately 8000. Therefore, the bit length of the transformed data becomes sufficiently smaller than the bit length of the LLR data.

[0045] As described above, the HARQ device 22 of this embodiment can generate transformed data by compressing LLR data, and can also restore the transformed data to the original LLR data.

[0046] 3, it can be seen that the magnitude of the positive average value μ(+) and the negative average value μ(-) is smaller when the SNR is low than when the SNR is high, and the range of the LLR difference D is smaller. Therefore, the degree of compression is greater when the SNR is low than when the SNR is high.

[0047] The bit length representing the difference D between the LLRs may be a predetermined value (for example, 9 bits), and the bit length representing the magnitude of the positive average value μ(+) and the negative average value μ(−) may be a predetermined value (for example, 9 bits). In this case, the capacity to be secured in the memory 35 for storing the transformation data can be kept constant.

[0048] (Conversion process of the conversion unit) Fig. 4 is a flowchart showing the flow of the conversion process (conversion method) in the conversion unit 32. As shown in Fig. 4, first, the conversion unit 32 initializes a variable i (i is an integer between 1 and n) to 1 (S11). Next, the conversion unit 32 calculates a positive average value μ(+) that is the average value of positive LLRs in the LLR data, and a negative average value μ(-) that is the average value of negative LLRs in the LLR data (S12).

[0049] Next, the conversion unit 32 determines whether LLR(i), which indicates the i-th LLR in the LLR data, is equal to or greater than 0 (S13). If LLR(i) is equal to or greater than 0 (YES in S13), the conversion unit 32 sets the bit value S(i), which indicates positive or negative, to 0, and calculates the difference D(i) of LLR(i) from the positive average value μ(+) (S14). Then, the process proceeds to step S16. On the other hand, if LLR(i) is less than 0 (NO in S13), the conversion unit 32 sets the bit value S(i), which indicates positive or negative, to 1, and calculates the difference D(i) of LLR(i) from the negative average value μ(-) (S15). Then, the process proceeds to step S16.

[0050] In step S16, the conversion unit 32 determines whether i=n. If i=n is not true (NO in S16), the conversion unit 32 increments the variable i by 1 (S17) and returns to step S13. On the other hand, if i=n is true (YES in S16), the conversion unit 32 creates conversion data including the magnitude of the positive average value μ(+), the magnitude of the negative average value μ(-), and S(i) and D(i) for each LLR(i) (S18). Then, the conversion unit 32 instructs the memory controller 34 via the bus matrix 33 to store the conversion data in the memory 35 (S19). Thereafter, the conversion process ends.

[0051] (Restoration process of the restoration part) Fig. 5 is a flowchart showing the flow of the restoration process in the restoration unit 36. As shown in Fig. 5, first, the restoration unit 36 ​​initializes a variable j (j is an integer between 1 and n) to 1 (S21). Next, the restoration unit 36 ​​instructs the memory controller 34 via the bus matrix 33 to acquire the conversion data stored in the memory 35, and acquires the magnitudes of the positive average value μ(+) and the negative average value μ(-) from the conversion data (S22).

[0052] Next, the restoration unit 36 ​​acquires a bit value S(j) indicating positive or negative and a difference D(j) for each LLR(j) from the converted data (S23), and determines whether the bit value S(j) indicating positive or negative is 0 or 1 (S24). If the bit value S(j) is 0, the restoration unit 36 ​​restores the LLR(j) by adding the difference D(j) to the positive average value μ(+) (S25). Then, the process proceeds to step S27. On the other hand, if the bit value S(j) is 1, the restoration unit 36 ​​restores the LLR(j) by adding the difference D(j) to the negative average value μ(-) (S26). Then, the process proceeds to step S27.

[0053] In step S27, the restoration unit 36 ​​determines whether j=n. If j=n is not true (NO in S27), the restoration unit 36 ​​increments the variable j by 1 (S28) and returns to step S23. On the other hand, if j=n is true (YES in S27), the restoration unit 36 ​​creates LLR data including the restored LLR(1) to LLR(n), and sends this to the combining unit 37 (S29). Thereafter, the restoration process ends.

[0054] [Embodiment 2] Another embodiment of the present disclosure will be described with reference to Fig. 6. The communication system 1 of this embodiment differs from the communication system 1 of the first embodiment in the operation of the conversion unit 32, but the other configurations are the same.

[0055] Incidentally, among the LLRs contained in the LLR data, there may be LLRs that deviate from the range of the positive average value μ(+)±3σ(+) and the range of the negative average value μ(-)±3σ(-) shown in Figure 3.

[0056] However, as mentioned above, the greater the LLR is from 0, the higher the probability that the corresponding bit is 0. Therefore, when LLR>μ(+)+3σ(+), the bit corresponding to the LLR is likely to be 0. Furthermore, the bit corresponding to LLR=μ(+)+3σ(+) is more likely to be 0 than the bit corresponding to LLR=μ(+)-3σ(+). From this, when LLR>μ(+)+3σ(+), even if LLR=μ(+)+3σ(+), the bit corresponding to LLR is still likely to be 0, and as a result, it is considered that there is little effect on error correction in the decoder 23 at the subsequent stage.

[0057] Furthermore, as described above, as the LLR approaches 0, it becomes unclear whether the corresponding bit is 0 or 1. Therefore, when LLR<μ(+)-3σ(+), it is highly likely that the bit corresponding to the LLR is unknown as being 0 or 1. Furthermore, it is highly likely that the bit corresponding to LLR=μ(+)-3σ(+) is unknown as being 0 or 1, compared to the bit corresponding to LLR=μ(+)+3σ(+). From this, when LLR<μ(+)-3σ(+), even if LLR=μ(+)-3σ(+), it is still highly likely that the bit corresponding to the LLR is unknown as being 0 or 1, and as a result, it is considered that there is little effect on error correction in the decoder 23 at the subsequent stage.

[0058] Similarly, when LLR<μ(-)-3σ(-), even if LLR=μ(-)-3σ(-), there is still a high probability that the bit corresponding to the LLR is 1, and as a result, it is thought that there will be little effect on error correction in the subsequent decoder 23. Also, when LLR>μ(-)+3σ(-), even if LLR=μ(-)+3σ(-), there is still a high probability that it is unknown whether the bit corresponding to the LLR is 0 or 1, and as a result, it is thought that there will be little effect on error correction in the subsequent decoder 23.

[0059] Therefore, in this embodiment, the conversion unit 32 sets an upper limit value and a lower limit value for the difference D of the LLR, and if the difference D exceeds the upper limit value, changes the difference D to the upper limit value, while if the difference D is below the lower limit value, changes the difference D to the lower limit value.

[0060] Specifically, when the LLR is positive, the conversion unit 32 sets the upper limit of the difference D to 3σ(+) and the lower limit of the difference D to −3σ(+). When the LLR is negative, the conversion unit 32 sets the upper limit of the difference D to 3σ(−) and the lower limit of the difference D to −3σ(−).

[0061] Then, when the LLR is positive, the conversion unit 32 changes the calculated difference D to D=3σ(+) when D>3σ(+), and changes it to D=-3σ(+) when D<-3σ(+).Furthermore, when the LLR is negative, the conversion unit 32 changes the calculated difference D to D=3σ(-) when D>3σ(-), and changes it to D=-3σ(-) when D<-3σ(-).

[0062] Therefore, the difference D falls between the lower limit and the upper limit, and the bit length of the bits indicating the difference D can be reduced. As a result, the data length of the converted data can be reduced, and the capacity of the memory 35 can be reduced. Furthermore, even if the conversion unit 32 changes the difference D as described above, it is considered that this will have little effect on error correction in the decoder 23 at the subsequent stage. Furthermore, by using 3σ as the upper and lower limits, most of the difference D of the LLR can fall within the range from the lower limit to the upper limit.

[0063] Fig. 6 is a flowchart showing the flow of the conversion process in the conversion unit 32. Compared to the conversion process shown in Fig. 4, the conversion process in the conversion unit 32 of this embodiment has the process shown in Fig. 6 added between step S16 and step S18.

[0064] 6, if i=n in step S16 (YES in S16), a bit value S(i) is determined for each LLR(i), and a difference D(i) is calculated. Then, the conversion unit 32 calculates a positive standard deviation σ(+), which is the standard deviation of positive LLRs, from the difference D(i) whose bit value S(i) is 0, and sets the upper limit of the difference D(i) to 3σ(+) and the lower limit of the difference D(i) to −3σ(+) (S31). Furthermore, the conversion unit 32 calculates a negative standard deviation σ(−), which is the standard deviation of negative LLRs, from the difference D(i) whose bit value S(i) is 1, and sets the upper limit of the difference D(i) to 3σ(−) and the lower limit of the difference D(i) to −3σ(−) (S32).

[0065] Next, the conversion unit 32 initializes a variable k (k is an integer between 1 and n) to 1 (S33). Next, the conversion unit 32 determines whether the bit value S(k) is 0 or 1 (S34).

[0066] When the bit value S(k) is 0, if the difference D(k) exceeds the upper limit 3σ(+), the conversion unit 32 changes the difference D(k) to the upper limit 3σ(+) (S35 and S36). On the other hand, if the difference D(k) is below the lower limit -3σ(+), the conversion unit 32 changes the difference D(k) to the lower limit -3σ(+) (S37 and S38). Thereafter, the process proceeds to step S43. Note that if the difference D(k) is within the range from the lower limit -3σ(+) to the upper limit 3σ(+), the difference D(k) is not changed. Also, steps S37 and S38 may be performed before steps S35 and S36.

[0067] On the other hand, when the bit value S(k) is 1, and the difference D(k) exceeds the upper limit 3σ(-), the conversion unit 32 changes the difference D(k) to the upper limit 3σ(-) (S39 and S40). On the other hand, when the difference D(k) is below the lower limit -3σ(-), the conversion unit 32 changes the difference D(k) to the lower limit -3σ(-) (S41 and S42). Thereafter, the process proceeds to step S43. Note that when the difference D(k) is within the range from the lower limit -3σ(-) to the upper limit 3σ(-), the difference D(k) is not changed. Also, steps S41 and S42 may be performed before steps S39 and S40.

[0068] Next, in step S43, the conversion unit 32 determines whether k=n. If k=n is not true (NO in S43), the conversion unit 32 increments the variable k by 1 (S44) and returns to step S34. On the other hand, if k=n is true (YES in S43), the conversion unit 32 ends the conversion process and proceeds to step S18.

[0069] [Embodiment 3] Another embodiment of the present disclosure will be described with reference to Figures 7 to 9. The communication system 1 of this embodiment is different from the communication systems 1 of the first and second embodiments in that the digital modulation method in the modulator 12 is 16QAM, but the other configurations are the same.

[0070] In a digital modulation method such as 16QAM, the modulator 12 generates a modulated signal by modulating multiple bits into one symbol and transmits the modulated signal to the receiving device 3 via the communication path 4. In this case, the demodulator 21 in the receiving device 3 generates LLR data including multiple LLRs, which are LLRs corresponding to a certain bit among the multiple bits and are included in multiple symbols, and performs this for each of the multiple bits to generate multiple pieces of LLR data. The receiving device 3 may also include multiple HARQ devices 22 that respectively use the multiple pieces of LLR data generated by the demodulator 21. The decoder 23 then rearranges the multiple pieces of decoding LLR data from the multiple HARQ devices 22 into multiple symbols and performs error correction and decoding on the multiple symbols.

[0071] In the case of 16QAM, the modulator 12 in the transmitting device 2 modulates four bits (b0, b1, b2, b3) into one symbol to create a modulated signal, and transmits the modulated signal to the receiving device 3 via the communication path 4. Therefore, in the receiving device 3 of this embodiment, the demodulator 21 creates LLR data including a plurality of LLRs, which are LLRs corresponding to bit b0 and are included in a plurality of symbols, and performs this for each of bits b1, b2, and b3 to create four pieces of LLR data corresponding to the four bits (b0, b1, b2, b3). The receiving device 3 includes four HARQ devices 22 that respectively use the four pieces of LLR data created by the demodulator 21. The decoder 23 rearranges the four pieces of restoration LLR data from the four HARQ devices 22 into a plurality of symbols and performs error correction and decoding on the plurality of symbols.

[0072] 7 is a diagram showing the arrangement of signal points in 16QAM. The I axis represents the I-phase (in-phase) symbols (b0, b2), and the Q axis represents the Q-phase (quadrature) symbols (b1, b3). The modulator 12 modulates a cosine wave (cos θ) carrying the I-phase symbols (b0, b2) and a sine wave (-sin θ) carrying the Q-phase symbols (b1, b3) into a combined wave and transmits the combined wave to the receiving device 3 via the communication path 4.

[0073] Meanwhile, the demodulator 21 separates the modulated wave into the cosine wave and the sine wave, and calculates the LLR of each bit. Therefore, the probability distribution of the LLR of bits b0 and b2 included in the I-phase symbol differs from the probability distribution of the LLR of bits b1 and b3 included in the Q-phase symbol.

[0074] Fig. 8 is a graph showing an example of a probability distribution of LLRs included in LLR data acquired by the HARQ device 22 of this embodiment. The upper part of Fig. 8 shows the probability distribution of LLRs of bits b1 and b3 included in the Q-phase symbol, and the lower part of Fig. 8 shows the probability distribution of LLRs of bits b0 and b2 included in the I-phase symbol.

[0075] The probability distribution shown in the upper part of Fig. 8 is the same as the probability distribution shown in Fig. 3. Therefore, the conversion unit 32 of the HARQ device 22 corresponding to each of the bits b1 and b3 included in the Q-phase symbol may be the same as the conversion unit 32 of the HARQ device 22 of the first embodiment.

[0076] On the other hand, the probability distribution shown in the bottom of Figure 8 has two positive Gaussian distributions and two negative Gaussian distributions, meaning that in this case there are two positive mean values ​​and two negative mean values.

[0077] However, the Gaussian distribution G2(+) having the mean value farther from 0 out of the two positive mean values ​​is more likely to have a corresponding bit that is 0 than the Gaussian distribution G1(+) having the mean value closer to 0. Furthermore, when upper and lower limit values ​​are set for the Gaussian distribution G1(+) having the mean value closer to 0, the bit corresponding to the upper limit value is more likely to be 0 than the bit corresponding to the lower limit value.

[0078] Therefore, in this embodiment, the conversion unit 32 of the HARQ device 22 sets the one of the two positive average values ​​closer to 0 as the positive average value μ(+). Similarly to the conversion unit 32 of the second embodiment, the conversion unit 32 sets an upper limit value 3σ(+) and a lower limit value −3σ(+), and changes the difference of the LLRs included in the Gaussian distribution G2(+) having an average value farther from 0 to the upper limit value 3σ(+). Even in this case, the possibility that the corresponding bit is 0 is still high, and as a result, it is considered that there will be little effect on error correction in the decoder 23 at the subsequent stage.

[0079] Similarly, the Gaussian distribution G2(-) having the mean value farther from 0 out of the two negative mean values ​​is more likely to have a corresponding bit that is 1 than the Gaussian distribution G1(-) having the mean value closer to 0. Furthermore, when upper and lower limit values ​​are set for the Gaussian distribution G1(-) having the mean value closer to 0, the bit corresponding to the upper limit value is more likely to be 1 than the bit corresponding to the lower limit value.

[0080] Therefore, in this embodiment, the conversion unit 32 of the HARQ device 22 sets the one of the two negative average values ​​closer to 0 as the negative average value μ(-). Similarly to the conversion unit 32 of the second embodiment, the conversion unit 32 sets an upper limit 3σ(-) and a lower limit −3σ(-), and changes the difference of the LLR included in the Gaussian distribution G2(-) having the average value farther from 0 to the upper limit 3σ(-). Even in this case, the possibility that the corresponding bit is 1 is still high, and as a result, it is considered that there will be little effect on error correction in the decoder 23 at the subsequent stage.

[0081] 8, the LLR ranges from -1024 to 1023, and is expressed by 11 bits. On the other hand, for positive LLR, the average value μ(+) is 180, and the standard deviation σ(+) is 40. For negative LLR, the average value μ(-) is -180, and the standard deviation σ(-) is 40.

[0082] In this case, the sign indicating whether the LLR is positive or negative can be expressed in 1 bit. If the difference D of the positive LLR falls within the range of ±3σ(+), it is in the range of -120 to 120 and can be expressed in 8 bits. If the difference D of the negative LLR falls within the range of ±3σ(-), it is in the range of -120 to 120 and can be expressed in 8 bits. The magnitudes (absolute values) of the positive average value μ(+) and the negative average value μ(-) can be expressed in 8 bits.

[0083] Therefore, if the number of LLRs included in the LLR data is n (n is an integer equal to or greater than 1), the bit length of the LLR data is 11×n (bits). Also, the bit length of the converted data is (1+8)×n+8+8=9×n+16 (bits).

[0084] Therefore, the condition under which the bit length of the transformed data is smaller than the bit length of the LLR data, that is, the condition under which the transformed data is more compressed than the LLR data, is n>8. Furthermore, the degree of compression is greater than in the example of the first embodiment.

[0085] Fig. 9 is a graph showing another example of the probability distribution of LLRs included in the LLR data acquired by the HARQ device 22 of this embodiment. The upper part of Fig. 9 shows the probability distribution of LLRs of bits b1 and b3 included in the Q-phase symbol, and the lower part of Fig. 9 shows the probability distribution of LLRs of bits b0 and b2 included in the I-phase symbol.

[0086] The probability distribution shown in Fig. 9 shows a case where the SNR is lower than that shown in Fig. 8. The probability distribution shown in the upper part of Fig. 9 is a Gaussian distribution with an average value of 0. In this case, similar to the conversion unit 32 of the HARQ device 22 of the first embodiment, the conversion unit 32 may calculate an average positive value μ(+) from a plurality of positive LLRs included in the LLR data, and calculate a difference D between each positive LLR and the average positive value μ(+). Furthermore, the conversion unit 32 may calculate an average negative value μ(-) from a plurality of negative LLRs included in the LLR data, and calculate a difference D between each negative LLR and the average negative value μ(-).

[0087] On the other hand, the probability distribution shown in the lower part of FIG. 9 has a Gaussian distribution G1(0) with a mean value of 0, a Gaussian distribution G2(+) with a positive mean value, and a Gaussian distribution G2(-) with a negative mean value.

[0088] However, a Gaussian distribution G2(+) with a positive mean value is more likely to have a corresponding bit that is 0 than a Gaussian distribution G1(0) with a mean value of 0. Furthermore, if an upper limit value 3σ(+) is set for a Gaussian distribution G1(0) with a mean value of 0, the bit corresponding to the upper limit value 3σ(+) is more likely to be 0 than a bit corresponding to 0.

[0089] Therefore, in this embodiment, the conversion unit 32 of the HARQ device 22 sets a positive upper limit 3σ(+) and a positive lower limit 0, and changes the difference of the LLR included in the Gaussian distribution G2(+) with a positive average value to the upper limit 3σ(+). Even in this case, the possibility that the corresponding bit is 0 is still high, and as a result, it is considered that there is little influence on error correction in the decoder 23 at the subsequent stage.

[0090] Similarly, a Gaussian distribution G2(-) with a negative mean value is more likely to have a corresponding bit that is 1 than a Gaussian distribution G1(0) with a mean value of 0. Furthermore, if a lower limit value of -3σ is set for a Gaussian distribution G1(0) with a mean value of 0, the bit corresponding to the lower limit value -3σ is more likely to be 1 than a bit corresponding to 0.

[0091] Therefore, in this embodiment, the conversion unit 32 of the HARQ device 22 sets a negative lower limit -3σ(-) and a negative upper limit 0, and changes the difference of the LLR included in the Gaussian distribution G2(-) with a negative average value to the lower limit -3σ(+). Even in this case, the possibility that the corresponding bit is 1 is still high, and as a result, it is considered that there is little influence on error correction in the decoder 23 at the subsequent stage.

[0092] 9, the LLR ranges from -1024 to 1023, and is expressed by 11 bits. On the other hand, for positive LLR, the average value μ(+) is 63, and the standard deviation σ(+) is 21. For negative LLR, the average value μ(-) is -63, and the standard deviation σ(-) is 21.

[0093] In this case, the sign indicating whether the LLR is positive or negative can be expressed in 1 bit. If the difference D of the positive LLR falls within the range of ±3σ(+), it is in the range of -63 to 63 and can be expressed in 7 bits. If the difference D of the negative LLR falls within the range of ±3σ(-), it is in the range of -63 to 63 and can be expressed in 7 bits. The magnitudes (absolute values) of the positive average value μ(+) and the negative average value μ(-) can be expressed in 6 bits.

[0094] Therefore, if the number of LLRs included in the LLR data is n (n is an integer equal to or greater than 1), the bit length of the LLR data is 11×n (bits). Also, the bit length of the converted data is (1+7)×n+6+6=7×n+12 (bits).

[0095] Therefore, the condition under which the bit length of the transformed data is smaller than the bit length of the LLR data, i.e., the condition under which the transformed data is more compressed than the LLR data, is n>3. Also, the degree of compression is greater than in the example of the first embodiment. Furthermore, comparing the examples of Fig. 8 and Fig. 9, it can be seen that the degree of compression is greater when the SNR is small than when the SNR is large.

[0096] (Special Notes) In the above embodiment, an LLR of 0 is regarded as a positive LLR, but it may be regarded as a negative LLR. Furthermore, an LLR of 0 can be expressed by 1 bit, so it may be stored in the memory 35 as it is.

[0097] In addition, although the above embodiment uses the mean value as the representative value of the Gaussian distribution, any representative value such as the median, mode, etc. In addition, in the above embodiment, 3σ is used as the upper and lower limit values, but any value representing the upper and lower limit values ​​such as the half-value width can be used.

[0098] [Software implementation example] The functions of the HARQ device 22 (hereinafter referred to as the "device") can be realized by a program that causes a computer to function as the device, and a program that causes a computer to function as each control block of the device (particularly the conversion unit 32 and the restoration unit 36).

[0099] In this case, the device includes a computer having at least one control device (e.g., a processor) and at least one storage device (e.g., a memory) as hardware for executing the program. The control device and storage device execute the program, thereby realizing the functions described in each of the above embodiments.

[0100] The program may be non-transitory and may be recorded on one or more computer-readable recording media. The recording media may or may not be included in the device. In the latter case, the program may be supplied to the device via any wired or wireless transmission medium.

[0101] Furthermore, some or all of the functions of the control blocks can be realized by logic circuits. For example, an integrated circuit in which a logic circuit that functions as each of the control blocks is formed is also included in the scope of the present invention. In addition, the functions of the control blocks can also be realized by, for example, a quantum computer.

[0102] Furthermore, each process described in each of the above embodiments may be executed by AI (Artificial Intelligence). In this case, the AI ​​may run on the control device or on another device (for example, an edge computer or a cloud server).

[0103] 〔summary〕 The HARQ device 22 according to aspect 1 of the present invention includes an acquisition unit 31 that acquires LLR data including a plurality of LLRs of 1 bit, and a conversion unit 32 that converts the LLR data to create converted data and stores the converted data in a memory 35. The conversion unit 32 obtains a positive representative value that is a value that represents a positive LLR in the LLR data and a negative representative value that is a value that represents a negative LLR in the LLR data, and, for each LLR included in the LLR data, if the LLR is positive, calculates a difference of the LLR from the positive representative value, while, if the LLR is negative, calculates a difference of the LLR from the negative representative value, and stores data including a sign indicating whether the LLR is positive or negative, the difference of the LLR, the positive representative value, and the negative representative value as the converted data in the memory 35.

[0104] According to the above configuration, the transformation data includes a sign indicating whether the LLR is positive or negative and a difference between the LLRs instead of the LLRs. The difference between the LLRs can have a shorter bit length than the LLRs. Therefore, the transformation data can be compressed from the original LLR data.

[0105] The HARQ device 22 according to a second aspect of the present invention, in accordance with the first aspect, further includes a restoration unit 36 ​​that reads the transformed data from the memory 35 and restores the transformed data to LLR data, and the restoration unit 36 ​​may restore the LLRs by adding the LLR difference to a positive representative value included in the transformed data when the sign indicating positive or negative of the LLR and the LLR difference included in the transformed data are positive, and by adding the LLR difference to a negative representative value included in the transformed data when the sign is negative, thereby restoring the LLRs and creating LLR data including the restored LLRs. In this case, the original LLR data can be restored from the transformed data.

[0106] In the HARQ device 22 according to aspect 3 of the present invention, in the above aspect 1 or 2, the conversion unit 32 may set an upper limit and a lower limit for the LLR difference, and if the LLR difference exceeds the upper limit, change the LLR difference to the upper limit, while if the LLR difference is below the lower limit, change the LLR difference to the lower limit. In this case, the LLR difference falls within a range from the lower limit to the upper limit. This makes it possible to align the bit length of a bit string indicating the LLR difference to a bit length corresponding to the range.

[0107] In the HARQ device 22 according to aspect 4 of the present invention, in the above-mentioned aspect 3, the conversion unit 32 may set the upper and lower limit values ​​of the LLR difference when the LLR is positive based on the standard deviation of the positive LLR in the LLR data, and may set the upper and lower limit values ​​of the LLR difference when the LLR is negative based on the standard deviation of the negative LLR in the LLR data.

[0108] The distribution of the LLRs is approximated to a Gaussian distribution. Therefore, according to the above configuration, the conversion unit 32 sets the upper limit and the lower limit based on the standard deviation of the LLRs, so that most of the differences in the LLRs can fall within the range from the lower limit to the upper limit.

[0109] In the HARQ device 22 according to aspect 5 of the present invention, in the above aspects 1 to 4, the conversion unit 32 may select the representative value closest to 0 as the positive representative value when there are multiple positive representative values ​​in the LLR data, and may select the representative value closest to 0 as the negative representative value when there are multiple negative representative values ​​in the LLR data.

[0110] According to the above configuration, among a plurality of probability distributions each having a plurality of positive representative values, a probability distribution having another representative value closest to 0 is more likely to have a bit of 0. Furthermore, when an upper limit and a lower limit are set for the probability distribution having the representative value closest to 0, the bit corresponding to the upper limit is more likely to have a bit of 0 than the bit corresponding to the lower limit. For this reason, even if the difference between the LLRs included in a probability distribution having another representative value is changed to the upper limit, the bit is still more likely to have a bit of 0, and as a result, it is considered that there is little impact on error correction at a later stage.

[0111] Similarly, among a plurality of probability distributions each having a plurality of negative representative values, a probability distribution having another representative value is more likely to have a bit of 1 than a probability distribution having a representative value closest to 0. Furthermore, when an upper limit and a lower limit are set for a probability distribution having a representative value closest to 0, the bit corresponding to the lower limit is more likely to be 1 than the bit corresponding to the upper limit. For this reason, even if the difference between the LLRs included in a probability distribution having another representative value is changed to the lower limit, the bit is still more likely to be 0, and as a result, it is thought that there will be little impact on error correction at a later stage.

[0112] The positive representative value may be any one of the average value, median value, and mode value of the positive LLRs in the LLR data, and the negative representative value may be any one of the average value, median value, and mode value of the negative LLRs in the LLR data.

[0113] A communication device 3 according to a sixth aspect of the present invention is a communication device 3 that performs HARQ and includes a demodulator 21 that demodulates a signal that has been coded and modulated by another communication device 2 to create LLR data, the HARQ device 22 according to the second aspect above that creates transformed data by converting the LLR data, stores the transformed data in a memory 35, and reads the transformed data from the memory 35 to restore the LLR data, and a decoder 23 that performs error correction and decoding on the LLR data from the HARQ device 22. In this case, the same effects as those of the second aspect above can be achieved.

[0114] A communication device 3 according to a seventh aspect of the present invention is the same as in the sixth aspect, in which the modulated signal modulates a plurality of bits into one symbol, the demodulator 21 creates LLR data including a plurality of LLRs that correspond to a certain bit among the plurality of bits and are included in a plurality of symbols, and creates a plurality of pieces of LLR data by performing this for each of the plurality of bits, and the HARQ device 22 may be a plurality of HARQ devices 22 that respectively use the plurality of LLR data created by the demodulator 21. In this case, the method can also be applied to a modulation method that transmits symbols including a plurality of bits, such as multi-level QAM.

[0115] A conversion method according to aspect 8 of the present invention is a conversion method in HARQ device 22 that creates conversion data by converting LLR data including a plurality of LLRs of 1 bit, and includes the steps of: calculating a positive representative value that is a value that represents positive LLRs in the LLR data; and a negative representative value that is a value that represents negative LLRs in the LLR data; calculating, for each LLR included in the LLR data, if the LLR is positive, a difference of the LLR from the positive representative value; and, if the LLR is negative, calculating a difference of the LLR from the negative representative value; and storing, for each LLR, data including a sign indicating whether the LLR is positive or negative, the difference of the LLR, the positive representative value, and the negative representative value in memory 35 as the conversion data.

[0116] According to the above method, the same effects as those of the first aspect can be achieved.

[0117] The HARQ device 22 according to each aspect of the present invention may be realized by a computer. In this case, the control program for the HARQ device 22, which causes the computer to operate as each unit (software element) of the HARQ device 22, thereby realizing the HARQ device 22 on the computer, and the computer-readable recording medium on which the control program is recorded, also fall within the scope of the present invention.

[0118] The present invention is not limited to the above-described embodiments, and various modifications are possible within the scope of the claims. Embodiments obtained by appropriately combining the technical means disclosed in different embodiments are also included in the technical scope of the present invention. Furthermore, new technical features can be formed by combining the technical means disclosed in each embodiment. [Explanation of symbols]

[0119] 1. Communication Systems 2. Communication equipment (transmitting equipment) 3. Communication equipment (receiving equipment) 4 Communication Channel 11 encoder 12 Modulator 21 Demodulator 22 HARQ equipment 23 Decoder 31 Acquisition Department 32 Conversion unit 33 Bus Matrix 34 Memory Controller 35 memory 36 Restoration Section 37 Joint

Claims

1. An apparatus for HARQ, an acquisition unit that acquires LLR data including a plurality of LLRs of bits; a conversion unit that converts the LLR data to generate conversion data and stores the conversion data in a memory; The conversion unit determining a positive representative value that is a value that represents a positive LLR in the LLR data and a negative representative value that is a value that represents a negative LLR in the LLR data; For each LLR included in the LLR data, if the LLR is positive, a difference of the LLR with respect to the positive representative value is calculated, and if the LLR is negative, a difference of the LLR with respect to the negative representative value is calculated; The HARQ device stores, for each of the LLRs, data including a sign indicating whether the LLR is positive or negative, a difference between the LLRs, the positive representative value, and the negative representative value in the memory as the transformation data.

2. a restoration unit that reads the transformed data from the memory and restores the transformed data to LLR data; The restoration unit is Regarding a sign indicating positive or negative of the LLR included in the transformation data and a difference between the LLRs, when the sign indicates positive, the difference between the LLRs is added to a positive representative value included in the transformation data, and when the sign indicates negative, the difference between the LLRs is added to a negative representative value included in the transformation data, thereby restoring the LLRs; The HARQ device according to claim 1 , which generates LLR data including a plurality of restored LLRs.

3. The conversion unit setting upper and lower limits of the difference between the LLRs; 2. The HARQ device according to claim 1, wherein when the difference between the LLRs exceeds the upper limit, the difference between the LLRs is changed to the upper limit, and when the difference between the LLRs is below the lower limit, the difference between the LLRs is changed to the lower limit.

4. 4. The HARQ device according to claim 3, wherein the conversion unit sets upper and lower limit values ​​of the LLR difference when the LLR is positive based on a standard deviation of positive LLRs in the LLR data, and sets upper and lower limit values ​​of the LLR difference when the LLR is negative based on a standard deviation of negative LLRs in the LLR data.

5. The conversion unit If there are a plurality of positive representative values ​​in the LLR data, a representative value closest to 0 is selected as the positive representative value; The HARQ device according to claim 1 , wherein when there are a plurality of negative representative values ​​in the LLR data, a representative value closest to 0 is selected as the negative representative value.

6. 6. The HARQ device according to claim 1, wherein the positive representative value is one of an average value, a median value, and a mode value of positive LLRs in the LLR data, and the negative representative value is one of an average value, a median value, and a mode value of negative LLRs in the LLR data.

7. A communication device that performs HARQ, a demodulator that demodulates a modulated signal that has been coded and modulated by another communication device to generate LLR data; the HARQ device according to claim 2, which generates transformed data by transforming the LLR data, stores the transformed data in a memory, and reads the transformed data from the memory to restore the LLR data; a decoder that performs error correction and decoding on the LLR data from the HARQ device.

8. the modulated signal is modulated into one symbol by multiple bits, the demodulator generates LLR data including a plurality of LLRs, each LLR corresponding to a certain bit among the plurality of bits and included in a plurality of symbols, and performs this for each of the plurality of bits to generate a plurality of the LLR data; The communication device according to claim 7 , wherein the HARQ device comprises a plurality of HARQ devices each using a plurality of LLR data generated by the demodulator.

9. A control program for causing a computer to function as the HARQ device according to claim 1, the control program causing the computer to function as the conversion unit.

10. A conversion method in an HARQ device for generating converted data by converting LLR data including a plurality of LLRs of bits, comprising: determining a positive representative value that is a value that represents positive LLRs in the LLR data and a negative representative value that is a value that represents negative LLRs in the LLR data; For each LLR included in the LLR data, if the LLR is positive, calculating a difference of the LLR with respect to the positive representative value, and if the LLR is negative, calculating a difference of the LLR with respect to the negative representative value; and storing, for each of the LLRs, data including a sign indicating whether the LLR is positive or negative, a difference between the LLRs, the positive representative value, and the negative representative value in a memory as the transformation data.

Images