JP7054762B2 - Wireless communication system and wireless communication method - Google Patents

Description

The present invention relates to a repetitive error correction processing technique in a wireless communication system.

In a wireless transmission device (FPU: Field Pick-up Unit) for transmitting television broadcast program material, a QAM (Quadrature Amplitude Modulation) method is adopted as a microwave band wireless transmission method (for example, non-patented). See Document 1). In the error correction of this wireless transmission method, a concatenated code method is adopted. In the concatenated code system, an RS (Reed-Solomon) code is used as the external code and a convolutional code is used as the internal code, and these are connected via an interleaver. On the receiver side, for the demodulated signal, Viterbi decoding is used as error correction for the internal code, and RS decoding is used for error correction for the external code.

FIG. 1 shows an example of a wireless communication system adopting the wireless transmission method disclosed in Non-Patent Document 1. In the following description, the processing of blocks 11 to 15 is performed on the transmitter side, and the processing of blocks 17 to 21 is performed on the receiver side.

Information signals such as images and data to be transmitted are input to the RS coding unit 11. The RS coding unit 11 packets the input information signal in units of a predetermined length, and RS-encodes the packet. For example, as the RS code used in the FPU, a (204,188) code having a code length of 204 bytes and a tissue length of 188 bytes is adopted. The order of the signal after RS coding by the RS coding unit 11 is rearranged by the interleaving unit 12 in order to improve the error correction capability in the receiver.

The signal after sorting by the interleave unit 12 is converted into (m-1) bit parallel data by the S / P conversion unit 13. m indicates the number of bits to be loaded per transmission symbol. For example, when m = 4, it becomes a digital multi-value modulation method such as 16QAM method (16-value quadrature amplitude modulation method), and when m = 6, it becomes a digital multi-value modulation method such as 64QAM method (64-value quadrature amplitude modulation method). When converting to parallel data, it is divided into the lower 1 bit (hereinafter referred to as "lower bit") of the parallel data and the remaining (m-2) bit (hereinafter referred to as "upper bit"), and is lower. The bits are input to the convolutional coding unit 14, and the high-order bits are input to the modulation unit 15.

The convolutional coding unit 14 performs convolutional coding by calculating an exclusive OR for a combination of predetermined positions in the information bit sequence. In the modulation unit 15, the high-order bit and the data of the signal 2 bits after the convolutional coding by the convolutional coding unit 14 are converted into a radio signal according to the 2 ^m QAM modulation method. Although the single carrier method is used as the secondary modulation method in Non-Patent Document 1, it is also possible to use a modulation method such as OFDM (Orthogonal Frequency Division Multiplexing), which is essentially the present invention. There are no restrictions on the modulation method.

The signal modulated by the modulation unit 15 is input to the demodulation unit 17 via the propagation path 16. The demodulation unit 17 performs demodulation processing corresponding to the modulation in the modulation unit 15. As the demodulation result, the demodulation unit 17 outputs a rigid determination decoding result corresponding to the information bit LLR (Log Likelihood Ratio; log-likelihood ratio) for the upper bit to the P / S unit 19, and outputs the coded bit LLR for the lower bit. It is output to the bitabi decoding unit 18. The information bit LLR is the reliability information of the information bit with respect to the high-order bit. The coded bit LLR is the reliability information of the coded bit with respect to the lower bits.

In the Viterbi decoding unit 18, the coded bit LLR obtained in the demodulation unit 17 is expanded on the trellis diagram, the maximum likelihood LLR sequence is selected, and the rigid determination decoding result corresponding to the sequence is converted into the P / S conversion unit. Output to 19. The P / S conversion unit 19 outputs the rigid determination decoding result for the upper bit and the rigid determination decoding result for the lower bit to the deinterleave unit 20. The signal output from the P / S conversion unit 19 is rearranged in the reverse order of the interleaving unit 12 by the deinterleaving unit 20, and the order is restored. The result of the deinterleave unit 20 is input to the RS decoding unit 21 and RS decoding processing is performed. In the case of the (204,188) code, if the bit error rate before RS decoding is 2 × 10 ^-4 or less, it can be pseudo-error-free by RS decoding, and material transmission without transmission error is possible. Become.

Various inventions have been proposed so far regarding the repetitive error correction processing technique in the QAM method. For example, in the invention disclosed in Patent Document 1, on the transmission device side, a 1-bit signal of the serial / parallel-converted transmission data is input to the convolutional coder, and the output 2 bits and the convolutional coder are input to the signal. Depending on the combination with other signals that did not, it is divided into a real part and an imaginary part, and the signal points to be transmitted independently are determined. On the receiving device side, the real number part and the imaginary number part are independently determined for the area and the metric is set, and the result of the Viterbi decoding is input to the convolutional coder having the same configuration as the coder on the transmitting device side, and the output and the area judgment information are input. Decrypt the transmitted data from.

Japanese Unexamined Patent Publication No. 2001-345785

ARIB STD-B11 2.2 version, "Portable microwave band digital wireless transmission system for transmission of television broadcast program material", Association of Radio Industries and Businesses JPO: Standard Technology Collection [Technology Classification] 3-3-2 MIMO Peripheral Technology / Error Control Technology / Error Correction Decoding Technology, <https://www.jpo.go.jp/resources/report/sonota/document/ hyoujun_gijutsu / h29_mimo.pdf ＞.

The FPU is often received in an environment where the reception level is low and the CNR (Carrier to Noise power Ratio) is low. In such a poor environment, the bit error rate becomes large, and the error rate after Viterbi decoding may exceed 2 × 10 ^-4 , making correct viewing impossible. The CNR required for correct transmission (the bit error rate after external decoding is 0) is referred to as a required CNR. Improving the required CNR by improving the error correction capability is one of the permanent challenges of radios.

The present invention has been made in view of the above-mentioned conventional circumstances, and an object of the present invention is to improve the required CNR of a wireless communication system that wirelessly transmits a signal from a transmitting device to a receiving device.

In the present invention, in order to achieve the above object, a wireless communication system that wirelessly transmits a signal from a transmitting device to a receiving device is configured as follows.
That is, the transmission device performs the following processing. The convolutional coding means performs convolutional coding on the second bit of the information bit sequence obtained by RS coding for the signal to be transmitted, excluding the predetermined first bit. The modulation means performs modulation processing on the first bit and the coded bit obtained by the convolutional coding means.

In addition, the receiving device performs the following processing. The first demodulation means performs demodulation processing on the received signal from the transmitting device, and calculates the bit log-likelihood ratio with respect to the coded bits. The maximal post-probability decoding means maximizes the post-probability based on the bit log likelihood ratio to the coded bits calculated by the first demodulation means and the first prior information to the information bits of the second bit. The information bit of the second bit is decoded, and the logarithmic likelihood ratio with respect to the second bit and the second prior information with respect to the information bit of the second bit are calculated. The second demodulation means determines the log-likelihood ratio with respect to the first bit based on the received signal, the second prior information calculated by the maximum a posteriori decoding means, and the third prior information with respect to the information bit of the first bit. calculate. The decoding means decodes the signal to be transmitted based on the log-likelihood ratio to the first bit calculated by the second demodulation means and the log-likelihood ratio to the second bit calculated by the maximum a posteriori decoding means. do. The prior information feedback means generates the first prior information and feeds it back to the maximum a posteriori decoding means based on the information of the decoding result output from the decoding means, and also generates the third prior information to generate the second demodulation means. Give feedback to.

According to such a configuration, the maximum a posteriori decoding and the RS decoding are repeatedly repeated in the receiving device. As a result, each error correction capability can be gradually improved, so that a large improvement effect can be obtained after a predetermined number of repetitions. Therefore, it is possible to provide a wireless communication system that can improve the required CNR.

Here, as an example of the configuration, the transmission device further performs the following processing. That is, the RS coding means packetizes the signal to be transmitted in units of a predetermined length, and RS-encodes the packet. The first interleaving means rearranges the order after RS coding by the RS coding means to acquire an information bit sequence. In the information bit sequence acquired by the first interleaving means, the first bit portion is input to the modulation means and the second bit portion is input to the convolutional coding means.

In addition, the prior information feedback means of the receiving device performs the following processing. The second interleaving means sorts the information of the decoding result output from the decoding means in the same order as the first interleaving means of the transmitting device. The prior information generation means distributes the information after sorting by the second interleaving means into the information for the first bit or the information for the second bit, and feeds back the information for the first bit as the third information to the second demodulation means. The information for 2 bits is fed back to the maximum post-probability decoding means as the first information.

Further, the decoding means of the receiving device performs the following processing. The log-likelihood ratio rearranging means rearranges the order of the log-likelihood ratio to the first bit calculated by the second demographic means and the log-likelihood ratio to the second bit calculated by the maximum a posteriori decoding means. The log-likelihood ratio means hard-determines the log-likelihood ratio after sorting by the log-likelihood ratio sorting means, and calculates the decoding result. The conversion means converts the log-likelihood ratio after sorting by the log-likelihood ratio sorting means into reliability. The deinterleaving means rearranges the decoding result by the rigid determination means and the conversion result by the conversion means in the reverse order of the second interleaving means of the transmitting device. The erasure means packetizes the decoding result after sorting by the deinterleaved means, and erases the signals in the packet in N patterns (multiple N) in ascending order of reliability in the packet. The decoding means performs RS decoding for each of the N-pattern packets obtained by the vanishing means. The syndrome calculation means calculates the syndrome for each of the N packets decoded by the RS decoding means, and determines whether the packet can be correctly decoded or the packet has an error remaining. The selection means selects a packet to be output as a decoding result from the N packets decoded by the RS decoding means based on the determination result by the syndrome calculation means.

According to such a configuration, while repeatedly decoding the received signal, the reliability of the log-likelihood ratio to the first bit (for example, the upper bit) and the log-likelihood ratio to the second bit (for example, the lower bit). The reliability can be gradually improved, and a large improvement effect can be obtained after a predetermined number of repetitions.

As the maximum a posteriori decoding means, for example, a BCJR decoding means that performs BCJR (Bahl-Cocke-Jelinek-Raviv) decoding processing can be used. Further, as another example, an SOVA decoding means that performs SOVA (Soft Output Vitterbi Algorithm) decoding processing can be used.

According to the present invention, it is possible to improve the required CNR of a wireless communication system that wirelessly transmits a signal from a transmitting device to a receiving device.

It is a figure which shows an example of the structure of a wireless communication system. It is a figure which shows the structural example of the receiver of the wireless communication system which concerns on one Embodiment of this invention. It is a figure which shows the CNR vs. bit error characteristic before RS decoding.

Hereinafter, an embodiment of the present invention will be described in detail with reference to the drawings.
First, the configuration of the receiver of the wireless communication system according to the embodiment of the present invention will be described in detail with reference to FIG. Since the transmitter side has the same configuration as that shown in FIG. 1, detailed description thereof will be omitted. As described with reference to FIG. 1, a signal that has undergone error correction coding and modulation processing on the transmitter side is transmitted to the propagation path 16 and reaches the receiver.

The signal received by the receiver is input to the high-order bit LLR calculation unit 34 and the coded bit LLR calculation unit 32.
Before explaining the error correction and decoding on the receiver side, the correspondence between the error correction code and the modulation / demodulation will be described. The coding bits that are the output of the convolutional coding unit 14 on the transmitter side are '0' and '1', but in modulation / demodulation and soft judgment error correction processing, bit '0' is +1 and bit '1' is-. It is replaced with 1.

In the coded bit LLR calculation unit 32, based on the received signal y _j , the coded LLR L _j ^{c (a), for the coded bit which is the output of the convolutional code unit 14 on the transmitting side shown in (Equation 1),} Calculate ^BCJR . Here, j indicates a sequence number.

The calculation method of L _j ^{c (a), BCJR} is to calculate L _j c (a) ^{, BCJR} from the signal after equalization after performing waveform equalization such as ZF (Zero Forcing) or MMSE (Minimum Mean Squared Error). There are a method of calculating, a method of calculating a received signal replica such as MLD (Maximum Likelihood Detection), and a method of calculating L _j ^{c (a) and BCJR} based on the Euclidean squared distance from the received signal. Since the present invention does not depend on the calculation method of _Lj ^{c (a), BCJR} , detailed description thereof will be omitted.
The coded LLR L _j ^{c (a), BCJR} obtained by the demodulator 21 is input to the BCJR decoding unit 33. The name BCJR is derived from the acronyms of the inventors, Bahl, Cocke, Jelinek, and Raviv.

ここで、ｙ_1～j-1 は、１～ｊ－１までの受信系列、ｙ_j+1～J は、ｊ＋１～Ｊまでの受信系列を示している。The maximum a posteriori decoding based on the BCJR algorithm will be described below. For a detailed explanation, refer to Non-Patent Document 2 and the like. In the trellis diagram corresponding to the convolutional coding unit 14, if the state of time j is S _j and the state of time j + 1 is S _{j + 1} , the input information bits u _j to the convolutional coding unit 14 are from S _j . It corresponds to the state transition to S _{j + 1} . Therefore, the posterior probability p (u _j | y) of the information bit with respect to the reception sequence y is (Equation 2).

Here, y _{1 to j-1} indicate a reception sequence from 1 to j-1, and y _{j + 1 to J} indicate a reception sequence from j + 1 to J.

The following substitution is performed for this (Equation 2).

Substituting (Equation 3) into (Equation 2) and rewriting it gives (Equation 4).

Here, α (S _j-1 ) is often referred to as a forward-looking metric, and β (S _j ) is often referred to as a backward metric. These α (S _j-1 ) and β (S _j ) are recursively expressed as in (Equation 5) and (Equation 6).

When the logarithm is taken with respect to (Equation 5) and (Equation 6), it becomes (Equation 7) and (Equation 8).

Further, in order to reduce the amount of calculation, the processing of Σ () in (Equation 5) and (Equation 6) is replaced with the maximum value selection max (), and (Equation 9) and (Equation 10) have large characteristics. No deterioration occurs.

Further, when the logarithm is taken with respect to the second equation of (Equation 3), it becomes (Equation 11).

Ｌ_j ^c(a),BCJR は、（式１）で示した符号化ビットに対する対数尤度比であり、Ｌ_j ^u(a,l),BCJR は、下位ビットに対する情報ビットｕ^l の事前情報である。Here, Γ (S _j-1 , S _j ) is expressed by (Equation 12).

L _j ^{c (a) and BCJR} are log-likelihood ratios to the coded bits shown in (Equation 1), and L _j ^{u (a, l) and BCJR} are prior information of the information bits u ^l to the lower bits. Is.

By the above processing, the BCJR decoding unit 33 performs an approximation processing similar to the maximum a posteriori decoding called Max-Log-MAP decoding according to the above BCJR algorithm, and the posterior information L _j ^{u (p, l)} of the information bit u ^l . ^{), BCJR} is output. The ex post facto information L _j ^{u (p, l), BCJR} indicates the reliability of the information bit u ^l which is the BCJR decoding result.

An object of the present invention is to improve the required CNR by iteratingly processing the results of the BCJR decoding unit 33 and the RS decoding unit 43 via the interleave 46. Since the knowledge of u ^l has not been obtained, the prior information L _j ^{u (a, l) and BCJR} are 0, and the Γ (S _j-1 , S _j ) corresponding to (Equation 12) is (Equation 14). Become.

外部情報Ｌ_j ^c(e),BCJR は、下位ビットに対する情報ビットｕ^l の事前情報Ｌ_j ^u(a,l),MLD として上位ビットＬＬＲ算出部３４に出力される。Further, the BCJR decoding unit 33 obtains the external information _Lj ^{c (e) and BCJR} for the encoded bit of the lower bit by (Equation 15).

The external information L _j ^{c (e) and BCJR} are output to the high-order bit LLR calculation unit 34 as prior information L _j ^{u (a, l) and MLD} of the information bit u ^l for the low-order bit.

In the upper bit LLR calculation unit 34, the received signal y _j , the advance information L _j ^{u (a, l), MLD} of the information bit u ^l for the lower bit from the BCJR decoding unit 33, and the upper order from the advance information generation unit 48. The post-information bit LLR L _j ^{u (p, h), MLD} for the upper bit is calculated based on the pre-information L _j ^{u (a, h), MLD} of the information bit u ^h for the bit. The calculation method of L _j ^{u (p, h), MLD} is a method of calculating from the signal after equalization after performing waveform equalization such as ZF, or a method of calculating a received signal replica like MLD and receiving signal. There is a method to calculate directly by comparing with. Hereinafter, the case where the MLD method is used will be described as an example, but the same applies to the case where the ZF method and the MMSE method are used.

ここで、インデックスｑは、受信候補点番号でｑ＝１，・・・，Ｑであり、受信候補点数をＱとしたとき、ｒ^ ＝［ｒ₁^ ，・・・，ｒ_Q^ ］である。ｂ_q は、受信候補点番号ｑの受信候補点を構成するビットのインデックスである。ｂ_q （ｒ_q^ ）は、受信候補点ｒ_q^ を構成する信号点ベクトルのｑ番目のビット（‘０’または‘１’）である。 In the high-order bit LLR calculation unit 34, the received signal y is according to the reception candidate point._jReceived signal replica y that is a replica of_jCalculate ^. And the received signal y_jAnd the received signal replica y_j^ And the estimation result of the received noise power σ²Based on ^ and bit LLR L_j ^{u (p, h), MLD}Is calculated and output. Bit LLR L_j ^{u (p, h), MLD}Each component L of_j ^{u (p), MLD}(B_{q q}) Is calculated by, for example, (Equation 16).

Here, the index q is a reception candidate point number q = 1, ..., Q, and when the number of reception candidate points is Q, r ^ = [r.₁^ ^ , ..., r_Q^ ^ ]. b_{q q}Is an index of the bits constituting the reception candidate point of the reception candidate point number q. b_{q q}(R_{q q}^) Is the reception candidate point r_{q q}It is the qth bit ('0' or '1') of the signal point vector constituting ^.

Here, using the approximate expression of (Equation 17), (Equation 16) becomes (Equation 18).

The LLR sorting unit 35 includes the high-order bit LLR L _j ^{u (p, h), MLD} output from the high-order bit LLR calculation unit 34 and the low-order bit LLR L _j ^{u (p, l} ) output from the BCJR decoding unit 33. ^{), BCJR} are processed to return to the original order after being rearranged in the predetermined order on the transmitting side, and the post-information LLR _Lj ^{u (p)} returned to the original order is output to the rigid determination unit 36. do.

The rigid determination device 36 outputs the information bit '0' when the sign of the posterior information LLR L _j ^{u (p)} is positive, and outputs the information bit '1' when the sign is negative. This is due to the correspondence between the error correction code and the modulation / demodulation described above.
The result of the hardness determination is input to the S / P converter 38. The S / P converter 38 parallel-converts the serial bits resulting from the rigid determination into the number of bits corresponding to the RS code (generally 1 byte = 8 bits).

これは、Ｌ_j ^u(p) ＝ｌｎ｛Ｐ（ｕ＝＋１）／Ｐ（ｕ＝－１）｝を逆変換して導出している。このＬＬＲ／確率変換器３７の出力Ｐ_u,j は、情報ビットｕに関する誤り確率である。Further, the post-information LLR L _j ^{u (p)} is also input to the LLR / probability conversion unit 37. The LLR / probability conversion unit 37 performs the operation shown in (Equation 19) using the post-information LLR _Lj ^{u (p)} .

This is derived by inversely transforming L _j ^{u (p)} = ln {P (u = + 1) / P (u = -1)}. The outputs _{Pu and j} of the LLR / probability converter 37 are error probabilities related to the information bit u.

In addition, in order to reduce the dynamic range of the arithmetic processing, the conversion shown in (Equation 20) may be used.

This information bit probability _{Pu, j} is input to the byte probability converter 39. The byte probability converter 39 converts the information bit probabilities P _{u and j} into byte probabilities P _{u_byte} according to (Equation 21).

Further, in order to reduce the dynamic range of the arithmetic processing, the product may be converted into a sum by using (Equation 22).

These byte probabilities P _{u_byte} indicate the reliability of the byte, and the smaller the value, the lower the reliability, and the higher the possibility that an error has occurred.
The byte-based information signal obtained by the S / P conversion 38 is input to the deinterleaved section 40-1, and the byte probability P _{u_byte} obtained by the byte probability converter 39 is input to the deinterleaved section 40-2. .. These deinterleaved units 40-1 and 40-2 rearrange the input in the reverse order of the interleaved unit 12, and return the input to the original order.

The deinterleaved result of the byte probability P _{u_byte} by the deinterleaved unit 40-2 is input to the ascending order sorting unit 41. The ascending order sorting unit 41 grasps the position in the RS packet processed by the RS decoder in ascending order of byte probability.

The disappearance unit 42 calculates the disappearance pattern for a plurality of subsequent RS decoding units 43, but before explaining the processing of the disappearance unit 42, a brief explanation of the disappearance decoding will be given.
The minimum distance d _min of the RS (204,188) code described above is d _min = 204-188 = 16.

ここで、ｅは、誤りの位置が予め分かっている場合の消失訂正バイト数を示しており、ｔは、誤り位置が不明な場合の誤り訂正バイト数を示している。As a feature of RS decoding, if it is (Equation 23), correct decoding is possible.

Here, e indicates the number of erasure correction bytes when the error position is known in advance, and t indicates the number of error correction bytes when the error position is unknown.

When the positions of all errors are unknown, t ≦ 8 is set in order to satisfy (Equation 23), and error correction is possible up to 8 bytes in the RS packet (204 bytes). If all the error positions are known, e ≦ 16 and up to 16 bytes can be corrected. By presenting the error position to the RS decoding unit in this way, the error correction capability is increased up to twice as compared with the case where the error position is not presented.

The vanishing unit 42 creates a plurality of vanishing patterns indicating erroneous positions based on the signal obtained from the ascending order rearranging unit 41 for the byte-based information signal after deinterleaving by the deinterleaving unit 40-1. Specifically, in the first vanishing pattern, the vanishing byte number e is set to 0, and there is no error position. In the second disappearance pattern, for example, the number of lost bytes e is set to 2, and the position of the lost bytes is a position where the byte probability P _{u_byte} having the smallest value among the results of the ascending order sorting unit 41 is set. In the third loss pattern, for example, the number of lost bytes e is 4, and 3 bytes are set as the positions of the lost bytes in ascending order sorting unit 41 in ascending order of byte probability P _{u_byte} . Similarly, after the fourth disappearance pattern, the number of lost bytes e is set to 6, 8, 10, 12, and 14, and the positions of the lost bytes are assigned in ascending order from the result of the sorting unit 41 in ascending order.

The disappearance patterns calculated by the disappearance unit 42 are RS decoding unit 43-1, RS decoding unit 43-2, RS decoding unit 43-3, RS decoding unit 43-4, ..., RS in order from the first disappearance pattern. It is input to the decoding unit 43-N. The RS decoding units 43-1 to 43-N perform RS decoding on the presented disappearance pattern.

The decoding results D _k (k = 1 to the maximum number of disappearance patterns) from the RS decoding units 43-1 to 43-N are input to the syndrome calculation units 44-1 to 44-N, respectively. The syndrome calculation unit 44 calculates the syndrome _Sk for each decoding result D _k . The syndrome is used for error detection by utilizing the fact that the code generated by the RS code unit is divided by the RS code generation polynomial and the division result becomes 0. That is, the syndrome _Sk is calculated from the RF decoding result, and if the syndrome _Sk is 0, it means that the decoding is correct, and if the syndrome _Sk is not 0, it is determined that an error remains. ..

Each syndrome _Sk is input to the selection unit 45. The selection unit 45 selects the decoding result D _k based on the syndrome _Sk as shown below.
The selection unit 45 outputs W = ω as the normal decoding result W of the packet when there is _k in which the syndrome Sk obtained by the syndrome calculation units 44-1 to 44-N has _Sk = 0. At the same time, the decoding result D _k at which _Sk = 0 is selected and output. Here, ω is a prior information coefficient, which is a predetermined coefficient larger than 0, and for example, ω = 0.8. If there is no decoding result with _Sk = 0, it is assumed that correct decoding could not be performed, and W = 0 is set.

Here, according to the definition of (Equation 1), if the bit constituting the decoding result D _k is '0', it is replaced with +1 and if it is '1', it is replaced with -1. Further, the result is multiplied by the normal decoding result W for the entire packet. That is, if there is no correctly decoded result, it is replaced with 0.

Specific examples of these processes will be described.
Assuming that the result D _k that can be correctly decoded is {"10110010", "00111100", ...}, When ω = 0.8, the replacement result DW is {"-0. 8 +0.8-0.8-0.8 +0.8 +0.8-0.8 +0.8 "," +0.8 +0.8-0.8-0.8-0.8-0.8 " +0.8 +0.8 ", ...}. If the correct decoding result does not exist, all the packets are 0 {"0 0 0 0 0 0 0 0", "0 0 0 0 0 0 0 0", ...}.

This result is input to the interleave unit 46. The interleave unit 46 performs the same order rearrangement as the interleave unit 23 on the transmitter side. The interleaved result is converted from a parallel signal in byte units to a serial signal in bit units via the P / S converter 47, and is input to the advance information generation unit 48.

The prior information generation unit 48 distributes the input RS decoding enable / disable information into information for the upper bit and information for the lower bit according to the modulation method information. The information on whether or not RS decoding is possible for the lower bits is input to the BCJR decoding unit 33 as the prior information bits LLR L _j ^{u (a, l), BCJR} . The RS decoding enable / disable information for the high-order bit is input to the high-order bit LLR calculation unit as the prior information bit LLR L _j ^{u (a, h), MLD} .

As described in (Equation 12), Γ (S _j-1 , S _j ) is the sum of the coding bits LLR L _j ^{c (a), BCJR} and the prior information bits LLR L _j ^{u (a, l), BCJR} . Information on whether or not RS decoding is possible is input to the BCJR decoding unit 33 as advance information bits LLR L _j ^{u (a, l) and BCJR} . In these prior information bits LLR, a set signal for each packet is distributed by the interleaving unit 46. That is, the prior information 0 when the decoding is not correct and the prior information ± ω when the decoding is correct are dispersed and input to the BCJR decoding unit 33.

These processes are shown by (Equation 24). The signal obtained by multiplying the RS decoding result D and the normal decoding result W is sorted by the interleave function Π [], and then parallel serial conversion is performed by {} _{P → S} , and the prior information bit LLR L _j ^{u (a, l) Add as BCJR} .

In BCJR decoding, even if an erroneous series estimation is performed in the initial processing, it is correct by inputting the prior information bits LLR _Lj ^{u (a, l), BCJR} as in (Equation 24). Series estimation can be performed, and the bit error rate can be reduced.

By repeating the above-described processing a plurality of times, each time the BCJR decoding unit 33, the high-order bit LLR calculation unit 34, and the RS decoding units 43-1 to 43-N repeat the iterative processing, their respective error correction capabilities are gradually increased. By improving, a large improvement effect can be obtained after a predetermined number of repetitions.

FIG. 3 shows the configuration before RS decoding when BCJR decoding and RS (204,188) decoding for a convolutional code having a constraint length of 7 and a coding rate of 1/2 are repeatedly processed in the configuration disclosed in Non-Patent Document 1. It is a figure which showed the CNR vs. bit error rate characteristic. 32QAM is used as the modulation method. As described above, the bit error rate before RS decoding is 2 × 10 ^-4 , which is the bit error rate that becomes error-free in the subsequent RS decoding. As can be seen from this result, it is possible to realize an improvement of the required CNR of about 7 dB by performing the iterative process.

As described above, in this example, the transmitter and the receiver operate as follows.
In the transmitter, the RS coding unit 11 packetizes the signal to be transmitted in units of a predetermined length, and RS-encodes the packet. The interleaving unit 12 rearranges the order after RS coding by the RS coding unit 11 to acquire an information bit sequence. The S / P conversion unit 13 distributes the information bit sequence obtained by the interleaving unit 12 into upper bits (u ^h ) or lower bits ( ^ul ). The convolutional coding unit 14 performs convolutional coding on the lower bits distributed by the S / P conversion unit 13. The modulation unit 15 performs modulation processing on the high-order bits distributed by the S / P conversion unit 13 and the coded bits obtained by the convolutional coding unit 14.

In the receiver, the coded bit LLR calculation unit 32 demodulates the received signal (y _j ) from the transmitting device, and the bit log-likelihood ratio to the coded bit (L _j ^{c (a), BCJR} ). Is calculated. The BCJR decoding unit 33 determines the bit log likelihood ratio (L _j ^{c (a), BCJR} ) to the coded bit calculated by the coded bit LLR calculation unit 32, and prior information (L _j ^u ) to the information bits of the lower bits. Based on ^{(a, l), BCJR} ; first prior information), the information bits of the lower bits are decoded so that the posterior probability is maximized, and the logarithmic likelihood ratio to the lower bits (L _j ^{u (p))} . ^{, L), BCJR} ) and the prior information (L _j ^{u (a, l), MLD} ; second prior information) for the information bits of the lower bits are calculated. The high-order bit LLR calculation unit 34 receives a received signal (y _j ), prior information (L _j ^{u (a, l), MLD} ; second prior information) calculated by the BCJR decoding unit 33, and the information bit of the high-order bit. The logarithmic likelihood ratio (L _j ^{u (p, h), MLD} ) to the high-order bits is calculated based on the prior information (L _j ^{u (a, h), MLD} ; third prior information).

After that, the receiver receives the logarithmic likelihood ratio (L _j ^{u (p, h), MLD} ) to the upper bits calculated by the upper bit LLR calculation unit 34 and the logarithmic likelihood to the lower bits calculated by the BCJR decoding unit 33. Decoding processing is performed to decode the signal to be transmitted based on the degree ratio (L _j ^{u (p, l), BCJR} ). In this decoding process, the LLR sorting unit 35 is calculated by the BCJR decoding unit 33 and the logarithmic likelihood ratio (L _j ^{u (p, h), MLD} ) to the high-order bits calculated by the high-order bit LLR calculation unit 34. Sort the order of the log-like likelihood ratio (L _j ^{u (p, l), BCJR} ) to the lower bits. The hardness determination unit 36 determines the log-likelihood ratio after sorting by the LLR sorting unit 35, and calculates the decoding result. The LLR / probability conversion unit 37 converts the log-likelihood ratio after sorting by the LLR sorting unit 35 into reliability. The deinterleaved units 40-1 and 40-2 rearrange the decoding results by the rigid determination unit 36 and the conversion results by the LLR / probability conversion unit 37 in the reverse order of the interleaving unit 12 of the transmitter. The vanishing unit 42 packetizes the decoding result after sorting by the deinterleaved unit 40-1, and erases the signals in the packet in N patterns (multiple N) in ascending order of reliability in the packet. The RS decoding unit 43-1-N performs RS decoding for each of the N pattern packets obtained by the vanishing unit 42. The syndrome calculation units 44-1 to 44-N calculated the syndrome for each of the N packets decoded by the RS decoding unit 43-N, and the packets were correctly decoded or the packets with residual errors. Judge if it is. Based on the determination result by the syndrome calculation units 44-1 to 44-N, the selection unit 45 outputs the packet to be output as the decoding result from the N packets decoded by the RS decoding units 43-1 to 43-N. select.

Further, the receiver receives prior information (L _j ^{u (a, l), BCJR} ; first prior information) for the information bits of the lower bits based on the decoding result information (D / W) output from the selection unit 45. ) Is generated and fed back to the BCJR decoding unit 33, and advance information (L _j ^{u (a, h), MLD} ; third prior information) for the information bit of the upper bit is generated and sent to the upper bit LLR calculation unit 34. Prior information to give feedback Perform feedback processing. In this prior information feedback process, the interleaving unit 46 rearranges the decoding result information (D / W) output from the selection unit 45 in the same order as the interleaving unit 12 of the transmitter. The prior information generation unit 48 distributes the information sorted by the interleaving unit 46 into the information for the upper bit or the information for the lower bit, and the information for the upper bit is the prior information for the information bit of the upper bit (L _j ^{u (a, h).} ) ^{, MLD} _; It feeds back to the BCJR decoding unit 33.

With such a configuration, BCJR decoding and RS decoding are repeatedly repeated in the receiver. As a result, each error correction capability can be gradually improved, so that the reliability of the log-likelihood ratio for the upper bits and the reliability of the log-likelihood ratio for the lower bits can be gradually improved, and the predetermined number of iterations can be reduced. After that, a great improvement effect can be obtained. Therefore, it is possible to provide a wireless communication system that can improve the required CNR.

Here, in the wireless communication system of this example, the transmitter corresponds to the transmitting device according to the present invention, and the receiver corresponds to the receiving device according to the present invention. It may be a wireless communication system of another form, such as a configuration in which wireless communication devices including a transmitter and a receiver communicate wirelessly with each other.

Further, in the wireless communication system of this example, the BCJR decoding unit 33 that performs the BCJR decoding process is used as the maximum a posteriori decoding means according to the present invention, but the present invention is not limited to this. For example, instead of the BCJR decoding unit 33, an SOVA decoder that performs SOVA decoding processing may be used, and decoding processing that is close to maximum a posteriori decoding may be performed.

Further, in the wireless communication system of this example, the upper bit corresponds to the first bit according to the present invention, and the lower bit corresponds to the second bit according to the present invention, but the present invention is not limited thereto. For example, the above allocation may be reversed (that is, the lower bits may be the first bit according to the present invention and the upper bits may be the second bit according to the present invention).

Although the present invention has been described above based on one embodiment, it is needless to say that the present invention is not limited to the wireless communication system described here and can be widely applied to other wireless communication systems.
Further, the present invention can be provided, for example, as a method or method for executing the process according to the present invention, a program for realizing such a method or method, a storage medium for storing the program, or the like.

This application claims the benefit of priority on the basis of Japanese application Japanese Patent Application No. 2019-195303 filed on October 28, 2019, the entire disclosure of which is incorporated herein by reference.

The present invention can be used in a wireless communication system that wirelessly transmits a signal from a transmitting device to a receiving device.

11: RS encoder, 12: interleave part, 13: S / P conversion part, 14: convolutional coding part, 15: modulation part, 16: propagation path, 17: demodulation part, 18: Viterbi decoding part, 19: P / S conversion unit, 20: deinterleaved unit, 21: RS decoding unit, 32: coded bit LLR calculation unit, 33: BCJR decoding unit, 34: upper bit LLR calculation unit, 35: LLR sorting unit, 36: hard Judgment unit, 37: LLR / probability conversion unit, 38: S / P conversion unit, 39: byte probability conversion unit, 40-1, 40-2: deinterleaved unit, 41: ascending order sorting unit, 42: disappearance unit, 43-1 to 43-N: RS decoding unit, 44-1 to 44-N: Syndrome calculation unit, 45: Selection unit, 46: Interleave unit, 47: P / S conversion unit, 48: Preliminary information generation unit

Claims (5)

In a wireless communication system that wirelessly transmits a signal from a transmitter to a receiver
The transmitter is
A convolutional coding means that performs convolutional coding on the second bit other than the predetermined first bit among the information bit series obtained by RS coding for the signal to be transmitted.
A modulation means for performing a modulation process on the first bit and the coding bit obtained by the convolutional coding means is provided.
The receiving device is
A first demodulation means that performs demodulation processing on a received signal from the transmission device and calculates a bit log-likelihood ratio with respect to the coded bit.
The second bit so as to maximize the posterior probability based on the bit logarithmic likelihood ratio to the coded bit calculated by the first demodulatoring means and the first prior information to the information bit of the second bit. The maximum post-probability decoding means for decoding the information bit of the above and calculating the logarithmic likelihood ratio to the second bit and the second prior information for the information bit of the second bit.
The log-likelihood ratio for the first bit is calculated based on the received signal, the second prior information calculated by the maximum a posteriori decoding means, and the third prior information for the information bit of the first bit. Second demographic means to do
The signal to be transmitted is set based on the log-likelihood ratio to the first bit calculated by the second demodulation means and the log-likelihood ratio to the second bit calculated by the maximum a posteriori decoding means. Decoding means to decode,
Based on the information of the decoding result output from the decoding means, the first prior information is generated and fed back to the maximum a posteriori decoding means, and the third prior information is generated to the second demographic means. A wireless communication system characterized by having a prior information feedback means for feeding back. In the wireless communication system according to claim 1,
The transmitter is
An RS coding means for packetizing the signal to be transmitted in units of a predetermined length and RS-coding the packet.
Further provided with a first interleaving means for reordering the order after RS coding by the RS coding means to acquire the information bit sequence.
The prior information feedback means of the transmitter is
A second interleaving means that sorts the information of the decoding result output from the decoding means in the same order as the first interleaving means of the transmitting device.
The information after sorting by the second interleaving means is distributed to the information for the first bit or the information for the second bit, and the information for the first bit is fed back to the second demodizing means as the third information. A wireless communication system comprising a pre-information generation means for feeding back information for a second bit to the maximum post-probability decoding means as the first information. In the wireless communication system according to claim 2,
The decoding means of the receiving device is
Log-likelihood ratio rearrangement that rearranges the order of the log-likelihood ratio to the first bit calculated by the second demodulation means and the log-likelihood ratio to the second bit calculated by the maximum a posteriori decoding means. Means and
The log-likelihood ratio sorting means hard-determines the log-likelihood ratio after sorting, and the hard-determining means for calculating the decoding result.
A conversion means for converting the log-likelihood ratio after sorting by the log-likelihood ratio sorting means into reliability, and a conversion means.
A deinterleaving means that rearranges the decoding result by the rigid determination means and the conversion result by the conversion means in the reverse order of the first interleaving means of the transmission device.
A vanishing means that packetizes the decoding result after sorting by the deinterleaving means and eliminates the signals in the packet in N patterns (multiple N) in ascending order of reliability in the packet.
An RS decoding means that performs RS decoding for each of the N pattern packets obtained by the disappearing means, and an RS decoding means.
A syndrome calculation means that calculates a syndrome for each of the N packets decoded by the RS decoding means and determines whether the packet can be correctly decoded or a packet in which an error remains.
A wireless communication system comprising: a selection means for selecting a packet to be output as a decoding result from among N packets decoded by the RS decoding means based on a determination result by the syndrome calculation means. In the wireless communication system according to claim 1,
The maximum a posteriori decoding means is a wireless communication system characterized by being a BCJR decoding means that performs BCJR decoding processing or an SOVA decoding means that performs SOVA decoding processing. In a wireless communication method in which a signal is transmitted wirelessly from a transmitting device to a receiving device.
The transmitter is
Of the information bit series obtained by RS coding for the signal to be transmitted, the second bit excluding the predetermined first bit is convolutionally coded.
Modulation processing is performed on the first bit and the coded bit obtained by the convolutional coding.
The receiving device
Demodulation processing is performed on the received signal from the transmitting device, and the bit log-likelihood ratio to the coding bit is calculated.
Decoding the information bit of the second bit so as to maximize the posterior probability based on the calculated bit log likelihood ratio to the coded bit and the first prior information to the information bit of the second bit. And calculated the logarithmic likelihood ratio to the second bit and the second prior information to the information bit of the second bit.
The log-likelihood ratio with respect to the first bit is calculated based on the received signal, the calculated second prior information, and the third prior information with respect to the information bit of the first bit.
The signal to be transmitted is decoded based on the log-likelihood ratio to the first bit and the calculated log-likelihood ratio to the second bit.
Based on the information indicating the result of the decoding, the 1 prior information is generated and fed back for decoding the information bit of the 2nd bit, and the 3rd prior information is generated and the logarithmic probability with respect to the 1st bit is generated. A wireless communication method characterized by feeding back for calculation of a degree ratio.

Images