JP2008160169A - Communication apparatus, turbo encoder, and communication method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To obtain a communication apparatus capable of performing faster processing than before and improving error rate characteristics after error correction. <P>SOLUTION: The communication apparatus according to the present invention employs a turbo encoder having an encoder 101 which encodes an information bit sequence and an encoder 104 which encodes the information bit sequence after interleaving processing. For example, the turbo encoder has an interleaver 102 which stores the information bit sequence in an M×N input buffer; prepares N-1 different random sequences as many as rows to generate M kinds of random sequences; maps a minimum to Nth rows of all the random sequences to generate M×N mapping patterns; maps an information bit sequence of interleaving length to the M×N mapping patterns; and reads the information bit sequence after the mapping, column by column, and outputs it to the encoder 103. <P>COPYRIGHT: (C)2008,JPO&INPIT

Description

  The present invention relates to a communication apparatus using an error correction code such as a turbo code.

  Hereinafter, a conventional communication apparatus will be described. For example, in wideband CDMA (W-CDMA: Code Division Multiple Access) using an SS (Spread Spectrum) system, a turbo code is proposed as an error correction code that greatly exceeds the performance of a convolutional code. This turbo code encodes a sequence obtained by interleaving an information bit sequence in parallel with a known encoded sequence, and is said to have characteristics close to the Shannon limit. One. In the W-CDMA, the performance of the error correction code greatly affects the transmission characteristics in voice transmission and data transmission. Therefore, the transmission characteristics are greatly improved by applying the turbo code.

  Here, operations of the transmission system and the reception system of the conventional communication apparatus using the turbo code will be specifically described. FIG. 3 (FIGS. 3-1 and 3-2) is a diagram showing a configuration of a turbo encoder used in the transmission system. 3A, reference numeral 101 denotes a first recursive systematic convolutional encoder that convolutionally encodes an information bit sequence and outputs redundant bits, 102 is an interleaver, and 103 is a result of rearrangement by the interleaver 102. This is a second recursive systematic convolutional encoder that convolutionally encodes the information bit sequence of the information and outputs redundant bits. FIG. 3-2 is a diagram illustrating an internal configuration of the first recursive systematic convolutional encoder 101 and the second recursive systematic convolutional encoder 103, and shows two recursive systematic convolutional encoders. Is an encoder that outputs only redundant bits. In addition, the interleaver 102 used in the turbo encoder performs processing for rearranging information bit sequences at random.

In the turbo encoder configured as described above, at the same time, the information bit sequence: x ₁ and the redundant bit sequence obtained by encoding the information bit sequence: x ₁ by the processing of the first recursive systematic convolutional encoder 101. : X ₂ and redundant bit sequence: x _{3 obtained} by encoding the information bit sequence after the interleaving process by the processing of the second recursive systematic convolutional encoder 103.

On the other hand, FIG. 4 is a diagram showing a configuration of a turbo decoder used in the reception system. In FIG. 4, 111 is a first decoder for calculating a log likelihood ratio from received signal: y ₁ and received signal: y ₂ , 112 and 116 are adders, and 113 and 114 are interleavers. , 115 is a second decoder for calculating a log-likelihood ratio from the received signal: y ₁ and the received signal: y ₃ , 117 is a deinterleaver, and 118 determines the output of the second decoder 115. And a determinator that outputs an estimated value of the original information bit sequence. The received signals: y ₁ , y ₂ , and y ₃ are signals that are affected by transmission line noise and fading on the information bit series: x ₁ and redundant bit series: x ₂ , x ₃ , respectively.

In the turbo decoder configured as described above, first, the first decoder 111 performs log likelihood ratio of estimated information bits: x _1k estimated from the received signal: y _1k and the received signal: y _2k : L (x _1k ) is calculated (k represents time). Here, the probability that the information bit: x _1k is 1 with respect to the probability that the information bit: x _1k is 0 is obtained. Note that Le (x _1k ), which will be described later, represents external information, and La (x _1k ) represents prior information that is the previous external information.

Next, the adder 112 calculates external information for the second decoder 115 from the log likelihood ratio as the calculation result. In the first decoding, since prior information is not obtained, Le (x _1k ) = 0.

Next, the interleavers 113 and 114 rearrange the signals so that the received signal: y _1k and the external information: Le (x _1k ) match the time of the received signal: y ₃ . Thereafter, in the second decoder 115, similarly to the first decoder 111, the received signal: y ₁ and the received signal: y ₃ and the previously calculated external information: La (x _1k ). The log likelihood ratio: L (x _1k ′) is calculated. Then, the adder 116 calculates external information: Le (x _1k ). At this time, the external information rearranged by the deinterleaver 117 is fed back to the first decoder 111 as prior information: Ln (x _1k ).

Finally, in this turbo decoder, the above processing is repeatedly performed for a predetermined number of times to calculate a higher-precision log likelihood ratio, and the determiner 118 is based on this log likelihood ratio. A determination is made to estimate the original information bit sequence. Specifically, for example, if the log likelihood ratio is “L (x _1k ′)> 0”, the estimated information bit: x _1k ′ is determined to be 1, and “L (x _1k ′) ≦ 0”. If so, the estimated information bit: x _1k ′ is determined to be 0.

  For example, in the W-CDMA standard, a prime interleaver (hereinafter referred to as PIL) is used as an interleaver, and has a very high-performance error correction capability as an internal interleaver of a turbo code.

  In addition, since a turbo code performs iterative decoding, a decoding delay time is large, and in a system that requires a processing speed exceeding 100 Mbps, parallel decoding processing is required. For decoding parallelization, when reading the data stored in the memory bank before and after interleaving, it is necessary not to access the same memory bank at the time of writing, but with the above-mentioned PIL, the circuit becomes complicated when mounting, etc. There are problems, and interleavers satisfying such conditions include block interleavers and Latin square / Latin rectangular interleavers (Latin Lattice / Rectangle Interleaver: hereinafter referred to as LRI (see Patent Documents 1 and 2 below)). The method for generating the LRI interleaver is shown below.

<LRI interleaver configuration method>
<Step 1>
(1) Set parameters.
K: Number of input information bits
P: Minimum prime number that satisfies the following: P ² ≧ K
N = P: Number of columns of interleave matrix
M = ceil (K / N): Number of rows of interleave matrix
(Ceil (x) is the smallest integer that exceeds x)
(2) Select the primitive source of P.
G ₀ : A primitive root

<Step 2>
Generate a basic pseudo-random number sequence.
C (0) = 1
For i = 1,…, P-3 {C (i + 1) = G ₀ × C (i) mod P}
Regarding this generation,
For i = 1,…, P-2 {C (i) = (G ₀ ) ⁱ mod P}
The same is true.

<Step 3>
The sequence C (•) is shifted one by one to form a matrix CL _j (i). However, the rightmost column is 0.
For j = 0,1,…, M-1 {
For i = 0,1, ..., N-2 {CL _j (i) = C (j + i mod N-1)}
CL _j (N-1) = 0
}

<Step 4>
An M × N input matrix U _j (i) is prepared.
For j = 0,1,…, M-1 {
For i = 0,1,…, N-1 {U _j (i) = i + N × j}
}

<Step 5>
An M × N output matrix U ′ _j (i) is prepared using CL _j (i).
For j = 0,1,…, M-1 {
For i = 0,1,…, N-1 {U ' _j (i) = U _(M-1-j) (CL _(M-1-j) (i))}
}

<Step 6>
Generated by reading the output sequence u (k) in column order.
For j = 0,1,…, M-1 {
For i = 0,1, ..., N-1 {u (j + M × i) = U ' _j (i)}
}

<Step 7>
Pruning If there is a difference between the size of the prepared matrix and the information length, the output value (u (k) ≧ K) having a size exceeding the information length is deleted.
An example of an interleave pattern generated by LRI is shown below. This is the case for K = 289, P = 17, M = 17, N = 17, and G ₀ = 3. The interleave pattern is generated by reading the following equation (1): U ′ _j (i) in the column direction.

Interleave pattern: u (i) = {273,261,240,..., 20,1,275,256,..., 51,34,17,0}. In the decoding process of this example, the received signal y _1k and the external information Le (x _1k ) are stored in another memory bank prepared as many as the number of rows (k before interleaving is 0, 1, 2, ..., K, K after interleaving is u (i)). Before interleaving, {0,1,2, ...}, {17,18,19, ...}, {34,35,36, ...} ... are processed at the same timing for each {} and interleaved. Later, {1,20,43, ...}, {26,44, ..., 3}, {47, ..., 9,47} ... The same memory bank is not accessed at the timing, and parallel processing can be performed without contradiction.

JP 2001-332981 A JP 2001-285077 A

  However, the block interleaver, which is an interleaver that enables parallel processing, has a very simple configuration, but has a large performance deterioration compared to the current PIL. LRI has the same or better performance than PIL when the interleaver size is small, but degrades when it becomes large.

  The present invention has been made in view of the above, and can be applied to all communication systems adopting a turbo code. Further, compared with the prior art, high-speed processing by parallelization is possible, and after error correction. An object of the present invention is to obtain a communication device that can improve the error rate characteristics.

  In order to solve the above-described problems and achieve the object, a communication apparatus according to the present invention includes a first encoder that performs convolutional coding or block coding of an information bit sequence and outputs first redundant data. A turbo encoder comprising: a second encoder that performs convolutional coding or block coding on the information bit sequence after the interleaving process and outputs second redundant data; The turbo encoder stores the information bit sequence in an M × N input buffer, prepares N−1 random sequences different from each other by the number of rows, generates M types of random sequences, and A minimum value is mapped to the Nth row of each row in all random sequences, an M × N mapping pattern is generated, and an interleave length information bit sequence is mapped to the M × N mapping pattern. Grayed reads information bit sequence after the mapping in units of columns, characterized in that it comprises an interleaver, to be output to the second encoder.

  According to the present invention, a parameter related to interleaver generation is not uniquely determined with respect to the input information length K as in the prior art, and a high interleave pattern is searched for a parallel number that satisfies a required processing speed. There is an effect that an interleave pattern having error correction capability can be obtained.

  Embodiments of a communication apparatus according to the present invention will be described below in detail with reference to the drawings. Note that the present invention is not limited to the embodiments.

Embodiment 1 FIG.
FIG. 1 is a configuration diagram illustrating a configuration example of a communication system according to Embodiment 1 including a turbo encoder 1 and a turbo decoder 11. In FIG. 1, a communication device on the transmission side (referred to as a transmission device) includes a turbo encoder 1, an interleaver generator 2, and a modulator 3. The communication device on the reception side (referred to as a reception device) And an interleaver generator 2.

  Here, the flow of the encoding process and the decoding process in the communication system employing the turbo code will be briefly described.

The turbo encoder 1 in the transmitter receives a message (u ₀ , u ₁ ,..., U _K-1 ) having an information length K and is encoded using the interleave pattern generated by the interleaver generator 2. v is generated. Then, in the modulator 2 in the transmission apparatus, the codeword v generated by the turbo encoder 1 is digitally modulated by a predetermined modulation method such as BPSK, QPSK, multilevel QAM, and the modulated signal x = (x ₀ , x ₁ ,..., x _N−1 ) are transmitted to the receiving device via the communication path 6.

On the other hand, in the receiving apparatus, the demodulator 4 performs the above-described BPSK, QPSK, multi-level QAM, etc. on the modulated signal y = (y ₀ , y ₁ ,..., Y _N−1 ) received via the communication path 6. Further, the turbo decoder 5 in the receiving apparatus performs decoding using the interleave pattern generated by the interleaver generator 2 on the demodulation result, and the decoding result (original Message u ₀ , u ₁ ,..., U _K-1 ).

  Note that a plurality of communication devices including both the transmission device and the reception device may be configured, and signals may be transmitted and received between the plurality of communication devices.

  Further, as shown in FIG. 3A, the turbo encoder 1 performs using a first recursive systematic convolutional encoder 101, an interleaver 102, and a second recursive systematic convolutional encoder 103. . FIG. 3-2 is an example of the recursive systematic convolutional encoder described above. For example, a recursive systematic convolutional encoder constituting a turbo code such as a recursive systematic convolutional encoder of the W-CDMA standard is used. Any container may be used. The turbo decoder 5 is configured corresponding to the recursive systematic convolutional encoder.

  Next, an interleave pattern generation algorithm in the interleaver generator 2 of the present embodiment is shown.

<Step 1>
(1) Set parameters.
K: Number of input information bits
N: Number of interleaved matrix columns (prime number)
M = ceil (K / N): Number of rows of interleave matrix
(Ceil (x) is the smallest integer that exceeds x)
s: Start position shift number at the time of reading out the basic pseudo random number sequence (2) Select a source element of N.
G ₀ : Primitive _{element of} N

<Step 2>
A basic pseudo-random number sequence is generated as N random sequences as follows.
C (0) = 1, C (N-1) = 0
For i = 0, ..., N-3 {C (i + 1) = G ₀ × C (i) mod N}
Regarding this generation,
For i = 0, ..., N-2 {C (i) = (G ₀ ) ⁱ mod N}
C (N-1) = 0
The same is true.

<Step 3>
The sequence C (•) is shifted s at a time to form a matrix CL _j (i).
For j = 0,1,…, M-1 {
For i = 0,1, ..., N-1 {CL _j (i) = C (i + s × j mod N)}
}

<Step 4>
An M × N input matrix U _j (i) is prepared.
For j = 0,1,…, M-1 {
For i = 0,1,…, N-1 {U _j (i) = i + N × j}
}

<Step 5>
An M × N output matrix U ′ _j (i) is prepared using CL _j (i).
For j = 0,1,…, M-1 {
For i = 0,1,…, N-1 {U ' _j (i) = U _(M-1-j) (CL _(M-1-j) (i))}
}

<Step 6>
Generated by reading the output sequence u (k) in column order.
For j = 0,1,…, M-1 {
For i = 0,1, ..., N-1 {u (j + M × i) = U ' _j (i)}
}

<Step 7>
Pruning If there is a difference between the size of the prepared matrix and the information length, the output value (u (k) ≧ K) having a size exceeding the information length is deleted.

  In steps 2 and 3, unlike the conventional LRI, the rightmost column is not fixed to 0, but the rightmost column may be fixed to 0 as described below. In this case, a basic pseudo-random number sequence is generated as follows as N−1 random sequences, and the Nth row in each row is fixed to 0.

<Step 2>
A basic pseudo-random number sequence is generated as N-1 random sequences as follows.
C (0) = 1
For i = 0, ..., N-3 {C (i + 1) = G ₀ × C (i) mod N}

<Step 3>
The sequence C (•) is shifted s at a time to form a matrix CL _j (i).
For j = 0,1,…, M-1 {
For i = 0,1, ..., N-2 {CL _j (i) = C (i + s × j mod N-1)}
CL _j (N-1) = 0
}

In step 1, N is selected so that N + 1 is a prime number. In step 2,
<Step 2>
A basic pseudo-random number sequence is generated as N random sequences as follows.
C (0) = 0
For i = 0, ..., N-3 {C (i + 1) = (G ₀ × (C (i) +1) mod N) -1}
Regarding this generation,
For i = 0, ..., N-2 {C (i) = ((G ₀ ) ⁱ mod N) -1}
The same is true.

In addition, when the primitive element G ₀ is 2, a multiplication by two is performed to obtain a basic pseudo-random number sequence. However, since a multiplication of 2 is possible by a 1-bit shift, a prime number with a primitive element of 2 is used. It is also possible to reduce the processing amount by limiting N only to this.

  FIG. 2 (FIGS. 2-1, 2-2, and 2-3) shows a performance comparison between the interleaver of the present embodiment and the conventional LRI and PIL of the W-CDMA standard. The simulation assumes an AWGN channel and BPSK modulation. For example, the turbo code recursive systematic convolutional encoder is the same as the W-CDMA standard, and the tail is a coding rate with 12 termination bits added. 1/3, and the decoding method is Max-Log-MAP decoding and the number of repetitions is 8. The horizontal axis represents the block error rate (Block Error Rate (BLER): Eb / N0 [dB]). Further, FIG. 2-1 shows a case of length 3000, FIG. 2-2 shows a length of 4000, and FIG. 2-3 shows a length of 5000. In configuring each interleaver, the combination of (m, s) is “ Search using at least one of “union bound” and computer simulation. (m, s) = (79,2) in the case of 3000, (101,1) in the case of 4000, and 5000 (157,3). In all cases, the BLER is improved in the present embodiment, and the same performance as the PIL employed in the current 3GPP having high error correction capability is shown. The set value of (m, s) is not limited to the above, and a combination having a high error correction capability is searched according to K.

Also, from (m, s) in each case, the number of rows is ceil (3000/101) = 30, ceil (4000/127) = 32, ceil (5000/157) = 32, respectively 30, 32, 32 parallelizations are possible. In the conventional LRI, since the number of rows is determined as P ² ≧ K (N = P), further parallelization is possible, but it is sufficient if a parallel number that satisfies the required processing speed of the communication system can be secured.

  As described above, according to the present embodiment, the performance is improved based on the number of rows or columns of an LRI interleaver having a Latin square or Latin rectangular structure. This enables not only a high parallel number as in the conventional LRI interleaver, but also an error correction capability equivalent to the 3GPP standard PIL having high performance. In addition, since the basic pseudo-random number generation method is the same as the 3GPP standard PIL, it is possible to improve the efficiency of interleaver development by effectively using conventional development assets.

Embodiment 2. FIG.
Next, the communication system according to the second embodiment will be described. In the system of the present embodiment, the configurations of the transmission device and the reception device are the same as those in FIG. 1 of the first embodiment described above. Hereinafter, processing different from the first embodiment will be described.

  Here, an interleave pattern generation algorithm in the interleaver generator 2 according to the second embodiment will be described.

<Step 1>
(1) Set parameters.
K: Number of input information bits
N: Number of columns in the interleave matrix
M = ceil (K / N): number of rows of interleave matrix, ceil (x) is the smallest integer exceeding x
s: Start position shift number when reading the basic pseudo-random number sequence (2) Select an integer relatively prime to N.
P ': N and a prime integer (co-prime, relative prime)

<Step 2>
A basic pseudo-random number sequence is generated as N random sequences as follows.
For i = 0, ..., N-1 {C (i) = (P '× i mod N)}
Regarding this generation,
C (0) = 0
For i = 1, ..., N-1 {C (i + 1) = C (i) + P 'mod N}
The same is true.

<Step 3>
The sequence C (•) is shifted one by one to form a matrix CL _j (i).
For j = 0,1,…, M-1 {
For i = 0,1, ..., N-1 {CL _j (i) = C (i + s × j mod N)}
}

<Step 4>
An M × N input matrix U _j (i) is prepared.
For j = 0,1,…, M-1 {
For i = 0,1,…, N-1 {U _j (i) = i + N × j}
}

<Step 5>
An M × N output matrix U ′ _j (i) is prepared using CL _j (i).
For j = 0,1,…, M-1 {
For i = 0,1,…, N-1 {U ' _j (i) = U _(M-1-j) (CL _(M-1-j) (i))}
}

<Step 6>
Generated by reading the output sequence u (k) in column order.
For j = 0,1,…, M-1 {
For i = 0,1, ..., N-1 {u (j + M × i) = U ' _j (i)}
}

<Step 7>
Pruning If there is a difference between the size of the prepared matrix and the information length, the output value (u (k) ≧ K) having a size exceeding the information length is deleted.

In step 2, the following may be performed.
<Step 2>
A basic pseudo-random number sequence is generated as N random sequences as follows.
For i = 0, ..., N-1 {C (i) = P '× i + Q mod N}
Here, Q is an arbitrary integer.

Or
<Step 2>
A basic pseudo-random number sequence is generated as N random sequences as follows.
For i = 0,…, N-1 {
C (i) = (floor (i / 2) x P 'mod N / 2) x 2 + 1 (when i is an even number)
C (i) = (floor (i / 2 + Q ') x P' mod N / 2) x 2 (when i is an odd number)
}
Here, floor (x) is a maximum integer not exceeding x, and Q ′ is an arbitrary integer.

Further, as N−1 random sequences, N−1 and a relatively prime integer P ″ are used to generate a basic pseudorandom number sequence as follows:
<Step 2>
For i = 0, ..., N-2 {C (i) = P '' × i mod N-1}
Or
For i = 0, ..., N-2 {C (i) = P '' × i + Q mod N-1}
<Step 3>
The sequence C (•) is shifted s at a time to form a matrix CL _j (i).
For j = 0,1,…, M-1 {
For i = 0,1, ..., N-2 CL _j (i) {CL _j (i) = C (i + s × j mod N)}
CL _j (N-1) = 0
}
And the rightmost column may be fixed at 0.

  In the present embodiment, the basic pseudo-random number sequence is generated based on the number of columns and a relatively prime integer. This pseudo-random number generation can be realized by repeating the addition of P ′, and random number generation is easy. Further, there are a large number of prime integers, and the number of columns is not limited to a prime number, and combinations with high error correction capability can be searched from a wide range when setting the number of rows or the number of columns.

  As described above, according to the present embodiment, the basic pseudo-random number of the LRI interleaver having a Latin square or Latin rectangular structure is generated using integers that are relatively prime to the number of columns. As a result, the number of choices of the number of columns increases, and it becomes easy to search for combinations having good characteristics.

Embodiment 3 FIG.
A communication system according to the third embodiment will be described. In the system of the present embodiment, the configurations of the transmission device and the reception device are the same as those in FIG. 1 of the first embodiment described above. Hereinafter, processing different from the first embodiment will be described.

  Here, an interleave pattern generation algorithm in the interleaver generator 2 according to the third embodiment will be described.

<Step 1>
(1) Set parameters.
K: Number of input information bits
N: Number of columns in the interleave matrix
M = ceil (K / N): number of rows of interleave matrix, ceil (x) is the smallest integer exceeding x
s: Start position shift number when reading the basic pseudo-random number sequence (2) Select an integer relatively prime to N.
P ': N and a prime integer (co-prime, relative prime)

<Step 2>
An interleave pattern is generated based on the following calculation formula.
u (k) = (P '× floor (k / M) + s × (M-1- (k mod M) mod N) + N × (k mod M)

<Step 3>
Pruning If there is a difference between the size of the generated pattern and the information length, the output value (u (k) ≧ K) having a size exceeding the information length is deleted.

  In the present embodiment, the basic pseudo-random number sequence is generated based on the number of columns and a relatively prime integer. This pseudo-random number generation can be realized by repeating the addition of P ′, and random number generation is easy. Further, there are a large number of prime integers, and the number of columns is not limited to a prime number, and combinations with high error correction capability can be searched from a wide range when setting the number of rows or the number of columns.

  As described above, according to the present embodiment, the basic pseudo-random number of the LRI interleaver having a Latin square or Latin rectangular structure is generated using integers that are relatively prime to the number of columns. As a result, the number of choices of the number of columns increases, and it becomes easy to search for combinations having good characteristics.

Embodiment 4 FIG.
FIG. 5 is a configuration diagram illustrating a configuration example of a communication system according to the fourth embodiment including a turbo encoder and a turbo decoder. In FIG. 5, a transmission side communication device (referred to as a transmission device) includes a bit adjuster 11, a turbo encoder 1, an interleaver generator 2a, a puncture unit 12, and a modulator 3, and a reception side communication device. (Referred to as a receiving device) includes a demodulator 4, a depuncture unit 13, a turbo decoder 5, an interleaver generator 2a, and a bit adjuster 14.

  Here, the flow of the encoding process and the decoding process in the communication system of Embodiment 4 will be briefly described.

The bit adjuster 11 in the transmission apparatus adjusts the message having the information length K based on the input bit length and the size of the interleaver generated by the interleaver generator 2a. If the input bit length is K and the size of the interleaver generated by the interleaver generator 2a is L (L ≧ K), there is a difference in (LK) bits between the two, so the original message (u ₀ , u ₁ ,. , u _K-1 ), a known signal of 0 or 1 is inserted in order to adjust (LK) bits. The insertion position is LK arbitrary places such as the beginning and end of the input bits. The turbo encoder 1 receives the bit-adjusted L-bit message and generates a codeword v encoded using the interleave pattern generated by the interleaver generator 2a. Then, the puncture unit 12 in the transmission apparatus generates a code word having a code length W according to the request of the communication system by puncturing from the output of the turbo encoder 1. The modulator 3 performs digital modulation on the codeword v generated by the puncture unit 12 by a predetermined modulation method such as BPSK, QPSK, multilevel QAM, and the like, and the modulated signal x = (x ₀ , x ₁ ,. , x _W-1 ) is transmitted to the receiving apparatus via the communication path.

On the other hand, in the receiving apparatus, the demodulator 4 performs the above-described BPSK, QPSK, multi-level QAM, etc. on the modulated signal y = (y ₀ , y ₁ ,..., Y _W-1 ) received via the communication path. Digital demodulation according to the modulation method is performed, and further, a depuncture unit 13 in the reception apparatus performs depuncture processing of the bits punctured by the transmission apparatus, and the turbo decoder 5 generates an interleaver generator for the demodulation result. Decoding is performed using the interleave pattern generated in 2a, the decoding result is output, and the bit adjuster 14 removes the adjustment bit inserted at the time of transmission, and the result (original message u ₀ , u ₁ ,. u corresponding to _K-1 ).

  Note that a plurality of communication devices including both the transmission device and the reception device may be configured, and signals may be transmitted and received between the plurality of communication devices.

  Further, as shown in FIG. 3A, the turbo encoder 1 performs using a first recursive systematic convolutional encoder 101, an interleaver 102, and a second recursive systematic convolutional encoder 103. . FIG. 3-2 is an example of the recursive systematic convolutional encoder described above. For example, a recursive systematic convolutional encoder constituting a turbo code such as a recursive systematic convolutional encoder of the W-CDMA standard is used. Any container may be used. The turbo decoder 5 is configured corresponding to the recursive systematic convolutional encoder.

  Next, an interleave pattern generation algorithm in the interleaver generator 2a of the present embodiment is shown.

<Step 1>
(1) Set parameters.
L: Number of input bits
N: Number of interleaved matrix columns (prime number)
M = ceil (L / N): Number of rows of interleaved matrix
(Ceil (x) is the smallest integer that exceeds x)
s: Start position shift number at the time of reading out the basic pseudo random number sequence (2) Select a source element of N.
G ₀ : Primitive _{element of} N

Specifically, L: Number of input bits, N: Number of columns of interleave matrix (prime number P), M = ceil (L / N): Number of rows of interleave matrix (ceil (x) is the smallest integer exceeding x) ), S: start position shift number at the time of reading out the basic pseudorandom number sequence, G ₀ : an example of the primitive _{element of} N is described.

A primitive element G ₀ list of prime numbers P and P corresponding to the input bit number L is defined. Here, corresponding to L = 40 to 82241, the prime number P (the primitive element G _{0 of} P) is {7 (3), 11 (2), 13 (2), 17 (3), 19 (2), 23 (5), 29 (2), 31 (3), 37 (2), 41 (6), 43 (3), 47 (5), 53 (2), 59 (2), 61 (2), 67 (2), 71 (7), 73 (5), 79 (3), 83 (2), 89 (3), 97 (5), 101 (2), 103 (5), 107 (2), 109 (6), 113 (3), 127 (3), 131 (2), 137 (3), 139 (2), 149 (2), 151 (6), 157 (5), 163 (2), 167 (5), 173 (2), 179 (2), 181 (2), 191 (19), 193 (5), 197 (2), 199 (3), 211 (2), 223 (3), 227 (2), 229 (6), 233 (3), 239 (7), 241 (7), 251 (6), 257 (3)} are prepared.

A) When L <= 992 1) Determine the number N of columns.
The smallest prime number P satisfying P * P ≧ L is retrieved and the number of columns N (= P) is set.
2) Select the primitive element G _{0 of} P.
3) Determine the number of rows M. M = ceil (L / N), ceil (x) is the smallest integer exceeding x.
4) Determine the number of shifts s.
When M = 7, s = 5
When M = 11, s = 10
Other than the above, s = 1
B) When L> 992 1) Determine the number of columns N.
The smallest prime number P satisfying 32 * P ≧ L is retrieved and the number of columns N (= P) is set.
2) Select the primitive element G _{0 of} P.
3) Determine the number of rows M. M = ceil (L / N), ceil (x) is the smallest integer exceeding x.
4) Determine the number of shifts s.
When M = 131, s = 2
When M = 179, s = 3
When M = 227, s = 2
Other than the above, s = 1

<Step 2>
Generate a basic pseudo-random number sequence.
C (0) = 1
For i = 0, ..., N-3 {C (i + 1) = G ₀ × C (i) mod N}

<Step 3>
The sequence C (•) is shifted s at a time to form a matrix CL _j (i).
For j = 0,1,…, M-1 {
For i = 0,1, ..., N-2 {CL _j (i) = C (i + s × j mod N)}
CL _j (N-1) = 0
}

<Step 4>
An M × N input matrix U _j (i) is prepared.
For j = 0,1,…, M-1 {
For i = 0,1,…, N-1 {U _j (i) = i + N × j}
}

<Step 5>
An M × N output matrix U ′ _j (i) is prepared using CL _j (i).
For j = 0,1,…, M-1 {
For i = 0,1,…, N-1 {U ' _j (i) = U _(M-1-j) (CL _(M-1-j) (i))}
}

<Step 6>
Generated by reading the output sequence u (k) in column order.
For j = 0,1,…, M-1 {
For i = 0,1, ..., N-1 {u (j + M × i) = U ' _j (i)}
}

In addition, when the primitive element G ₀ is 2, a multiplication by two is performed to obtain a basic pseudo-random number sequence. However, since a multiplication of 2 is possible by a 1-bit shift, a prime number with a primitive element of 2 is used. It is also possible to reduce the processing amount by limiting N only to this.

  In this embodiment, the number of columns is defined as a prime number and a pseudo-random number sequence is generated from the primitive element. However, a pseudo-random number sequence is generated using the relatively prime integer with the number of columns being an arbitrary integer. It is also possible to generate a pseudo-random number sequence without duplication by a method such as <Step 1> to <Step 6> described in Embodiments 1 and 2, and <Step 1> described in Embodiment 3. <Step 2> may be performed only.

  In the present embodiment, pruning processing in interleaver generation is eliminated by inserting known information such as 0 or 1 into the original information bits and adjusting the input length of the turbo code to the size of the interleaver. did. Thereby, the generation process of the interleaver can be simplified.

Embodiment 5. FIG.
Next, the communication system according to the fifth embodiment will be described. In the system of the present embodiment, the configurations of the transmission device and the reception device are the same as those in FIG. 1 of the first embodiment described above. Hereinafter, processing different from the first embodiment will be described.

  Here, an interleave pattern generation algorithm in the interleaver generator 2 according to the fifth embodiment will be described.

  The steps other than <Step 4> and <Step 5> are the same as those in the first and second embodiments. Only <Step 4> and <Step 5> are described.

<Step 4>
An M × N input matrix U _j (i) is prepared.
For i = 0,1,…, N-1 {
For j = 0,1,…, M-1 [U _j (i) = j + M × i}
}

<Step 5>
An M × N output matrix U ′ _j (i) is prepared using CL _j (i).
For j = 0,1,…, M-1 {
For i = 0,1,…, N-1 {U ' _j (i) = U _(M-1-j) (CL _(M-1-j) (i))}
}

Or
<Step 5>
An M × N output matrix U ′ _j (i) is prepared using CL _j (i).
For j = 0,1,…, M-1 {
For i = 0,1,…, N-1 [U ' _j (i) = U _j (CL _j (i))}
}

  When preparing an input matrix, numbers are written so as to be continuous in the row direction, but in this embodiment, numbers are written so as to be continuous in the column direction. Conventionally, the output series is read out so as to be continuous in the column direction, and the processing can be simplified by making the input / output series in the same direction for writing and reading the output series.

Embodiment 6 FIG.
Next, the communication system according to the sixth embodiment will be described. In the system of the present embodiment, the configurations of the transmission device and the reception device are the same as those in FIG. 1 of the first embodiment described above.

  Here, an interleave pattern generation algorithm in the interleaver generator 2 according to the sixth embodiment will be described. The processes other than <Step 1> are the same as those in the first embodiment described above.

<Step 1>
(1) Set parameters.
L: Number of input bits
N: Number of interleaved matrix columns (prime number)
M: Number of rows in interleave matrix
s: Start position shift number at the time of reading out the basic pseudo random number sequence (2) Select a source element of N.
G ₀ : Primitive _{element of} N

(3) A list of prime numbers P, prime elements G _{0 of} prime numbers P, and shift numbers s corresponding to the number of input bits L is defined. Here, corresponding to L = 40 to 8224, the prime number P (P primitive element G ₀ , shift number s) is {7 (3,5), 11 (2,10), 13 (2,1), 17 (3, 1), 19 (2, 15), 23 (5, 2), 29 (2, 5), 31 (3, 1), 37 (2, 1), 41 (6, 1), 43 (3,1), 47 (5,1), 53 (2,1), 59 (2,7), 61 (2,1), 67 (2,1), 71 (7,1), 73 ( 5,1), 79 (3,1), 83 (2,1), 89 (3,1), 97 (5,1), 101 (2,2), 103 (5,2), 107 (2 , 2), 109 (6,1), 113 (3,1), 127 (3,1), 131 (2,13), 137 (3,4), 139 (2,2), 149 (2, 2), 151 (6, 1), 157 (5, 1), 163 (2, 4), 167 (5, 2), 173 (2, 2), 179 (2, 3), 181 (2, 2 ), 191 (19, 1), 193 (5, 1), 197 (2, 3), 199 (3, 2), 211 (2, 2), 223 (3, 1), 227 (2, 3) , 229 (6, 2), 233 (3, 1), 239 (7, 1), 241 (7, 1), 251 (6, 1), 257 (3, 1)}.

A) When L ≦ 930 1) Determine the number of columns N.
The smallest prime number P satisfying P * (P−1) ≧ L is searched for, and the number of columns N (= P) and the number of rows M (= P−1) are set.
2) Read P's primitive element G ₀ from the list.
3) Read the shift number s corresponding to P from the list.
B) When L> 930 1) Determine the number N of columns.
The number of lines is M = 32.
The smallest prime number P satisfying M * P ≧ L is retrieved and the number of columns N (= P) is set.
2) Read P's primitive element G ₀ from the list.
3) Read the shift number s corresponding to P from the list.

<Step 2>
Generate a basic pseudo-random number sequence.
C (0) = 1
For i = 0, ..., N-3 {C (i + 1) = G ₀ × C (i) mod N}
C (N-1) = 0

<Step 3>
The sequence C (•) is shifted s at a time to form a matrix CL _j (i).
For j = 0,1,…, M-1 {
For i = 0,1, ..., N-1 {CL _j (i) = C (i + s × j mod N)}
}

<Step 4>
An M × N input matrix U _j (i) is prepared.
For j = 0,1,…, M-1 {
For i = 0,1,…, N-1 {U _j (i) = i + N × j}
}

<Step 5>
An M × N output matrix U ′ _j (i) is prepared using CL _j (i).
For j = 0,1,…, M-1 {
For i = 0,1,…, N-1 {U ' _j (i) = U _(M-1-j) (CL _(M-1-j) (i))}
}

<Step 6>
Generated by reading the output sequence u (k) in column order.
For j = 0,1,…, M-1 {
For i = 0,1, ..., N-1 {u (j + M × i) = U ' _j (i)}
}

<Step 7>
Pruning If there is a difference between the size of the prepared matrix and the information length, the output value (u (k) ≧ K) having a size exceeding the information length is deleted.

Note that the random sequence in step 2 may be generated as follows.
<Step 2>
For i = 0, ..., N-2 {C (i) = (G ₀ ) ⁱ mod N}
C (N-1) = 0

Next, an example of a decoding process when an interleave pattern is generated by the above interleave pattern generation algorithm will be described. In the decoding process, the received signal y _1k and the external information Le (x _1k ) are stored in another memory bank prepared as many as the number of rows M. For example, the first N pieces of data (0 ≦ k ≦ N−1) are stored in the memory bank 1 and the next N pieces of data (N ≦ k ≦ 2 × N−1) are stored in the memory bank 2 and the next N pieces of data. (2 × N ≦ k ≦ 3 × N−1) are stored in the memory bank 3 in order. In order to parallelize decoding before interleaving, decoding occurs for each memory bank, so the same memory bank is not accessed, and parallelization of the number of rows M (number of memory banks) is possible.

Here, in the present embodiment, it will be described that the same memory bank is not accessed in decoding after interleaving, and that the number of rows M (number of memory banks) can be parallelized. The sequence after interleaving is C (s × (M-1) mod N) + N × (M-1), C (s × (M-2) mod N) + N × (M-2 ), ..., C (s mod N) + N, C (0), C (1 + s × (M-1) mod N) + N × (M-1), C (1 + s × (M- 2) mod N) + N × (M-2), ..., C (1 + s mod N) + N, C (1), C (2 + s × (M-1) mod N) + N × ( M-1), C (2 + s × (M-2) mod N) + N × (M-2),..., C (2 + s mod N) + N, C (2),. Accesses to M memory banks occur periodically, and in order to perform parallel processing, if the memory banks accessed at the beginning of each process do not overlap, the same memory bank is not accessed during the process. When M × N sequences are divided into M, N is divided from the beginning, and N and M are in a relatively prime relationship, so the memory bank B (i) that is accessed first in each parallel processing is
B (i) = N × i mod M (0 ≦ i ≦ M-1)
And no B (i) is the same. The formula of B (i) is based on the same principle as a means for generating a pseudo random number based on a relatively prime relationship.

In addition, even when M ′ is less than M in parallel, access to the memory bank does not collide, and parallel processing can be performed. The memory bank B (i) that is first accessed in each parallel processing when M 'parallel is divided into M' pieces by W bits from the top,
B (i) = W × i mod M (0 ≦ i ≦ M-1)
It becomes. If W is relatively prime with M, another memory bank is accessed, and if M is 32, W is relatively prime if odd, so a positive integer less than or equal to the number of memory banks M is 1 ≦ W ≦ M Parallelization is possible.

In this embodiment, a pseudo-random number sequence is generated using the prime number P and its primitive element G _0. However, even if the pseudo-random number sequence is generated with a prime integer that is relatively prime to the prime number P as in the second embodiment, good. Further, as in the fourth embodiment, the size of the interleaver may not be adjusted using pruning, and a known bit may be inserted into the input bits and encoded as the same number of input bits as the size of the interleaver.

  As described above, according to the present embodiment, an LRI interleaver having a Latin square or Latin rectangular structure is designed so that the number of columns and the number of rows are relatively prime. As a result, the number of options for parallel processing increases, and in the implementation of a decoding circuit that requires high-speed processing, the options for parallel processing can be widened.

Embodiment 7 FIG.
Next, the communication system according to the seventh embodiment will be described. In the system according to the present embodiment, the configurations of the transmission device and the reception device are the same as those in the above-described sixth embodiment. Hereinafter, processing different from that of the sixth embodiment will be described.

  Specifically, in the interleave pattern generation algorithm of the seventh embodiment, the process of <Step 4> is different from that of the sixth embodiment.

<Step 4>
An M × N input matrix U _j (i) is prepared.
For j = 0,1,…, M-1 {
For i = 0,1, ..., N-1 {U _j (i) = j + M × i}
}

Next, an example of decoding processing when an interleave pattern is generated by the interleave pattern generation algorithm of the seventh embodiment will be described. In the decoding process, the received signal y _1k and the external information Le (x _1k ) are stored in another memory bank prepared as many as the number of rows M. For example, data (N) whose data number k is (k mod M = 0) is stored in the memory bank 1, data (N) whose data number k is (k mod M = 1) is stored in the memory bank 2, and so on. Data (N pieces) whose data number k is (k mod M = 2) is stored in the memory bank 3 in order. Access to M memory banks occurs periodically, and in order to parallelize, if the memory banks accessed at the beginning of each process do not overlap, the same memory bank is not accessed during the process. When M × N sequences are divided into M, N is divided from the beginning, and N and M are in a relatively prime relationship, so the memory bank B (i) that is accessed first in each parallel processing is
B (i) = N × i mod M (0 ≦ i ≦ M-1)
And no B (i) is the same. The formula of B (i) is based on the same principle as a means for generating a pseudo random number based on a relatively prime relationship.

In addition, even when M ′ is less than M in parallel, access to the memory bank does not collide, and parallel processing can be performed. The memory bank B (i) that is first accessed in each parallel processing when M 'parallel is divided into M' pieces by W bits from the top,
B (i) = W × i mod M (0 ≦ i ≦ M-1)
It becomes. If W is relatively prime with M, another memory bank is accessed, and if M is 32, W is relatively prime if odd, so a positive integer less than or equal to the number of memory banks M is 1 ≦ W ≦ M Parallelization is possible.

In this embodiment, the pseudo-random number sequence is generated using the prime number P and its primitive element G _0. However, as in the second embodiment, the pseudo-random number sequence may be generated using an integer that is relatively prime to the prime number P. . Further, as in the fourth embodiment, the size of the interleaver may not be adjusted using pruning, and a known bit may be inserted into the input bits and encoded as the same number of input bits as the size of the interleaver.

  As described above, according to the present embodiment, the LRI interleaver having a Latin square or Latin rectangular structure is designed so that the number of columns and the number of rows are relatively prime. As a result, the number of options for parallel processing increases, and in the implementation of a decoding circuit that requires high-speed processing, the options for parallel processing can be widened.

Embodiment 8 FIG.
Next, the communication system according to the eighth embodiment will be described. In the system according to the present embodiment, the configurations of the transmission device and the reception device are the same as those in the second embodiment described above. Hereinafter, processing different from the second embodiment will be described.

  Specifically, in the interleave pattern generation algorithm of the eighth embodiment, the process of <Step 4> is different from that of the second embodiment.

<Step 4>
An M × N input matrix U _j (i) is prepared.
For j = 0,1,…, M-1 {
For i = 0,1, ..., N-1 {U _j (i) = j + M × i}
}

Next, an example of decoding processing when an interleave pattern is generated by the interleave pattern generation algorithm of the eighth embodiment will be described. In the decoding process, the received signal y _1k and the external information Le (x _1k ) are stored in another memory bank prepared as many as the number of rows M. For example, data (N) whose data number k is (k mod M = 0) is stored in the memory bank 1, data (N) whose data number k is (k mod M = 1) is stored in the memory bank 2, and so on. Data (N pieces) of which the data number k is (k mod M = 2) is stored in the memory bank 3 in order. Access to M memory banks occurs periodically, and in order to parallelize, if the memory banks accessed at the beginning of each process do not overlap, the same memory bank is not accessed during the process. When M × N series are divided into W and processed in parallel, the memory bank B (i) that is divided into W pieces from the top and accessed first in each parallel process is
B (i) = N × i mod W
Therefore, parallel processing is possible when N is disjoint with W.

  Further, unlike the above, the same number of memory banks as the number of columns N are prepared, and M data (0 ≦ k ≦ M−1) from the top is stored in the memory bank 1 and the next M data (M ≦ k ≦ 2 × M−1) is stored in memory bank 2 and the next M data (2 × M ≦ k ≦ 3 × M−1) are stored in memory bank 3 in order, so that the number of columns It is possible to parallelize the same number as N or a divisor of N. If you prepare the same number of memory banks as the divisor N 'of the number of columns, it is possible to parallelize the same number N' or N 'divisor. explain. In the case where there are N memory banks, the same memory bank is never accessed because decoding is performed for each memory bank before interleaving. In the processing after interleaving, the order is not the same as before interleaving, and access to the memory bank occurs based on the generated random number sequence. There is no. Similarly, when parallelizing with a divisor of the number of columns N, one process in N parallel is serially performed d times using the divisor N ′ and the division number d such that “N = N ′ mod d”. Thus, N 'parallelism becomes possible. The same applies to parallelization using N 'memory banks.

  Further, as in the fourth embodiment, the size of the interleaver may not be adjusted using pruning, and a known bit may be inserted into the input bits and encoded as the same number of input bits as the size of the interleaver.

  As described above, according to the present embodiment, the basic pseudo-random number of the LRI interleaver having a Latin square or Latin rectangular structure is generated using integers that are relatively prime to the number of columns. As a result, the number of choices of the number of columns increases, and it becomes easy to search for combinations having good characteristics.

Embodiment 9 FIG.
Next, the communication system according to the ninth embodiment will be described. In the system according to the present embodiment, the configurations of the transmission device and the reception device are the same as those in the above-described sixth embodiment. Hereinafter, processing different from that of the sixth embodiment will be described.

  Specifically, in the interleave pattern generation algorithm of the ninth embodiment, the processes of <Step 1> and <Step 5> are different from those of the sixth embodiment.

<Step 1>
(1) Set parameters.
L: Number of input bits
N: Number of interleaved matrix columns (prime number)
M: Number of rows in interleave matrix
s: Start position shift number at the time of reading out the basic pseudo random number sequence (2) Select a source element of N.
G ₀ : Primitive _{element of} N

(3) A list of prime numbers P and prime elements G ₀ of prime numbers P corresponding to the input bit number L is defined. Here, corresponding to L = 40 to 8224, the prime number P (the primitive element G _{0 of} P) is {7 (3), 11 (2), 13 (2), 17 (3), 19 (2) , 23 (5), 29 (2), 31 (3), 37 (2), 41 (6), 43 (3), 47 (5), 53 (2), 59 (2), 61 (2) , 67 (2), 71 (7), 73 (5), 79 (3), 83 (2), 89 (3), 97 (5), 101 (2), 103 (5), 107 (2) , 109 (6), 113 (3), 127 (3), 131 (2), 137 (3), 139 (2), 149 (2), 151 (6), 157 (5), 163 (2) , 167 (5), 173 (2), 179 (2), 181 (2), 191 (19), 193 (5), 197 (2), 199 (3), 211 (2), 223 (3) , 227 (2), 229 (6), 233 (3), 239 (7), 241 (7), 251 (6), 257 (3)}.

A) When L ≦ 930 1) Determine the number of columns N.
The smallest prime number P satisfying P * (P−1) ≧ L is searched for, and the number of columns N (= P) and the number of rows M (= P−1) are set.
2) Read P's primitive element G ₀ from the list.
B) When L> 930 1) Determine the number N of columns.
The number of lines is M = 32.
The smallest prime number P satisfying M * P ≧ L is retrieved and the number of columns N (= P) is set.
2) Read P's primitive element G ₀ from the list.

<Step 5>
Generate M pseudo-random number sequences. Prepare number of lines M and disjoint integer Q,
For j = 0, ..., M-1 {D (j) = j × Q mod M}
Based on the pseudo random number D (j), an M × N output matrix U ′ _j (i) is prepared using CL _j (i).
For j = 0,1,…, M-1 {
For i = 0,1,…, N-1 {U ' _j (i) = U _{(D (j))} (CL _{(D (j))} (i))}
}

  As described above, according to the present embodiment, the LRI interleaver having a Latin square or Latin rectangular structure is designed so that the number of columns and the number of rows are relatively prime. As a result, the number of options for parallel processing increases, and in the implementation of a decoding circuit that requires high-speed processing, the options for parallel processing can be widened.

Embodiment 10 FIG.
Next, the communication system according to the tenth embodiment will be described. In the system according to the present embodiment, the configurations of the transmission device and the reception device are the same as those in the above-described eighth embodiment. Hereinafter, processing different from that of the eighth embodiment will be described.

  Specifically, in the interleave pattern generation algorithm of the tenth embodiment, the process of <Step 5> is different from that of the eighth embodiment.

<Step 5>
Generate M pseudo-random number sequences. Prepare number of lines M and disjoint integer Q,
For j = 0, ..., M-1 {D (j) = j × Q mod M}
An M × N output matrix U ′ _j (i) is prepared using CL _j (i) using the pseudo random number D (j).
For j = 0,1,…, M-1 {
For i = 0,1,…, N-1 {U ' _j (i) = U _{(D (j))} (CL _{(D (j))} (i))}
}

  In this embodiment, a pseudo-random number sequence for in-line replacement is generated using a positive integer that is relatively prime to the number N of rows. However, as in Embodiment 1, a pseudo-random sequence is generated using a primitive element of prime number P. It may be generated.

  As described above, according to the present embodiment, the LRI interleaver having a Latin square or Latin rectangular structure is designed so that the number of columns and the number of rows are relatively prime. As a result, the number of options for parallel processing increases, and in the implementation of a decoding circuit that requires high-speed processing, the options for parallel processing can be widened.

Embodiment 11 FIG.
Next, the communication system according to the eleventh embodiment will be described. In the system according to the present embodiment, the configurations of the transmission device and the reception device are the same as those in the above-described sixth embodiment. Hereinafter, processing different from that of the sixth embodiment will be described.

  Here, an interleave pattern generation algorithm according to the eleventh embodiment will be described.

<Step 1>
(1) Set parameters.
L: Number of input bits
N: Number of interleaved matrix columns (prime number)
M: Number of rows of interleaved matrix (2) Select N primitive elements.
G ₀ : Primitive _{element of} N

(3) A list of primitive elements G ₀ of prime numbers P and P corresponding to the input bit number L is defined. The primitive element of the prime number P is
(G ₀ ) ⁱ mod P (0 ≦ i ≦ P-1)
And G ₀ when all the numbers 0 ≦ i ≦ P−1 appear. For example, at P = 7, G ₀ = 3 and G ₀ = 5 are primitive elements. Here, a list is prepared corresponding to L = 40 to 8224, but a part of the list is omitted for convenience of explanation. (Prime number P: primitive element) = {(7: 3,5), (11: 2,6,7,8), (13: 2,6,7,11), (17: 3,5,6, 7,10,11,12,14), (19: 2,3,10,13,14,15), (23: 5,7,10,11,14,15,17,19,20,21) , (29: 2,3,8,10,11,14,15,18,19,21,26,27), (31: 3,11,12,13,17,21,22,24), ( 37: 2,5,13,15,17,18,19,20,22,24,32,35), (41: 6,7,11,12,13,15,17,19,22,24, 26,28,29,30,34,35), (43: 3,5,…), (47: 5,10,…), (53: 2,3,…), (59: 2,6, …), (61: 2,6,…), (67: 2,7,…), (71: 7,11,…), (73: 5,11,…), (79: 3,6, …), (83: 2,5,…), (89: 3,6,…), (97: 5,7,…), (101: 2,3,…), (103: 5,6, …), (107: 2,5,…), (109: 6,10,…), (113: 3,5,…), (127: 3,6,…), (131: 2,6, …), (137: 3,5,…), (139: 2,3,…), (149: 2,3,…), (151: 6,7,…), (157: 5,6, …), (163: 2,3,…), (167: 5,10,…), (173: 2,3,…), (179: 2,6,…), (181: 2,10, …), (191: 19,21…), (193: 5,10,…), (197: 2,3,…), (199: 3,6,…), (211: 2,3,… ), (223: 3,5, ...), (227: 2,5, ...), (229: 6,7, ...), (233: 3,5, ...), (239: 7,13, ... ), (241: 7,13, ...), (251: 6,11, ...), (257: 3,5, ...)}.

A) When L ≦ 930 1) Determine the number of columns N.
The smallest prime number P satisfying P * (P-1) ≧ L is searched and the number of columns N (= P) and the number of rows M (= P-1) are set.
2) Select P primitive element G ₀ with high decoding performance from the list.
B) When L> 930 1) Determine the number N of columns.
The number of lines is M = 32.
The smallest prime number P satisfying M * P ≧ L is retrieved and the number of columns N (= P) is set.
2) Select P primitive element G ₀ with high decoding performance from the list.

<Step 2>
Generate a basic pseudo-random number sequence.
C (0) = 1
For i = 0, ..., N-3 {C (i + 1) = G ₀ × C (i) mod N}
C (N-1) = 0

<Step 3>
The sequence C (•) is shifted one by one to form a matrix CL _j (i).
For j = 0,1,…, M-1 {
For i = 0,1, ..., N-1 {CL _j (i) = C (i + j mod N)}
}

<Step 4>
An M × N input matrix U _j (i) is prepared.
For j = 0,1,…, M-1 {
For i = 0,1,…, N-1 {U _j (i) = i + N × j}
}

<Step 5>
An M × N output matrix U ′ _j (i) is prepared using CL _j (i).
For j = 0,1,…, M-1 {
For i = 0,1,…, N-1 {U ' _j (i) = U _(M-1-j) (CL _(M-1-j) (i))}
}

<Step 6>
Generated by reading the output sequence u (k) in column order.
For j = 0,1,…, M-1 {
For i = 0,1, ..., N-1 {u (j + M × i) = U ' _j (i)}
}

<Step 7>
Pruning If there is a difference between the size of the prepared matrix and the information length, the output value (u (k) ≧ K) having a size exceeding the information length is deleted.

Note that the random sequence in step 2 may be generated as follows.
<Step 2>
For i = 0, ..., N-2 {C (i) = (G ₀ ) ⁱ mod N}
C (N-1) = 0

In the decoding process, the received signal y _1k and the external information Le (x _1k ) are stored in another memory bank prepared as many as the number of rows M. Data (N) with data number k (k mod M = 0) is stored in memory bank 1, data (N) with data number k (k mod M = 1) is stored in memory bank 2, data number Data (N pieces) where k is (k mod M = 2) is stored in the memory bank 3 in order. Access to M memory banks occurs periodically, and in order to parallelize, if the memory banks accessed at the beginning of each process do not overlap, the same memory bank is not accessed during the process. When M × N sequences are divided into M, N is divided from the beginning, and N and M are in a relatively prime relationship, so the memory bank B (i) that is accessed first in each parallel processing is
B (i) = N × i mod M (0 ≦ i ≦ M-1)
And no B (i) is the same. The formula of B (i) is based on the same principle as the means for generating a pseudo random number using a disjoint relationship.

Even if the parallel number is M or less, memory bank accesses do not collide and parallel processing can be performed. The memory bank B (i) that is first accessed in each parallel processing when W parallel is divided into N pieces from the beginning N '
B (i) = W × i mod M (0 ≦ i ≦ M-1)
It becomes. If W is relatively prime with M, another memory bank is accessed, and if M is 32, W is relatively prime if odd, so a positive integer less than or equal to the number of memory banks M is 1 ≦ W ≦ M Parallelization is possible.

Step 4
<Step 4>
An M × N input matrix U _j (i) is prepared.
For j = 0,1,…, M-1 {
For i = 0,1, ..., N-1 {U _j (i) = j + M × i}
}
It is also good.

In this embodiment, a pseudo-random number sequence is generated using the prime number P and its primitive element G _0. However, even if the pseudo-random number sequence is generated with a prime integer that is relatively prime to the prime number P as in the second embodiment, good. Further, as in the fourth embodiment, the size of the interleaver may not be adjusted using pruning, and a known bit may be inserted into the input bits and encoded as the same number of input bits as the size of the interleaver.

  As described above, according to the present embodiment, by designing an LRI interleaver having a Latin square or Latin rectangular structure so that the number of columns and the number of rows are in a relatively prime relationship, options for parallel processing are increased. In the implementation of a decoding circuit that requires high-speed processing, parallel processing also has an effect that a wide range of options can be taken. In addition, the random number sequence used in each row is easy to implement because it is only necessary to shift the start position when reading the basic pseudo-random number sequence one by one. High performance can be obtained by using a non-minimum primitive element with good original or decoding performance.

  As described above, the communication device according to the present invention can be applied to all communication devices using an error correction code such as a turbo code.

It is a figure which shows the structural example of the communication apparatus concerning this invention. It is a figure which shows the performance comparison with the turbo code | symbol interleaver concerning this invention, and a prior art example. It is a figure which shows the performance comparison with the turbo code | symbol interleaver concerning this invention, and a prior art example. It is a figure which shows the performance comparison with the turbo code | symbol interleaver concerning this invention, and a prior art example. It is a figure which shows the structural example of a turbo encoder. It is a figure which shows the structural example of a turbo encoder. It is a figure which shows the structural example of a turbo decoder. It is a figure which shows the structural example of the communication apparatus concerning this invention.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Turbo encoder 2, 2a Interleaver generator 3 Modulator 4 Demodulator 5 Turbo decoder 6 Communication path 11 Bit adjuster 12 Puncture device 13 Depuncture device 14 Bit adjuster 101 1st recursive systematic convolutional encoder 102 interleaver 103 second recursive systematic convolutional encoder 111 first decoder 112, 116 adder 113, 114 interleaver 115 second decoder 117 deinterleaver 118 determiner

Claims (35)

A first encoder that performs convolutional coding or block coding on the information bit sequence and outputs first redundant data; and a second encoder that performs convolutional coding or block coding on the information bit sequence after the interleaving process. A second encoder that outputs redundant data; and a communication device that employs a turbo encoder comprising:
The turbo encoder is
The information bit sequence is stored in an M × N input buffer, N-1 random sequences different from each other by the number of rows are prepared, M types of random sequences are generated, and all random sequences The minimum value is mapped to the Nth row in each row, an M × N mapping pattern is generated, an interleave length information bit sequence is mapped to the M × N mapping pattern, and the mapped information bit sequence is column-wise And an interleaver for outputting to the second encoder,
A communication apparatus comprising:   2. The communication apparatus according to claim 1, wherein the interleaver forms a Latin square or a Latin rectangular pattern in the M × (N−1) buffer as the M types of random sequences.   The interleaver shifts specific N−1 random sequences one or more at a time in units of rows, so that as M types of random sequences, Latin squares or M × (N−1) buffers The communication apparatus according to claim 1, wherein a Latin rectangular pattern is formed. The interleaver uses a prime number P and its primitive element G ₀ to generate a specific N−1 random sequences c (i),
c (i) = (G ₀ × c (i-1)) mod P
i = 1,2,…, P-2
c (0) = 1
The communication device according to claim 1, wherein the communication device is generated by (where N = P). The communication apparatus according to claim 4, wherein the interleaver generates a random sequence using only a prime number P of a primitive element G ₀ = 2. The interleaver uses a specific N−1 random sequences c ′ (i) using P ′,
c '(i) = (P' × i) mod N
i = 0,2, ..., N-2
4. The communication apparatus according to claim 1, 2 or 3, wherein N is generated by (where N and P 'are relatively prime).   The communication apparatus according to claim 1, 2 or 3, wherein the interleaver generates a random sequence using another method for obtaining a deterministic random sequence. A first encoder that performs convolutional coding or block coding on the information bit sequence and outputs first redundant data; and a second encoder that performs convolutional coding or block coding on the information bit sequence after the interleaving process. A second encoder that outputs redundant data; and a communication device that employs a turbo encoder comprising:
The turbo encoder is
The information bit sequence is stored in an M × N input buffer, N random sequences different from each other by the number of rows are prepared, M types of random sequences are generated, and an M × N mapping pattern is generated. An interleaver that maps an interleave length information bit sequence to the M × N mapping pattern, reads the mapped information bit sequence in units of columns, and outputs the information bit sequence to the second encoder;
A communication apparatus comprising:   The communication apparatus according to claim 8, wherein the interleaver forms a Latin square or a Latin rectangular pattern in the M × N buffer as the M kinds of random sequences.   The interleaver forms a Latin square or a Latin rectangular pattern in the M × N buffer as the M types of random sequences by shifting one or more specific N random sequences one by one in units of rows. The communication apparatus according to claim 8 or 9, characterized by the above. The interleaver uses a prime number P and its primitive element G ₀ to generate a specific N random sequences c (i),
c (i) = (G ₀ × c (i-1)) mod P
i = 1,2,…, P-1
c (0) = 1
The communication device according to claim 8, wherein the communication device is generated by (where N + 1 = P). The communication apparatus according to claim 11, wherein the interleaver generates a random sequence using only a prime number P of a primitive element G ₀ = 2. The interleaver uses a specific N random sequences c ′ (i), P ′,
c '(i) = (P' × i) mod (N + 1)
i = 0,2, ..., N-1
11. The communication device according to claim 8, 9 or 10, wherein N + 1 and P ′ are mutually prime.   The communication apparatus according to claim 8, 9 or 10, wherein the interleaver generates a random sequence using another method for determining a deterministic random sequence. A first encoder that performs convolutional coding or block coding on the information bit sequence and outputs first redundant data; and a second encoder that performs convolutional coding or block coding on the information bit sequence after the interleaving process. A second encoder that outputs redundant data; and a communication device that employs a turbo encoder comprising:
The turbo encoder is
The information bit sequence is stored in an M × N input buffer, M random sequences different from each other by the number of N−1 columns are prepared, and N−1 types of random sequences are generated. The Nth minimum value is mapped, an M × N mapping pattern is generated, an interleave length information bit sequence is mapped to the M × N mapping pattern, and the mapped information bit sequence is read in units of columns An interleaver for outputting to the second encoder;
A communication apparatus comprising:   The communication apparatus according to claim 15, wherein the interleaver forms a Latin square or a Latin rectangular pattern in the M × N buffer as the N-1 types of random sequences.   The interleaver shifts specific M random sequences one by one or more in units of columns to obtain the N-1 random sequences as Latin squares or M × (N−1) buffers in an M × (N−1) buffer. The communication device according to claim 15 or 16, wherein a latin rectangular pattern is formed. The interleaver uses a prime number P and its primitive element G ₀ to generate a specific M random sequences c (i),
c (i) = (G ₀ × c (i-1)) mod P
i = 1,2,…, P-1
c (0) = 1
The communication device according to claim 15, 16 or 17, characterized in that it is generated by (where M + 1 = P).   The communication apparatus according to claim 18, wherein the interleaver generates a random sequence using only a prime number P of a primitive element g = 2. The interleaver uses a specific M random sequences c ′ (i), P ′,
c '(i) = (P' × i) mod (M + 1)
i = 0,2, ..., M-1
18. The communication device according to claim 15, 16 or 17, wherein the communication device is generated by (where M + 1 and P 'are relatively prime).   The communication apparatus according to claim 15, 16 or 17, wherein the interleaver generates a random sequence by using another method for obtaining a deterministic random sequence. A first encoder that performs convolutional coding or block coding on the information bit sequence and outputs first redundant data; and a second encoder that performs convolutional coding or block coding on the information bit sequence after the interleaving process. A second encoder that outputs redundant data; and a communication device that employs a turbo encoder comprising:
The turbo encoder is
The information bit sequence is stored in an M × N input buffer, M random sequences different from each other by the number of columns are prepared, N types of random sequences are generated, and an M × N mapping pattern is generated. An interleaver that maps an interleave length information bit sequence to the M × N mapping pattern, reads the mapped information bit sequence in units of columns, and outputs the information bit sequence to the second encoder;
A communication apparatus comprising:   The communication apparatus according to claim 22, wherein the interleaver forms a Latin square or a Latin rectangular pattern in the M × N buffer as the N kinds of random sequences.   The interleaver forms a Latin square or a Latin rectangular pattern in the M × N buffer as the N types of random sequences by shifting specific M random sequences one by one or more in units of columns. 24. The communication apparatus according to claim 22 or 23. The interleaver uses a prime number P and its primitive element g to generate a specific M random sequences c (i).
c (i) = (g × c (i-1)) mod P
i = 1,2,…, P-1
c (0) = 1
25. The communication device according to claim 22, 23 or 24, wherein the communication device is generated by (where M + 1 = P).   26. The communication apparatus according to claim 25, wherein the interleaver generates a random sequence using only a prime number P with a primitive element g = 2. The interleaver uses a specific M random sequences c ′ (i), P ′,
c '(i) = (P' × i) mod (M + 1)
i = 0,2, ..., M-1
25. The communication device according to claim 22, 23 or 24, wherein the communication device generates (where M + 1 and P ′ are relatively prime).   25. The communication apparatus according to claim 22, 23, or 24, wherein the interleaver generates a random sequence using another deterministic method for determining a random sequence.   The turbo code may include a plurality of convolutional encoders or block encoders, and the interleaver at that time prepares a plurality of interleavers by changing specifications. The communication device according to any one of the above.   30. The communication apparatus according to claim 1, wherein N is a prime number, and M is an integer that is relatively prime to the prime number N. The interleaver defines an interleave pattern,
u (k) = (P '× floor (k / M) + s × (M-1- (k mod M) mod N) + N × (k mod M)
N: Number of columns in the interleave matrix
M = ceil (K / N): number of rows of interleaved matrix, ceil (x) is the smallest integer exceeding x
K: Number of input information bits
s: Start position shift number when reading the basic pseudo-random number sequence
P 'is an integer prime to N
23. The communication apparatus according to claim 1, 8, 15 or 22, wherein floor (x) is generated by a maximum integer not exceeding x. further,
Bit adjustment means for adjusting the information bit sequence to the size of the interleaver;
The communication apparatus according to claim 1, further comprising: Puncturing means for puncturing a code word output from the turbo encoder to a code length according to a request of a communication system;
The communication apparatus according to claim 32, comprising:   A turbo encoder comprising an interleaver in the communication device according to any one of claims 1 to 33.   A communication method comprising the interleaving process according to any one of claims 1 to 33.

Images