JP2000031837A - Encoding method - Google Patents

Abstract

(57) [Summary] [PROBLEMS] To realize a turbo coding method in which codes having different code lengths are combined. SOLUTION: Codes used in encoders 101 and 102 are systematic block codes having different code lengths, or systematic convolutional codes having different code lengths, or one is a systematic block code and the other is a systematic convolutional code. .

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an encoding method, and more particularly, to an encoding method for transmitting information through a noisy communication path.

[0002]

2. Description of the Related Art An error correction can be performed by encoding an information signal and decoding the information signal, and various techniques have been tried or put to practical use to date. When this error correction is performed, the signal to noise ratio (or Eb / N0, E
b: the power per bit, N0: the noise power per HZ), the problem is whether error correction is possible and the information signal can be demodulated at a predetermined error rate, no matter how low it is.
This is a very important problem when performing reliable wireless communication.

[0003] Shannon's theorem supports the theoretical limit of this problem. Let S be the average input power to the communication path and N be the average noise power. The capacity C (bits / sample) of this channel was calculated by Shannon. The formula given by Shannon is:

## EQU1 ## Channel capacity C = (1/2) .log ₂ (1+ (S / N)) Baseband discrete-time channel with its bandwidth (both sides)
Is 2W, the input signal of the receiver is 2W per second.
Nyquist sample data exists. If the noise is white Gaussian noise, 2W noise samples per second are independent, so such a discrete time channel is a memoryless noise channel. Therefore, the capacity Cb expressed in units of the number of bits per second is given by the following equation.

## EQU2 ## Cb = W.log ₂ (1+ (S / N))

The lower bound of Eb / N0 is obtained from the above equation. Suppose that the signal power can be expressed as S = Eb · R, and the noise power N = N0 · W. Since the information rate R is smaller than the communication capacity Cb, the following inequality is obtained from (Equation 2).

R / W <Cb / W = log ₂ (1 + ((R · Eb) /
(W · N0))) If the spectrum bit rate r is r = R / W, (Equation 4) is

## EQU4 ## Eb / N0> ( ^2r- 1) / r. When the spectral bit rate r approaches 0, the denominator and the numerator on the right side of (Equation 4) are differentiated with respect to r separately by applying de L'Opital's theorem to obtain the limit value.

Eb / N0> (2 ^r -1) / r → 2 ^r · log _e 2 →
0.693 ≒ -1.6 (dB) That is, Shannon's theorem states that the value of Eb / N0 is not better than about -1.6 (dB), regardless of what coding is performed on the Gaussian channel. Means.

[0005] A turbo code is a code which has approached Shannon's limit fairly asymptotically. This code is described in the literature "C. Berrou, A. Glavieux and G. Montorsi:" NEAR SHANNO
N LIMIT ERROR-CORRECTING CODING AND DECODING: TURBO
-CODES (1) ”, Proc.ofICC '93 (Geneva, Switzerland), pp.
1064-1070 "for the first time, and its performance was noted as a breakthrough. FIG. 4 is a diagram illustrating a configuration of the turbo encoder. The input information signal is transmitted to the encoder 40.
1 and encoded. The other of the input information signals is interleaved at a predetermined size by an interleave circuit 403, and then encoded by an encoder 402. The switch 404 alternately outputs the outputs from the encoders 401 and 402 in response to the interleaving. Further, this output is output from the parallel / serial converter
Are switched between uncoded bits (bits of input information) and coded bits (parallel / serial conversion) and output. Here, the encoding in the encoder 401 and the encoder 402 is the encoding by the systematic convolutional code (n1, k).

FIG. 5 shows a configuration of a turbo decoder.
The input signal is serial / parallel converted by the serial / parallel converter 501,
It is input to MAP decoders 502 and 503. The output of the MAP decoder 502 is input to an interleave circuit 504, and is rearranged by a length corresponding to the interleave at the time of encoding. These signals are decoded by the MAP decoder 503 and subjected to the reverse operation of the interleaving by the deinterleave circuit 506 and output as decoded outputs. On the other hand, part of the output of the MAP decoder 503 is subjected to the reverse operation of the interleaving by the deinterleave circuit 505,
Input to the decoder 502. This operation is repeated as necessary.

Here, MAP decoding will be described. Encoding can be performed correctly when the code word wj is transmitted when the received word y enters the decoding region Rj of the code word wj. Therefore, the correct decoding probability Pc in this case is calculated assuming that the probability that each codeword is sent is P (wj).

(Equation 6) In the above equation, the connection probability P (wj, y) = P (wj) · P
(Y | wj) should be maximized. After all, for a given received word y, the conditional probability P (y | w
What is necessary is just to judge that the code word which maximizes j) was sent.
The conditional probability P (y | wj) is called the posterior probability, and the decoding that estimates that the codeword that maximizes this probability has been sent is the maximum posteriori probability decoding (maximum a posteriori probability dec).
oding: MAP decryption). This algorithm is described in the literature "LRBahl, J. Cocke, F. Jelinek, and J. Raviv ,:
“Optimal decoding of linear codes for minimizing
symbolerror rate, ”IEEE Trans.Inform.Theory, Vol.IT
-20, pp. 284-287, 1974 ".

FIG. 6 shows a convolutional code ((n, k) = (3
7, 21), coding rate = １ ／), the interleaving size is 256 × 256, and the number of repetitions (iter
In this case, the bit error rate with respect to Eb / N0 at the time of changing the # ation # 1 is changed. As can be seen from the figure, the bit error rate improves as the number of repetitions increases.

[0009]

In the encoding method using the conventional turbo code described above, the codes used in the two encoders are the same, and the same convolutional code is used as a concatenated code for both. No consideration has been given to the case where the two codes are different types of codes or have different code lengths.

An object of the present invention is to provide an encoding method using two systematic codes of different types or two systematic codes having different code lengths as concatenated codes.

[0011]

In order to achieve the above object, the present invention provides an information signal encoded by a first encoder, and the information signal is interleaved by an interleave circuit and then interleaved by a second encoder. Encoding the first
And an output of the second encoder is switched by a switcher, and the output is switched between an uncoded bit and a coded bit. And a code used in the second encoder are systematic codes having different code lengths.

Further, the present invention provides an encoding method according to the above-mentioned encoding method, wherein the codes used in the first encoder and the second encoder are both systematic block codes. provide.

Further, the present invention provides an encoding method according to the above-mentioned encoding method, wherein both codes used in the first encoder and the second encoder are systematic convolutional codes. provide.

Further, according to the present invention, in the above-mentioned encoding method, the codes used in the first encoder and the second encoder are
An encoding method is characterized in that one is an organized convolutional code and the other is an organized block code.

[0015]

Embodiments of the present invention will be described below. FIG. 1 is a block diagram showing a configuration example of an encoder for performing encoding by the encoding method according to the present invention. The configuration of FIG. 1 is the same as that of FIG. 4, and the input information signal is input to encoder 101 and encoded. On the other hand, the other of the input information signals is interleaved by an interleave circuit 103 in a predetermined size.
2 is encoded. The switch 104 alternately outputs the outputs from the encoders 101 and 102 in response to the interleaving. Further, this output is input to the parallel / serial converter 105, which switches between uncoded bits (bits of input information) and coded bits (parallel / serial conversion).
Is performed and output. Here, the codes used in the encoders 101 and 102 are conventionally the same convolutional codes, but in FIG. 1, different codes or codes having different code lengths are used. However, there is no difference in that the two codes are concatenated codes.

FIG. 3 shows two encoders 101, 10 of FIG.
In illustration of concatenated codes used in the 2, reference numeral C1 is q ^k Motoue of (n1, k1) linear code, code C2 is the q Motoue (n2,
k) Linear codes, both systematic codes. The hatched portion is the code word of C1. The k words of the q ^element are arranged side by side to form a q ^{k element} word, and the k1 elements of the q ^{k element} are vertically arranged k 1 information symbols. And n1 -k1 beneath it to form a check symbol for C1. The encoder 101 uses this code C1
Is performed. Corresponding codewords C1 described above, which is configured all any check symbol any information symbol words on q origin in the k (next) word in the q ^k source aligned, code C2 is of Each of the words on the q ^{k element is} regarded as a sequence of k words on the q ^element , and it is encoded into a code word having a code length n2. Therefore, the hatched portion in FIG. 3 is the information symbol of symbol C2, and the blank portion is the check symbol of symbol C2.
02 performs encoding using the code C2.

Here, if n1 = n2, k1 = k, and the codes C1 and C2 are systematic convolutional codes, conventional turbo coding is performed. In the present invention, at least two codes C1 and C2 are used.
1, the code lengths n1 and n2 of C2 have different values. Further, as codes C1 and C2, under the condition that n1 ≠ n2,
Both use an organized block code, both use an organized convolutional code, or one uses an organized block code and the other uses an organized convolutional code.

Even when the above codes C1 and C2 are used, the coding operation described with reference to FIG. 1 is the same as that of the conventional turbo code. The configuration of a decoder corresponding to the encoder of FIG. 1 is shown in FIG. 2. The input signal is serial / parallel converted by a serial / parallel converter 201, and a MAP decoder 202
And 203. The output of the MAP decoder 202 is input to an interleave circuit 204, and is rearranged by a length corresponding to the interleave at the time of encoding. M
The AP decoder 203 decodes these signals, and the deinterleaving circuit 206 performs an operation reverse to interleaving to output the decoded output. On the other hand, the MAP decoder 20
Part of the output of No. 3 is subjected to an operation opposite to the interleaving in the deinterleave circuit 805, and is input to the MAP decoder 202. This operation is repeated as necessary. This operation is the same as the conventional turbo coding.

[0019]

According to the present invention, turbo coding using various systematic codes can be performed, and according to transmission path characteristics, required error rate characteristics, or available calculation resources (calculation time, memory). Appropriate encoding becomes possible.

[Brief description of the drawings]

FIG. 1 is a block diagram illustrating a configuration example of an encoder including an encoding method according to the present invention.

FIG. 2 is a block diagram illustrating a configuration example of a decoder for data encoded by the encoder of FIG. 1;

FIG. 3 is an explanatory diagram of a connection code;

FIG. 4 is a block diagram illustrating a configuration of a conventional turbo encoder.

FIG. 5 is a block diagram showing a configuration of a conventional turbo decoder.

FIG. 6 is a diagram illustrating a bit error rate characteristic of the turbo coding system.

[Explanation of symbols]

 101, 102 Encoders 103, 204 Interleave circuit 104 Switch 105 Parallel / Direct converter 201 Direct / Parallel converter 202, 203 MAP decoder 205, 206 Deinterleave circuit

Claims (4)

[Claims] An information signal is encoded by a first encoder, the information signal is interleaved by an interleave circuit, and then encoded by a second encoder.
And an output of the second encoder is switched by a switcher, and the output is switched between an uncoded bit and a coded bit. And a code used in the second encoder are systematic codes having different code lengths. 2. The encoding method according to claim 1, wherein:
A coding method, wherein both codes used in the first encoder and the second encoder are systematic block codes. 3. The encoding method according to claim 1, wherein:
A coding method, characterized in that the codes used in the first encoder and the second encoder are both systematic convolutional codes. 4. The encoding method according to claim 1, wherein:
An encoding method, characterized in that one of the codes used in the first encoder and the second encoder is an organized convolutional code, and the other is an organized block code.