JPS5991761A - Maximum likelihood decoder - Google Patents

Maximum likelihood decoder

Info

Publication number
JPS5991761A
JPS5991761A JP20042682A JP20042682A JPS5991761A JP S5991761 A JPS5991761 A JP S5991761A JP 20042682 A JP20042682 A JP 20042682A JP 20042682 A JP20042682 A JP 20042682A JP S5991761 A JPS5991761 A JP S5991761A
Authority
JP
Japan
Prior art keywords
signal
likelihood
state
difference
value
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Granted
Application number
JP20042682A
Other languages
Japanese (ja)
Other versions
JPH054865B2 (en
Inventor
Yoshizumi Eto
江藤 良純
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Hitachi Denshi KK
Hitachi Ltd
Original Assignee
Hitachi Denshi KK
Hitachi Ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Hitachi Denshi KK, Hitachi Ltd filed Critical Hitachi Denshi KK
Priority to JP20042682A priority Critical patent/JPS5991761A/en
Publication of JPS5991761A publication Critical patent/JPS5991761A/en
Publication of JPH054865B2 publication Critical patent/JPH054865B2/ja
Granted legal-status Critical Current

Links

Classifications

    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04LTRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L25/00Baseband systems
    • H04L25/02Details ; arrangements for supplying electrical power along data transmission lines
    • H04L25/06Dc level restoring means; Bias distortion correction ; Decision circuits providing symbol by symbol detection
    • H04L25/067Dc level restoring means; Bias distortion correction ; Decision circuits providing symbol by symbol detection providing soft decisions, i.e. decisions together with an estimate of reliability

Landscapes

  • Engineering & Computer Science (AREA)
  • Power Engineering (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • Signal Processing (AREA)
  • Signal Processing For Digital Recording And Reproducing (AREA)
  • Dc Digital Transmission (AREA)

Abstract

PURPOSE:To simplify the circuit and to attain high speed operation of the circuit by using a difference between likelihoods of two state at a state estimating section of the maximum likelihood decoder. CONSTITUTION:A receiving value yn at an input terminal 12 is converted into a binary code at an A/D converter 13 and quantized into 4-bit. In taking a likelihood m0(n) where a ternary signal goes to +1 at a time (n) and goes to (0) or -1 at a time (n+1), and a likelihood mE(n) where the signal goes to -1 at a time (n) and (0) or +1 at (n+1), the difference of the likelihoods DELTAm(n)=m0(n)- me(n) has only a half the amplitude of the yn and becomes 3-bit, and total 7-bit of the yn and the DELTAm(n) is applied to an address input to an ROM 14. The ROM 14 outputs the difference of likelihoods DELTAm(n+1) in 3-bit after one data period, a change d0(n+1) in 1-bit and a dE(n+1) in total 5-bit. The DELTAm(n+1) is stored temporarily at each period of data to a latch 15 of 3-bit and used as the address input of the ROM at the next time. The d0(n+1) and dE(n+1) are outputted from terminals 16, 17 and used as the input to a decoding value estimating section.

Description

【発明の詳細な説明】 〔発明の利用分野〕 本発明は、3値信号として受信された信号を2値の符号
に識別する際に、より符号誤シが少なくなる様、最尤復
号する復号器に関するものである。
[Detailed Description of the Invention] [Field of Application of the Invention] The present invention provides decoding that performs maximum likelihood decoding to reduce code errors when identifying a signal received as a ternary signal into a binary code. It is related to vessels.

〔従来技術〕[Prior art]

2値の符号信号を3値の符号信号に変換し、伝送あるい
は記録を行う場合がある。3値信号は、直流成分を含ま
ない様に構成する事が可能であシ、特に直流を伝送ある
いは記録できない、又はしにくい様なシステムに用いら
れる例が多い。
In some cases, a binary code signal is converted into a ternary code signal for transmission or recording. A ternary signal can be constructed so as not to include a direct current component, and is often used in systems where direct current cannot be transmitted or recorded, or it is difficult to do so.

2値の信号から3値の信号への変換の方法は、たとえば
(1)1が発生するたびに+1.−1と交互に変換し、
0は0のままとする、(2)0から1に変化する時に+
1,1から0に変化する時に−1とし、同一レベルが継
続する時はOとする、などの方法がある。3値から2値
への逆変換は、まず受信信号−io、5.−0.5のし
きい値を用いて+1゜0に識別し、上記変換の逆変換を
行えばよい。
The method of converting a binary signal to a ternary signal is, for example, (1) +1 . Convert alternately with -1,
0 remains 0, (2) + when changing from 0 to 1
There are methods such as setting it as -1 when changing from 1, 1 to 0, and setting it as O when the same level continues. The inverse conversion from ternary to binary is performed by first converting the received signal -io, 5. It is sufficient to identify +1°0 using a threshold of -0.5 and perform the inverse transformation of the above transformation.

一般に受信信号には雑音が重じようしているために、上
記の識別の過程でいわゆる符号誤シが生じる。符号誤り
をより少なくする識別方法として最尤復号法が知られて
いる。3値信号に対する最尤復号法は、)(、Koba
yashi HApplication ofprob
abilistic ])ecoding  to ]
)igita1Magnetic Recording
 3ystems : IBMJournal  of
 Re5earch and ])evelopmen
tJan、、1971  に詳しく述べられているが要
点を以下に記す。
Generally, since noise tends to be superimposed on the received signal, so-called code errors occur in the above-mentioned identification process. Maximum likelihood decoding is known as an identification method that reduces code errors. The maximum likelihood decoding method for ternary signals is )(, Koba
yashi HApplication ofprob
abilistic ]) encoding to ]
)igita1Magnetic Recording
3systems: IBM Journal of
Re5search and ])evelopmen
Although it is described in detail in tJan, 1971, the main points are described below.

上記(1)、(2)に示した3値信号は、+1になれば
次は0か−1になり、−1になれば次はOか+1になる
性質がある。すなわち、次が0か−1になる状態(これ
1oddとする)と0か+1になる状態(これをeve
nとする)の2つの状態がある。
The ternary signals shown in (1) and (2) above have the property that if the signal becomes +1, the next signal becomes 0 or -1, and if it becomes -1, the next signal becomes O or +1. In other words, the next state is 0 or -1 (this is called 1 odd), and the next state is 0 or +1 (this is called eve
There are two states: n).

各状態の確率に相当する値を尤度と呼び、より高い尤度
を選択してゆく方法を最尤復号法と呼ぶ。
A value corresponding to the probability of each state is called a likelihood, and a method of selecting a higher likelihood is called a maximum likelihood decoding method.

時刻nの状態oddの尤度’rmo(n)、状態eve
nの尤度をm、(n)とし、受信値をylとすると、雑
音がガウス分布に従う場合にはmo(n+1)=maX
imo(”)、ml(n)+yIl−0,5)  −(
1)rnx(n+1)=maX(’no(”)  yv
=  0.5.”x(nH・−C2)が成立する。ここ
でmax (A、 B )はAとBの大きい方をとる、
との意味である。
Likelihood of state odd at time n 'rmo(n), state eve
Let the likelihood of n be m, (n) and the received value be yl, then if the noise follows a Gaussian distribution, mo(n+1)=maX
imo(”), ml(n)+yIl−0,5) −(
1) rnx(n+1)=maX('no('') yv
= 0.5. "x(nH・-C2) holds. Here, max (A, B) is the larger of A and B,
It means.

上式は以下全意味する。時刻n十iで状態がoddの場
合に、時刻nでは状態がoddの場合とevenの場合
があり得る。時刻nでod’dの確率がm、(n)、e
 ”、/ e nの確率がm、(n)であると考えた場
合に、oddおよびevenの状態が時刻n+1でod
dHなシ得る確率は各々mo(n)。
The above formula has the following meanings. If the state is odd at time n1i, the state may be odd or even at time n. The probability of odd'd at time n is m, (n), e
”, / e If we consider that the probability of n is m, (n), then the states of odd and even become odd at time n+1.
The probability of obtaining each dH is mo(n).

”E(n) +yn  o、sである。この理由は上記
の文献に詳述されており、ここでは説明を省略する。実
際に時刻n+1でoddになるのは、これらの確率の大
きい方であるから、これ’t”o(n+1)とするのが
(1)式でおる。(2)式のmx (n+1 )につい
ても同様である。すなわち、mO+ ”z K適癌な初
期値を与えれば、各時刻毎の受信値を用い、各状態の尤
度が得られ、がっ”aX(A、 B )においてA。
"E(n) + yn o, s. The reason for this is detailed in the above-mentioned literature, and the explanation will be omitted here. In reality, it is the larger of these probabilities that becomes odd at time n+1. Therefore, we can use equation (1) to define this as 't''o(n+1). The same holds true for mx (n+1) in equation (2). That is, if mO+'zK is given an appropriate initial value, the likelihood of each state can be obtained using the received value at each time, and A in aX(A, B).

Bのいずれが大きいかにより、2つの隣接した時刻間に
状態の変化があったかどうかが分る。たとえば、(1)
式でmQ (n)>mz (n)+y、 −o、sの場
合には、時刻n十iがoddであれば、時刻nもodd
であり、’no(n)<mx(n)−+−)’、  0
.5  ノ”A合KB、時刻n+1がod、dであれば
、時刻nはevenである。
Depending on which of B is larger, it can be determined whether there has been a change in state between two adjacent times. For example, (1)
In the formula, if mQ (n) > mz (n) + y, -o, s, if time n + i is odd, time n is also odd.
and 'no(n)<mx(n)-+-)', 0
.. 5. If time n+1 is od and d, time n is even.

この様に、状態の変化の様子を各時刻毎に図示してゆく
と、第1図のトレリス、嵌崗と呼ばれる図が得られる。
In this way, by illustrating how the state changes at each time, a diagram called a trellis or fitting shown in FIG. 1 is obtained.

各時刻間には必らず2本の線が存在するが、あるものは
途中で途切れ、Φるものは連続する。この内、連続した
ものを、正しい状態の変化と推定し、この状態にもとづ
いて、各時刻の復号値を推定してゆく方法が最尤復号法
である。
There are always two lines between each time, but some are interrupted midway, while others are continuous. The maximum likelihood decoding method is a method in which successive changes are estimated as correct state changes, and the decoded value at each time is estimated based on this state.

最尤復号法により、たとえば受信した信号対雑音比が1
3dBの場合には、通常の復号法による符号誤り率は約
10−3であるが、とh−tz約10−5に改善できる
と言われている。
By maximum likelihood decoding, for example, the received signal-to-noise ratio is 1
In the case of 3 dB, the code error rate by the normal decoding method is about 10-3, but it is said that it can be improved to h-tz about 10-5.

最尤復号法を実現する従来知られている回路構成を第2
図に示す。受信端子1に3値の受信信号yl1与える。
The conventionally known circuit configuration that realizes the maximum likelihood decoding method is
As shown in the figure. A ternary received signal yl1 is given to the receiving terminal 1.

メモリ2.3には尤度mo(n)。The memory 2.3 contains the likelihood mo(n).

m、(n)が一時記憶されている。4,5は尤度計算回
路で各々、”1i(n) + Ym  O,5、および
mo(l  yn  o、sを求める。比較回路6,7
は各各、mo(n)とm”(n) +yn  O,5,
l1lz(n)とmQ (n ) −y++−0,5の
大小を判定し、選択回路8゜9は比較結果を用いてその
大きい方を出力する◇この出力結果が、(1)、 (2
)式のmo (n+1)、rng(n+1)に相当する
から、これらを次の尤度としてメモリ2.3に記憶する
。一方、上記の比較結果音用いて、復号値推定回路10
により出力端子11に復号値を得る。復号値推定回路1
0の詳細は上述の文献に述べられており、ここでは省略
する。
m, (n) are temporarily stored. 4 and 5 are likelihood calculation circuits that calculate ``1i(n) + Ym O,5'' and mo(lyn o,s), respectively. Comparison circuits 6 and 7
are each mo(n) and m''(n) +yn O,5,
The selection circuit 8゜9 determines the magnitude of l1lz(n) and mQ(n) -y++-0,5, and outputs the larger one using the comparison result. ◇This output result is (1), (2
), these are stored in the memory 2.3 as the next likelihood. On the other hand, using the above comparison result sound, the decoded value estimation circuit 10
A decoded value is obtained at the output terminal 11. Decoded value estimation circuit 1
The details of 0 are described in the above-mentioned document and are omitted here.

状態の変化の推定全第2図の回路で行うと、以下の問題
が生じる。状態の変化は、メモリ、尤度計算回路、比較
回路、選択回路を用いた閉回路よシ得られる。この様に
、回路数が多く、回路規模が大きくなると言った欠点が
ある。また、閉回路であるために全体の動作速度は、各
回路の遅延時間の総和で制限される。すなわち、回路数
が多いことはt高速動作を実現しにくい要因にもなって
いる。
If the estimation of state changes is carried out using the circuit shown in FIG. 2, the following problems arise. The change in state is obtained through a closed circuit using a memory, a likelihood calculation circuit, a comparison circuit, and a selection circuit. As described above, there are drawbacks such as a large number of circuits and a large circuit scale. Furthermore, since it is a closed circuit, the overall operating speed is limited by the sum of the delay times of each circuit. That is, the large number of circuits is also a factor that makes it difficult to realize high-speed operation.

〔発明の目的〕[Purpose of the invention]

本発明は、3値信号に対する最尤復号器に占める状態推
定部を実現する回路を簡略化し、全体の回路規模を縮少
すると同時に、回路の高速動作を〔発明の概要〕 と全特徴とする。
The present invention simplifies the circuit that implements the state estimating section of a maximum likelihood decoder for ternary signals, reduces the overall circuit scale, and at the same time has high-speed operation of the circuit [Summary of the Invention]. .

更に詳しく言えば、「0または正、0または負」のレベ
ルの〈シ返しよシなる3値信号と、上記3値信号の2種
の状態の尤度差を入力とし、1デ一タ周期後の尤度差お
よび上記状態の変化の有無を出力する状態推定部と、上
記状態の変化の有無を入力とし2値の復号値を出力する
復号値推定部とで復号器を禍成したものである。
More specifically, the input is a 3-value signal with a level of "0 or positive, 0 or negative" and the likelihood difference between the two states of the above 3-value signal, and one data period is A decoder consisting of a state estimator that outputs the subsequent likelihood difference and the presence or absence of a change in the above state, and a decoded value estimator that receives the presence or absence of a change in the state as input and outputs a binary decoded value. It is.

筐ず、本発明の詳細な説明する。The present invention will now be described in detail.

上記(1)、 (2)式の両辺より、各々m、(n)を
減する。また、新たに 3m(n)=mo(n)−mz(n)   −・・−・
−−−−・−−−−−(3)とおく。これよシ m(、(n−1−1)−m、(n)−max(3m(n
)、yl−0,5)・・川・(4)mii(n+1) 
”1lz(n)””aX(3m(n)  y n  o
、s 、 o )・・・・・・(5) を得る。さらに(4)、 (5)式の差を求めるとΔr
n(n+1)=max(3m (n) 、 )’m −
0,5)−max iΔm (n ) −y、  0.
5 、0 )・・・・・・・・・(6) を得る。
Subtract m and (n) from both sides of equations (1) and (2) above, respectively. Also, newly 3m(n)=mo(n)-mz(n) −・・・・
−−−−・−−−−− (3). This is m(, (n-1-1)-m, (n)-max(3m(n
), yl-0,5)...river(4)mii(n+1)
"1lz(n)""aX(3m(n) y no o
, s, o)...(5) is obtained. Furthermore, finding the difference between equations (4) and (5), Δr
n(n+1)=max(3m(n), )'m−
0,5)-max iΔm (n)-y, 0.
5,0)......(6) is obtained.

(6)式は上記(1)、(2)式と全く同じ機能を有す
る。
Equation (6) has exactly the same function as Equations (1) and (2) above.

すなわち、3m(n)>Yカー0.5の場合には、時刻
11+1がoddであれば時刻nもoddでろシ、不等
式が逆になれば、時刻nはevenである。また、3m
(n)−3’、−0,5)0 の場合には、時刻n+1
がevenであれば時刻nはoddであり、不等式が逆
になれば時刻nはevenである。しかるに、ここで一
時記憶すべき値は尤度差Δm(n)のみとなシ、従来必
要としていた値の半分となる。
That is, in the case of 3m(n)>Y car 0.5, if time 11+1 is odd, time n is also odd, and if the inequality is reversed, time n is even. Also, 3m
(n)-3', -0,5) In the case of 0, time n+1
If is even, time n is odd, and if the inequality is reversed, time n is even. However, the value to be temporarily stored here is only the likelihood difference Δm(n), which is half of the value conventionally required.

さらに、(6)式の()の中の大小関係により4つの場
合が生じる手をここで述べたが、夷はこれは下記の3つ
の場合で十分である。すなわち、〔1〕 3m(n)≧
Y、+0.5       ・・・・・・・・・(7)
(2)  y、+o、5〉3m(n))yll−0,5
−−−−・−・−・<8)〔3〕 3m (n )≦Y
、 −0,5・・・・・・・・・(9)さらに、状態の
変化を示す符号として、下記のdo (n+1) 、 
di (n+1) を11rたに考える。すなわち do(n+1)=0 :時刻n、n+1ともodddo
(11+1)=1 :時刻nはeven、n+1はod
ddi(n+1)=0 :時刻n、n−1−1ともev
endg(n+1)=t :時刻nはOdd、n−1−
1はevenこれらと、(6)式を用いると各場合の状
態推定部の結果は以下の様に求まる。
Furthermore, although four cases have been described here depending on the magnitude relationship in parentheses in equation (6), the following three cases are sufficient. That is, [1] 3m(n)≧
Y, +0.5 ・・・・・・・・・(7)
(2) y, +o, 5〉3m(n)) yll-0,5
−−−−・−・−・<8) [3] 3m (n)≦Y
, -0,5... (9) Furthermore, as a code indicating a change in state, do (n+1) below,
Let us consider di (n+1) as 11r. In other words, do(n+1)=0: Both times n and n+1 are odd.
(11+1)=1: Time n is even, n+1 is od
ddi(n+1)=0: both time n and n-1-1 ev
endg(n+1)=t: Time n is Odd, n-1-
1 is even.Using these and equation (6), the results of the state estimator in each case can be found as follows.

〔1〕  3m (n+x ) = y、 +0.5 
   ・・・叩・・・・・α0do(n+1)=0  
      由由・旧・・(ロ)dE(n+1)=1 
       ・・・・・・・・・・・・(6)〔2〕
 3m(n+1>=ΔmCn)     −・・・−団
・−・−α3do(n+1)=0        ・・
・・旧・団・α呻di(n+1):=0       
  ・旧・川・川(至)〔3〕  3m (n+1) 
= y、−0,5−由由用DfjdoCn+1)=1 
        ・・・・・・・・・・・・αηdx(
”+1)=0         ・・・・・・・・・川
(至)以上の本発明の原理によれば、尤度差Δmのみを
記憶すれi−t’、受信値y、との大小関係よシ決まる
3種の場合に応じて、次の尤度差と、状態変化の有無が
1.tO−08)式により自動的に与えられる。
[1] 3m (n+x) = y, +0.5
...Strike...α0do(n+1)=0
Yuyu/old...(b)dE(n+1)=1
・・・・・・・・・・・・(6) [2]
3m (n+1>=ΔmCn) −・・−group・−・−α3do(n+1)=0 ・・
・・Old・Dan・α groan di(n+1):=0
・Old・river・river (to) [3] 3m (n+1)
= y, -0,5-DfjdoCn+1) = 1
・・・・・・・・・・・・αηdx(
According to the principle of the present invention described above, only the likelihood difference Δm is stored and the magnitude relationship between it' and the received value y is determined. Depending on the three types of cases determined, the following likelihood difference and the presence or absence of a state change are automatically given by equation 1.tO-08).

〔発明の実施例〕[Embodiments of the invention]

第3図に本発明による最尤符号器の第一の実施例の構成
を示す。ここでは、入力端子12の受信値y1はA/D
変換器13で2進行号に変換される。Y、Ifi4ビッ
ト程度で量子化しても、量子化しない場合の最尤復号器
の性能に比べて、劣化は極めて少ない。また、(10,
(ハ)、(ト)式より、Δmはy3の1/2の振幅変化
しかなくy、*4ビットで量子化すれば、Δmは3ビツ
トで与えられる事が分る。14′はROM (Read
 On ly Memory )であり、Ynの4ビツ
トとΔm(n)の3ビツトの合計7ビツトを番地入力と
している。またROMの出力idJm(n+1)の3ビ
ツト、do(n+1)。
FIG. 3 shows the configuration of a first embodiment of a maximum likelihood encoder according to the present invention. Here, the received value y1 at the input terminal 12 is the A/D
The converter 13 converts it into a binary sign. Even if Y, Ifi is quantized to about 4 bits, the performance of the maximum likelihood decoder without quantization is significantly less degraded. Also, (10,
From equations (c) and (g), it can be seen that Δm has an amplitude change of only 1/2 of y3, and if quantized with y, *4 bits, Δm can be given by 3 bits. 14' is ROM (Read
Only Memory), and a total of 7 bits, 4 bits of Yn and 3 bits of Δm(n), are used as address input. Also, 3 bits of the ROM output idJm(n+1), do(n+1).

d、(n+1)各1ビツトの合計5ビツトである。また
、ROMの各番地のデータは、(7)〜(9)式の場合
に応じて切〜Q8)式で与えられるものである。
There are 5 bits in total, 1 bit each for d and (n+1). Further, the data at each address of the ROM is given by formulas (7) to (9) depending on the case of formulas (7) to (9).

Δm (n+1 )は3ビツトのラッチ15にデータの
周期毎に一時記憶され、次の時刻のROMの番地入力と
して使われる。状態変化を示す結果do(n+1) 、
 di(n+1)は端子16,17.1:り出力さn、
後に述べる復号値推定部の入力として用いられる。
Δm (n+1) is temporarily stored in the 3-bit latch 15 for each data cycle, and is used as the ROM address input at the next time. A result indicating a state change do(n+1),
di(n+1) is the terminal 16, 17.1: output n,
It is used as an input to the decoded value estimator described later.

第3図を第2図と比較すると明らかな様に、第一の実施
例は入カフビット、出力5ビツトの通常規模のROM3
ビツトのラッチのみで構成され、回路規模は極めて小さ
い。また、回路の動作速度もROMとラッチの2回路の
遅延のみで制限され、高速化が容易に(はかれる。
As is clear from comparing FIG. 3 with FIG. 2, the first embodiment has a normal size ROM 3 with input cuff bits and output 5 bits.
Consisting only of bit latches, the circuit scale is extremely small. Furthermore, the operating speed of the circuit is limited only by the delay of the two circuits, the ROM and the latch, making it easy to increase the speed.

信号全高速に処理する方法として、信号を複数の系列に
分割して、並列に処理する多相処理で行うことができる
。最尤復号の場合には、尤度は1時刻前の尤度と受信値
より求めるために、これを分割して並列処理はできない
。しかし、本発明の第3図の構成を拡張すれば、多相処
理が可能となの構成を示すもので2相処理により高速化
をはかったものである。
As a method for processing signals at full speed, polyphase processing can be used in which the signal is divided into a plurality of sequences and processed in parallel. In the case of maximum likelihood decoding, the likelihood is determined from the likelihood one time before and the received value, so it cannot be divided and processed in parallel. However, if the configuration of FIG. 3 of the present invention is extended, it shows a configuration in which multiphase processing is possible, and speeding up is achieved by two-phase processing.

(7)、(財)式によりΔm(n)、y1全与えれば\
Δm(n−1−1)、do(n+i)、dE(n+1)
  が求まるが、ここで得らnたΔm(n+1)  と
次の時刻に得られるyn+1によりΔm (n+2 >
 、 ctOCn+2 )、 dg (n+2)が求ま
ることは容易に理解できる。そこで、受信値Y、に偶数
時刻のy2i+と奇数時刻のY2にや!に分けて考え、
Y2k + y2に+1とΔm(2k) ’に与え、Δ
m(2に+2)、do(2に+1)、do(2に+2)
、(IK(2に+1 ) 、 d z (2に+2) 
 を得る数表を作るごとができる。第4図において、入
力端子12に与えられた受信値ynは、A/D変換器1
3で2追打号化され、ソフトレジスター18を用いて直
並列変換されy2にとY2に+x  に分割される。R
OM14’は上述の数表を格納したもので、Δrn(2
k)、 Y2に+yzk+lk’ti地とし、Δm(2
に+2)’irラッチ15K、do(2に+1)、do
(2に+2)kシフトレジスタ19に、di(2に+1
)、a、(2に+2)全シフトレジスタ20に出力する
。ラッチ15は受信1直ynのデータ周期の2倍の周期
で動作するために、第3図と同一速度のROMとラッチ
を用いれば、全体の動作速度は第3図の2倍に高速化で
きる。なお、do’(2に+1)、do(2,に+2)
、dr、(2に+1)。
(7), if Δm(n) and y1 are all given by the formula, \
Δm(n-1-1), do(n+i), dE(n+1)
is calculated, but by Δm(n+1) obtained here and yn+1 obtained at the next time, Δm(n+2 >
, ctOCn+2), and dg(n+2) can be easily understood. Therefore, the received value Y is y2i+ at even time and Y2 at odd time! Think about it separately,
Give +1 to Y2k + y2 and Δm(2k)', Δ
m (+2 to 2), do (+1 to 2), do (+2 to 2)
, (IK (+1 to 2), d z (+2 to 2)
You can create a table of numbers to get . In FIG. 4, the received value yn applied to the input terminal 12 is input to the A/D converter 1.
3, it is converted into two additional stroke codes, serial-parallel converted using the soft register 18, and divided into y2 and Y2 +x. R
OM14' stores the above numerical table, and Δrn(2
k), Y2 is set to +yzk+lk'ti, and Δm(2
+2)'ir latch 15K, do (+1 to 2), do
(+2 to 2) k shift register 19, di(+1 to 2)
), a, (2+2) are output to all shift registers 20. Since the latch 15 operates at a cycle twice as long as the data cycle of the first reception cycle, if a ROM and latch of the same speed as in Fig. 3 are used, the overall operating speed can be made twice as fast as in Fig. 3. . In addition, do' (+1 to 2), do (+2 to 2,)
, dr, (+1 to 2).

d K (2に+’2 )はシフトレジスタ19.20
で並直列変換されてdo (n+ 1 ) + dE 
(n+ 1 )  として端子17.18に出力δれ、
全体としては第3図と同一の入出力関係を持つ状態推定
部として動作する。
d K (+'2 to 2) is a shift register 19.20
It is parallel-serial converted into do (n+1) + dE
Output δ to terminal 17.18 as (n+1),
As a whole, it operates as a state estimation section having the same input/output relationship as in FIG.

すなわち、第4図の構成は、大容量のROM(y。That is, the configuration shown in FIG. 4 uses a large capacity ROM (y.

全4ビツトとすれば、番地が11ビツト、データ出力は
7ビツト)は必要とするが、2相処理と等価な動作を災
現しておシ、動作速度が2倍に向上している。さらに大
容量のROM’に用いれば、同様の)jjc埋で3相、
4相などの処理も可能である。
If all 4 bits are used, 11 bits are required for the address and 7 bits for the data output), but the operation is equivalent to two-phase processing, and the operating speed is doubled. If used for a larger capacity ROM', a similar) jjc buried 3-phase,
Processing such as 4-phase is also possible.

第3図、第4図では、Y、’74ビットで童子化する例
を述べた。このピット数金減せば所要の回路規模はさら
に少なくなるが、反面量子化誤差により最尤復号法の効
果が少なくなる。ここでは、3ビツトの量子化の例を考
えてみる。すなわち、+1と−1の間のレベルを8個の
代表値で表わすことKなる。たとえば、代表値の例とし
てel、5,515 が適当である。ところで、±0.5ケしきいイ直として
、受信1直y1を+1.0.−1に判定する通常の復号
法から容易((想像できるが、受信値y、が土0.5近
傍の値をとる時が最尤復号法による符号誤りの改善が太
きい。何故ならば、もし雑音がなければ、受信;直y1
は±1,0の31直しかない。
In FIGS. 3 and 4, an example of converting into a doji using Y and '74 bits was described. If the number of pits is reduced, the required circuit scale will be further reduced, but on the other hand, the effectiveness of the maximum likelihood decoding method will be reduced due to quantization errors. Here, consider an example of 3-bit quantization. That is, K represents the level between +1 and -1 using eight representative values. For example, a suitable representative value is el, 5,515. By the way, assuming a +/-0.5 degree threshold, the first reception line y1 is +1.0. It is easy to use the normal decoding method that determines -1. If there is no noise, receive; direct y1
There are only 31 corrections of ±1,0.

また、雑音が小さい時はynは土1,0の近傍に集中し
、雑音による符号誤りは発生しない。また、本来0とな
るべき受信値が、たとえば+1近傍の値をとる程の大き
な雑音はめったに発生しないし、発生したら最尤復号法
を用いたとしてもこれ’kOと判定する事は困難である
。したがって、通常の復号法では符号誤りとなるが、最
尤復号法では正しく復号される可能性のめるのは、受信
値が±0.5近傍の場合である。
Furthermore, when the noise is small, yn is concentrated near 1 and 0, and no code error occurs due to noise. In addition, noise so large that the received value, which should normally be 0, takes a value near +1 rarely occurs, and when it occurs, it is difficult to determine that it is 'kO' even if maximum likelihood decoding is used. . Therefore, a normal decoding method would result in a code error, but the maximum likelihood decoding method is likely to correctly decode it when the received value is around ±0.5.

そこで、量子化の際の代表値を、前述の様に等間隔にと
らず、±0.5近傍を他よ1fflかくする。
Therefore, the representative values during quantization are not taken at equal intervals as described above, but those around ±0.5 are hidden by 1 ffl from the others.

たとえば とする。この場せの符号誤シ率の測定結果を第5図に示
すが、同じ8個の代表値を等間隔に選んだ場合よりも符
号誤シ率の改善が犬8<、16個の代表値を用いた場合
の特性に近づく事が分る。
For example. Figure 5 shows the measurement results of the code error rate in this situation. It can be seen that the characteristics are close to those obtained when using .

本発明の構成の実施例は、第3図と同一である。The embodiment of the structure of the present invention is the same as that shown in FIG.

この場合に、A/D変換器は4ビツトを用い、yllと
しては4ビットヲ考えるが、ROM14の中になylは
3ビツト相当になシ、Δmは2ピツトとなる。したがっ
て、ROMの番地は4+2=6ビツト〜出力はdOとd
、を加えて2+(IX2)=4ビットとなシ、先に第3
図全周いて説明した数値例よりは入出力とも1ビツト少
なくなる。特にROMの容量が1/2X415=0.4
倍になるなど、性能劣化が少なくて回路規模を小さくで
きる。
In this case, the A/D converter uses 4 bits, and yll is considered to be 4 bits, but yl in the ROM 14 is not equivalent to 3 bits, and Δm is 2 pits. Therefore, the ROM address is 4+2=6 bits ~ the output is dO and d
, and 2+(IX2)=4 bits.
Compared to the numerical example explained by going all the way around the figure, both input and output are 1 bit less. Especially the capacity of ROM is 1/2 x 415 = 0.4
The circuit scale can be made smaller with less performance deterioration.

第6図はアナログ回路で状態推定部を実現する実施例で
ある。端子13に与えられるy、はNPNトランジスタ
21のベースに接続されるが、バイアスを適当に設定す
ることにより、上記べ一δの電位’1Y11−0.5に
選ぶ。NPNトランジスタ22のベースにはΔm(n)
’に接続し、両トランジスタのエミッタを共通に接続す
ると、エミッタにはmax(Δm(n) l yll 
−0,5)−εが得られる。ここでεはトランジスタの
ベース、エミッタ開成位差である。一方、yllはNP
N)ランジスタ23のベースにも接続し、バイアスの設
定によpベース電位−iy、+0.5とする。トランジ
スタ23のエミッタ電位はy、+0.5−εとなる。さ
らに、トランジスタ21.22のエミッタ=zPNPト
ランジスタ24のベースに、トランジスタ23のエミッ
タ=’fPNP)ランジスタ25のベースに接続し、ト
ランジスタ24.25のエミッタを接続すると、エミッ
タより、m(max(Δm(n) + Yn  O,5
)−ε。
FIG. 6 shows an embodiment in which the state estimator is implemented using an analog circuit. The voltage y applied to the terminal 13 is connected to the base of the NPN transistor 21, and by appropriately setting the bias, the voltage y is selected to be the potential '1Y11-0.5 of the above base δ. Δm(n) at the base of the NPN transistor 22
' and connect the emitters of both transistors in common, the emitter has max(Δm(n) l yll
−0,5)−ε is obtained. Here, ε is the difference in open potential between the base and emitter of the transistor. On the other hand, yll is NP
N) It is also connected to the base of the transistor 23, and the p base potential is set to -iy and +0.5 by bias setting. The emitter potential of the transistor 23 is y, +0.5-ε. Furthermore, when the emitters of transistors 21 and 22 are connected to the base of transistor 24, the emitter of transistor 23 is connected to the base of transistor 25, and the emitters of transistors 24 and 25 are connected, m(max(Δm (n) + Yn O, 5
)−ε.

y、+o、5−t )+ε=tnin (maxiΔm
(”) + yn  o、5) 。
y, +o, 5-t )+ε=tnin (maxiΔm
(”) + yn o, 5).

y、+o、s)  が得られる。ここで、調(A、B)
とはA、Bの小さい方、の意味である。上記間【)を(
7)〜(9)式の場合に応じて計算してみると、いずれ
の場合も(10、Q3 、 Qft式で表わせるΔm(
n+1)に一致することが分る。すなわち、トランジス
タ24.25のエミッタよりΔm(n−1−1)が得ら
れる。
y, +o, s) is obtained. Here, key (A, B)
means the smaller of A and B. Between the above [) and (
When calculating according to the cases of formulas 7) to (9), in all cases, Δm(
n+1). That is, Δm(n-1-1) is obtained from the emitters of transistors 24 and 25.

これを標本低保持回路26で、−データ周期間保持し、
トランジスタ22のベースに、次の時刻の尤度として供
給する。なお、トランジスタ21゜22のペース電圧の
大小全比較器27で、ざらにトランジスタ21,23の
ベース電圧の大小を比較器28で判定すれば、各々do
 (n+ 1) 、d、(n + 1)が得られるので
、端子16.17に出力する。第6図の構成では、エミ
ツタヲ接続した2個のスイッチ回路と標本値保持回路で
尤度を求める閉回路が成シ立っており、藁速化が容易に
達成しやすい。
This is held in the sample low holding circuit 26 for -data period,
It is supplied to the base of the transistor 22 as the likelihood of the next time. Note that if the comparator 27 roughly determines the magnitude of the base voltage of the transistors 21 and 23 by the comparator 28, each do
Since (n+1), d, and (n+1) are obtained, they are output to terminals 16 and 17. In the configuration shown in FIG. 6, a closed circuit for determining the likelihood is established with the two switch circuits connected in an emitter direction and the sample value holding circuit, and it is easy to achieve high speed.

また、この様な構成は、(3)〜α印式で表現される。Moreover, such a configuration is expressed by the formula (3) to α.

本発明の原理によシ可能になっているものである。This is made possible by the principles of the present invention.

以上の実施例においては、受信信号y、より、状態変化
の有無を示す符号do(n+1)、dπ(n+1)を得
る状態推定部について述べた。ここで、do(n+i)
、 ax(n+1) *用いた復号値の推定について述
べる。ここで、本来の3値信号は、復号値が1になる度
に+1.−1と変化し、0の時は0になる場合を例にと
る。第7図にevenとoddの2つの状態間の変化、
do(n+1)、 ax(n+1)と復号値の関係の実
例を示す。これよす復号値全得るには (1)  do(n+1)k仮の復号値とする。
In the embodiments described above, the state estimation unit obtains the codes do(n+1) and dπ(n+1) indicating the presence or absence of a state change from the received signal y. Here, do(n+i)
, ax(n+1) *The estimation of the decoded value using the following will be described. Here, the original ternary signal is +1 every time the decoded value becomes 1. Let us take as an example the case where the value changes to -1 and becomes 0 when it is 0. Figure 7 shows the changes between the two states of even and odd.
An example of the relationship between do(n+1), ax(n+1) and decoded values will be shown. To obtain all the correct decoded values, (1) do (n+1)k provisional decoded values.

0)たたし、do(n+1)=1となった場合には、前
回do(n+1)とdi(n+1)の論理和が1になっ
た時刻のdo(n+1)’r反転する。
0), and when do(n+1)=1, do(n+1)'r of the previous time when the logical sum of do(n+1) and di(n+1) became 1 is inverted.

とすれば良い。これを実現する実施例が第8図である。It's fine if you do this. An embodiment that realizes this is shown in FIG.

入力端子29.30に与えられたdo(n+1)。do(n+1) given to input terminal 29.30.

di+(n+1)ノ内、do(n+1)はN段のD形フ
リップ70ツゾ31−1〜31−Nで遅延される。一方
、ORゲート32で得られるdo(n+1)+ciE(
n+1)も、N段のD形フリップフロップ33−1〜3
3−Nで遅延される。do$−よびdo’+d、が共に
1となる位置がANDゲート34−1〜34−Nで俟出
され、その時刻のdoはBxclusive  ORゲ
ー) 35−1〜35−Nで反転される。この結果が次
段の7リツプフロツプ31−2〜31−Nに転送される
が、その直後、フリップ70ツブ31.33−t、駆動
する1デ一タ周期間隔のクロックを端子36にもらい、
このクロックの後半で、フリップフロップ33−2以後
の内容をOとする。この結果、フリップフロップ33は
すべて0か、あるいは前回のdo+dg=1となった時
刻のみが1となっており、d□=1となる度にその時刻
のdoが反転され、出力38に復号値としてとシ出され
る。なお、Nが有限であるために、do=1となった時
には、前回のdo+dE=lなる情報が7リツプフロツ
プ33−N全通9過ぎてしまっている場合がある。その
際は端子39に2m (n )を入力し、この正負を比
較器40で判定し、尤度の尚い状態を現在の状態と推定
する方法が、前述の文献に述べられているが、この効果
については本発明には直接関係がないので省略する。
di+(n+1) and do(n+1) are delayed by N stages of D-type flips 70 31-1 to 31-N. On the other hand, do(n+1)+ciE(
n+1) are also N-stage D-type flip-flops 33-1 to 33-3.
Delayed by 3-N. The positions where do$- and do'+d are both 1 are determined by AND gates 34-1 to 34-N, and do at that time is inverted by Bxclusive OR games 35-1 to 35-N. This result is transferred to the next stage of 7 flip-flops 31-2 to 31-N, but immediately after that, the flip-flop 70 block 31.33-t receives a clock with a one-digit period interval to the terminal 36 to drive it.
In the second half of this clock, the contents after the flip-flop 33-2 are set to O. As a result, the flip-flops 33 are all 0, or only the previous time when do+dg=1 is 1, and every time d□=1, do at that time is inverted, and the output 38 is a decoded value. It is brought out as a. Note that since N is finite, when do=1, the previous information of do+dE=1 may have passed through 7 lip-flops 33-N. In this case, the above-mentioned literature describes a method of inputting 2m (n) to the terminal 39, determining whether it is positive or negative using the comparator 40, and estimating the state with a lower likelihood as the current state. This effect is not directly related to the present invention and will therefore be omitted.

以上では、受信信号として3値信号が得られる例を示し
た。しかし、受信信号は2値信号であっても、これ全3
値信号に変換後、最尤復号することによ多、2値信号の
ままで復号するより符号誤シ率を改善できる場合がある
Above, an example was shown in which a ternary signal is obtained as a received signal. However, even if the received signal is a binary signal, all three
By performing maximum likelihood decoding after converting to a value signal, the code error rate can often be improved compared to decoding the binary signal as it is.

2値信号を3櫃信号に変換する方法として、2値信号を
一定時間遅延し、原2値信号に加算あるいは減算する方
法がある。
As a method of converting a binary signal into a 3-level signal, there is a method of delaying the binary signal for a certain period of time and adding or subtracting it from the original binary signal.

たとえば、2値信号としてFM信号を考える。For example, consider an FM signal as a binary signal.

FM信号とは、たとえばデータの変化点では必らずレベ
ルが変化し、データが0の際にはデータの中央でも変化
するものである。また、上記一定時間として、データ周
期Tの1/2全考える。この時の原データ、FM信号お
よびT/2遅延されたFM信号との和を第9図Ql)、
←)、0に示す。(C)より明らかな様に、原データが
1となる毎に±1と変化し、原データがOの時は0とな
る3値信号が得られておL(b)k直接復号する代りに
、(C)ffi本発明の最尤復号器を用いて信号する事
が可能である。
In the FM signal, the level always changes at a data change point, for example, and even at the center of the data when the data is 0. Also, consider the entire 1/2 of the data period T as the above-mentioned fixed time. The sum of the original data, FM signal, and T/2 delayed FM signal at this time is shown in Figure 9 Ql).
←), shown in 0. (C) As is clearer, a ternary signal that changes ±1 every time the original data becomes 1 and becomes 0 when the original data is O is obtained. (C)ffi can be signaled using the maximum likelihood decoder of the present invention.

また、次の実施例として、N几2の2値信号を考え、一
定遅延時間としてデータ周期Tk考える。
Further, as the next example, a binary signal of N≠2 is considered, and a data period Tk is considered as a constant delay time.

この際の、原NRZ信号と、T遅延されたN几Z信号の
尭ヲ第9図(d)、(e)に示す。(e)より明らかな
様に、原データが0から1あるいは1から0に変化する
毎に±1と変化し、その他はOとなる3値信号が得られ
、やはり本発明の最尤復号器を用いて復号できる。たた
し、(e)と(→の関係は、(C)と(a)の関係とは
若干異なるため、(e)では第8図の復号推定部で復号
後の符号を積分することで(a)の符号が得られる。
At this time, the original NRZ signal and the T-delayed NRZ signal are shown in FIGS. 9(d) and 9(e). (e) As is clearer, a ternary signal is obtained which changes by ±1 every time the original data changes from 0 to 1 or from 1 to 0, and becomes O otherwise, and the maximum likelihood decoder of the present invention also It can be decrypted using However, the relationship between (e) and (→ is slightly different from the relationship between (C) and (a), so in (e), by integrating the code after decoding in the decoding estimation section in Figure 8, The code in (a) is obtained.

コノ他、2値信号として、Modified FM。Kono et al., Modified FM as a binary signal.

Miller 5quareと呼ばれる符号など、多く
の符号に本発明が適用できる。
The present invention is applicable to many codes, such as a code called Miller 5quare.

なお、信号を一定時間τたけ遅延して原信号と加減算す
ると言うことは、 ある。このフィルタ操作により、原信号に含まれる雑音
の周波数分布が変化し、全雑音電力が増加する場曾もあ
り、減少する場合もある。フィルタ操作で雑音が増力口
してしまっては3値化して最尤復号する効果がlぐなる
が、原信号の雑音が高周波に多い場合にば2cos−の
特性のフィルタで雑音が減少するために、原信号と遅延
信号の加算操作により、また原信号の雑音が低周波に多
い場合には25in−の特性のフィルタで雑音が減少す
るために、原信号と遅延信号の減算操作により、フィル
タ操作による雑音増加を発生させることなく3値化でき
、最尤復号の効果を発揮できる。この様に、加減算のい
ずれを選ぶか、あるいは遅延時間でとして何を用いるか
は、加減算による雑音増加がなるべく少なくなる様に、
あるいは加減算により雑音が減少する様に決めれば良い
Note that there is a case where a signal is delayed by a certain amount of time τ and then added to or subtracted from the original signal. This filter operation changes the frequency distribution of noise contained in the original signal, and the total noise power may increase or decrease in some cases. If noise is amplified by the filter operation, the effect of ternary conversion and maximum likelihood decoding will be diminished, but if the original signal has a lot of noise in the high frequencies, a filter with a 2cos characteristic will reduce the noise. In addition, if there is a lot of noise in the original signal at low frequencies, a filter with a 25in- characteristic reduces the noise by adding the original signal and the delayed signal, and by subtracting the original signal and the delayed signal. Ternaryization can be performed without increasing noise due to operation, and the effect of maximum likelihood decoding can be achieved. In this way, the selection of addition and subtraction, and the delay time to be used, are determined in such a way that the increase in noise due to addition and subtraction is minimized.
Alternatively, it may be determined such that noise is reduced by addition and subtraction.

〔発明の効果〕〔Effect of the invention〕

以上述べた様に、本発明においては、3値信号を最尤復
号する復号器の状態推定部において、従来、尤度計算、
比較、選択、一時記憶の4動作が必要であったものを、
2つの状態の尤度差を用いることで、最初の3動作を一
括して実現することが可能になった。こ几により、全体
の回路規模が縮少する他、上記4動作の総遅延時間で制
限されていた動作速度の同上ケはかることが可能になっ
た。さらに、従来、不可能であった多相化による並列処
理も可能になり、これによっても一層の動作速度の向上
がはかれるようになった。また、3値信号の重子化の際
に、不等間隔の量子化を行えば、さらに回路規模を小さ
くでさる様になった。
As described above, in the present invention, in the state estimating section of a decoder that performs maximum likelihood decoding of a ternary signal, likelihood calculation,
What used to require four actions: comparison, selection, and temporary storage.
By using the likelihood difference between the two states, it became possible to realize the first three operations at once. This method not only reduces the overall circuit scale, but also makes it possible to increase the operating speed, which was previously limited by the total delay time of the four operations mentioned above. Furthermore, parallel processing through multi-phase processing, which was previously impossible, has become possible, and this has also led to further improvements in operating speed. Furthermore, when quantizing a ternary signal, the circuit size can be further reduced by performing quantization at nonuniform intervals.

この結果、符号誤り率の改善に効果がある復号法ではあ
るが、回路規模が複雑で、高速動作は難しいとさ扛てい
た最尤復号器の問題点を解決できる様になった。
As a result, although this is a decoding method that is effective in improving the bit error rate, it has become possible to solve the problems of the maximum likelihood decoder, which had a complicated circuit scale and was difficult to operate at high speed.

さらに、復号前に3値信号に変換することにより、本発
明の最尤復号器の効果を2値信号の復号にも利用できる
ようになった。
Furthermore, by converting into a ternary signal before decoding, the effect of the maximum likelihood decoder of the present invention can also be used for decoding a binary signal.

【図面の簡単な説明】[Brief explanation of the drawing]

第1図は3値信号の状態変化を示すトレリス線図、第2
図は従来の最尤復号器の構成を示す図、第3図、第4図
、第6図、第8図はいずれも本発明による最尤復号器の
実施例の構成図、第5図は、本発明の符号誤シ率に及ぼ
す効果の実測例、第7図、第9図は本発明の・1ぎ号変
化金示す図である。 1.12.13,29,30,36.39・・・入力端
子、11,16,17.38・・・出力端子、2゜3・
・・メモ1バ 4,5・・・尤度計算回路、6,7゜2
7.28.40・・・比較回路、8,9・・・選択回路
、10・・・復号値推定部、13・・・A/D変換器、
14゜14′・・・RO1’VI、15・・・ラッチ、
18・・・直並列変換器、19.20・・・並直列変換
器、21,22゜23.24.25・・・トランジスタ
、26・・・標本値保持回路、31,33.41・・・
D形フリップフロ。 ツブ、32・・・ORゲート、34・・・ANDゲート
、35・・・排他的ORゲート、37・・・NANDゲ
ート。 第 1 図 +*II               yt−t  
   yt     −yttt′″fJ 2 区 第 3 冴 j 第 4−2 遁 5 図 4開対雑挺b(dす 第    乙   しイ] y 7 口 d、    ρ  / θ  / 6  θ  θ θ
 θ /  ρdp     ρ  0  θ ρ  
θ  /  り  6  /  θ ρdo −rdE
    ρ   /   θ   /   θ   /
   ρ  θ  /   /  ρ4a号櫃 θ θ
 θ / 6 θ θ θ / /第 82 第 q 2 (e)
Figure 1 is a trellis diagram showing state changes of a ternary signal;
The figure shows the configuration of a conventional maximum likelihood decoder, FIGS. 3, 4, 6, and 8 are configuration diagrams of embodiments of the maximum likelihood decoder according to the present invention, and FIG. FIGS. 7 and 9 are diagrams illustrating the effect of the present invention on the code error rate. 1.12.13,29,30,36.39...Input terminal, 11,16,17.38...Output terminal, 2゜3.
... Memo 1 bar 4,5... Likelihood calculation circuit, 6,7゜2
7.28.40... Comparison circuit, 8, 9... Selection circuit, 10... Decoded value estimator, 13... A/D converter,
14°14'...RO1'VI, 15...Latch,
18...Serial-to-parallel converter, 19.20...Parallel-serial converter, 21,22゜23.24.25...Transistor, 26...Sample value holding circuit, 31,33.41...・
D-type flip-flop. 32...OR gate, 34...AND gate, 35...exclusive OR gate, 37...NAND gate. Figure 1 +*II yt-t
yt -yttt'''fJ 2nd Ward 3rd Saej 4th-2nd Release 5 Figure 4 Open pair miscellaneous gun b (dsu th Otoshii) y 7 Mouth d, ρ / θ / 6 θ θ θ
θ / ρdp ρ 0 θ ρ
θ / ri 6 / θ ρdo −rdE
ρ / θ / θ /
ρ θ / ρ4a box θ θ
θ / 6 θ θ θ / /82nd q 2 (e)

Claims (1)

【特許請求の範囲】 1、「0または正、0または負」のレベルのくシ返しよ
りなる3値信号と、上記3値信号の2樵の状態の尤度差
を入力とし、1デ一タ周期後の尤度差および上記状態の
変化の有無を出力とする状態推定部と、上記状態の変化
の有無を入力とし2値の復号値全出力とする復号値推定
部を有してなる事を特徴とする最尤復号器。 2、上記3値信号の2進符号値と、上記尤度差の2進符
号値を番地とし、1デ一タ周期後の尤度差の2進符号値
と上記状態変化の有無の符号値を出力符号とするメモリ
より上記状態推定部が構成された$を特徴とする第1項
記載の最尤復号器。 3、上記3値信号の2進符号値系列の隣接した複数の符
号値と、上記尤度差の2進符号値の上記複数値毎の値と
を番地とし、上記複数値データ周期後の尤度差の2進符
号値と、上記複数値データ周期間の上記状態変化の有無
を出力符号とする事を特徴とする第2項記載の最尤復号
器。 4、上記3値信号の2進符号は、上記正および負の信号
レベルの中心近傍の量子化を他より細かくした事を特徴
とする第2項記載の最尤復号器。 5、上記状態推定部は、3値信号と尤度差をペース入力
としエミッタを共通接続した2個のNPNトランジスタ
と、上記共通エミッタおよび上記3値信号をベース入力
としエミッタを共通接続した2個のPNP )ランジス
タと、上記PNPトランジスタのエミッタ出力を1デ一
タ周期遅延する回路とよシなる事を特徴とする第1項記
載の最尤復号器。 6、上記3値信号は、2値打号系列と、上記2値打号系
列を一定時間たけ遅延した系列との和または差よシ合成
されたことを特徴とする第1項記載の最尤復号器。
[Claims] 1. A ternary signal consisting of a comb pattern of levels "0 or positive, 0 or negative" and the likelihood difference between the two woodcutter states of the ternary signal are input, and a 1-digit signal is input. a state estimating section that outputs the likelihood difference after the data period and the presence or absence of a change in the state; and a decoded value estimator that receives the presence or absence of a change in the state as an input and outputs the entire binary decoded value. A maximum likelihood decoder characterized by: 2. The binary code value of the ternary signal and the binary code value of the likelihood difference are the addresses, and the binary code value of the likelihood difference after one data period and the code value of the presence or absence of the state change. 2. The maximum likelihood decoder according to claim 1, wherein the state estimator is constructed from a memory whose output code is . 3. A plurality of adjacent code values of the binary code value series of the above-mentioned ternary signal and a value of each of the above-mentioned plurality of binary code values of the above-mentioned likelihood difference are set as addresses, and the likelihood after the above-mentioned multi-value data period is 3. The maximum likelihood decoder according to claim 2, wherein the binary code value of the degree difference and the presence or absence of the state change between the plurality of data periods are output codes. 4. The maximum likelihood decoder according to item 2, wherein the binary code of the ternary signal is quantized in the vicinity of the center of the positive and negative signal levels more finely than the others. 5. The state estimator includes two NPN transistors whose emitters are connected in common and whose base inputs are the ternary signal and the likelihood difference; and two NPN transistors whose emitters are connected in common and whose base inputs are the common emitter and the ternary signal. 2. The maximum likelihood decoder according to claim 1, characterized in that it is similar to a PNP transistor (PNP) transistor and a circuit that delays the emitter output of the PNP transistor by one data period. 6. The maximum likelihood decoder according to item 1, wherein the ternary signal is synthesized by the sum or difference of a binary encoding sequence and a sequence obtained by delaying the binary encoding sequence by a certain amount of time. .
JP20042682A 1982-11-17 1982-11-17 Maximum likelihood decoder Granted JPS5991761A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
JP20042682A JPS5991761A (en) 1982-11-17 1982-11-17 Maximum likelihood decoder

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
JP20042682A JPS5991761A (en) 1982-11-17 1982-11-17 Maximum likelihood decoder

Publications (2)

Publication Number Publication Date
JPS5991761A true JPS5991761A (en) 1984-05-26
JPH054865B2 JPH054865B2 (en) 1993-01-21

Family

ID=16424097

Family Applications (1)

Application Number Title Priority Date Filing Date
JP20042682A Granted JPS5991761A (en) 1982-11-17 1982-11-17 Maximum likelihood decoder

Country Status (1)

Country Link
JP (1) JPS5991761A (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JPH0563582A (en) * 1991-03-15 1993-03-12 Internatl Business Mach Corp <Ibm> Coding/decoding method and device in dictionary edition manner

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JPH0563582A (en) * 1991-03-15 1993-03-12 Internatl Business Mach Corp <Ibm> Coding/decoding method and device in dictionary edition manner

Also Published As

Publication number Publication date
JPH054865B2 (en) 1993-01-21

Similar Documents

Publication Publication Date Title
US8022850B2 (en) Multiple-bit, digital-to-analog converters and conversion methods
JP3590209B2 (en) Modulation coding method
JP2819006B2 (en) Thermometer binary encoding method
EP1137001A1 (en) Reproduction of recorded data using a soft input/soft output demodulator
JPH01256251A (en) Encoding method and decoding method
JPH0821958B2 (en) Coding method
CN110875740B (en) Digital-to-analog converter
US7656337B2 (en) Method and system for bit polarization coding
US7339500B2 (en) Encoding method and decoding method
US5781130A (en) M-ary (d,k) runlength limited coding for multi-level data
JP2001266499A (en) Unit and method for recording data, device and method for reproducing data, unit and method for recording and reproducing data
JPS5991761A (en) Maximum likelihood decoder
JP4489973B2 (en) Decoding to encode from n-bit source word to corresponding m-bit channel word and vice versa where the transform is parity inversion
JP3951441B2 (en) Code state determination method and encoding apparatus
WO1998017005A1 (en) Method and circuit for digital modulation and method and circuit for digital demodulation
KR100727690B1 (en) Apparatus for implementing an extended partial response maximum likelihoodeprml channel
JP3858362B2 (en) Decoding apparatus and method, and data reproducing apparatus
JPH11154873A (en) Encoding circuit, encoding method, digital signal transmission equipment, and digital magnetic recording device
JP2002313035A (en) Encoding method and recording medium, decoding method, and recording medium reproducing device
KR20070069288A (en) Circuit for converting gray code
JP3083119B2 (en) Encoding / decoding circuit using adaptive delta modulation scheme
EP0521744A2 (en) Maximum likelihood decoder
JP2978181B2 (en) NTM modulation method
US5675330A (en) M=5(4,11)runlength limited code for multi-level data
JPH0536213A (en) Code conversion system