SU1580567A1 - Codec of nonsystematic convolution code - Google Patents

Abstract

The invention relates to computing and communication technology. Its use in digital communication systems improves noise immunity. A codec consists of transmitting side 1, receiving side 2, and communication channel 3. Transmitting side 1 has encoder 4. Receiving side 2 contains calculator 6 syndrome, calculator 7 evaluations, analyzer 9 syndrome, switch 10 and majority element 11. Due to the introduction of encoder 5 on the transmitting side, two decoders 12, 13, the pulse generator 14, the key element 15, the OR element 16, the time interval generator 17 and the threshold counter 18, the codec realizes obtaining a quasi-orthogonal non-systematic convolutional code, the minimum codeword weight of which is larger than orthogonal non-systematic convolutional code. 2 Il.

Description

FIG. one

The invention relates to computing and communication technology and can be used in digital communication systems.

The purpose of the invention is to improve noise immunity.

Figure 1 shows the block diagram for the decade in FIG. 2, a diagram of its specific implementation.

The codec of a non-systematic convolutional code consists of transmitting 1 and receiving 2 sides and channel 3 of communication. Transmitting side 1 contains the first 4 and second 5 coders, the receiving side 2 - the calculator 6 syndrome, the evaluator 7 evaluations, block 8 modulo adders, the analyzer 9 syndrome, the switch 10, the majority element 11, the first 12 and the second 13 A generator, a pulse generator 14, a key element 15, an OR element 16, a time interval generator 17i and a threshold counter 18. Figure 1 also shows input 19, the first 20, and second 21 codec outputs.

The first encoder 4, which generates the code words of an orthogonal unsystematic convolutional code, is intended to form the check sequence symbols over the information symbols. As a coder, a register is used (FIG. 2).

22 shift associated with adders

23 and 24 modulo two. The length of the shift register, the number of modulo-adders and their connections are determined by the generating polynomials of the orthogonal non-systematic convolutional code

The second encoder 5 is designed to form additional check bits in the generator polynomials of a quasi-orthogonal non-systematic convolutional code that can detect erroneous decoding of received information signals at the receiving side 2. The second encoder 5 is the shift register 25 associated with the first 26 and second 27 modulo-two adders, the number of register-25 divisions and communications is due to the type of additional polynomials in the generating polynomials of the quasi-orthogonal non-systematic convolutional Yes.

Syndrome calculator 6 is designed to multiply the quasi-orthogonal unsystematic check sequences received from channel 3

a convolutional code on the generating polynomials of the orthogonal nonsystematic convolutional code P. .. (x and P01 (x) and to form the syndrome sequence from the check sequence that is adopted and the check sequence formed on the receiving side 2. The calculator 6 of the syndrome represents are the shift registers 28 and 29 and the adder 30 modulo two; the connections of the bits of the registers 28 and 29 to the adder 30 are determined by the generating polynomials P0 ((x) and Horn (x) of the orthogonal non-systematic convolutional code.

The estimate calculator 7 is intended to form the evaluation signals EJ, orthogonal with respect to the information symbol, from the orthogonal non-systematic convolutional code codes and represent shift registers 31 and 32 associated with modulators 33 and 34 two. The length of the shift registers, the number of modulo-two adders, and their connections are determined by the structure of checks for each specific polynomial of the orthogonal non-systematic convolutional code.

Block 8 is intended to exclude from the consideration at the inputs of the calculator 7 estimates of the additional check sequences M (x) x xPv | (x) and M (x) Par (x) formed by multiplying the information sequence M (x) by additional generating polynomials Rd | ( x) and Rl2 (x) in an encoder that generates code words of a quasiorthogonal non-systematic convolutional code. Block 8 is the moduli two adders 35 and 36.

The syndrome analyzer 9 is designed to detect errors in the test symbols and is a shift shift register 37 for recording the sequence of syndromes S and a logic unit 38 connected to it, generating the following signals:

T, S,

S, S

 +

+ S, S6 + S5S4 + SjS4S

The serial output of the shift register 37 is connected to the key element 15 and the threshold counter 18. The switch 10 is intended to exclude from the review at the input of the majority element 11 that estimate E-, the number of which coincides with the number of the non-zero function T :, if.

The majority element 11 is designed to decide on the reliability of the received symbols on most E estimates: at the first output, as well as forming at the second output based on the E estimates of the signal T0. In this way,

„

(E, E2-E (E, - E2-E (E,

VE4), -Е4)

(E, -E2 (K, +

E, E4) +

Е2-Е, -Е4) + (Е, - E2-EVE4)

The first 12 and second 13 decoders are multiplication schemes for the decoded information sequence by polynomials of the form P (x)

Р0, (х) -Р „(х) and Р (х) Р

og

R

about r qz

(x)

Ra, (x) and Pa2 (x). The first decoder 12 is formed by shift register 39 and modulators 40 - 42 modulo two, the second decoder 13 - shift register 43 and modulo 44 adder 44 (as in this example, RL, (x) Pa 2 (x)).

The pulse generator 14 generates the pulses necessary for the operation of a time interval formaker 17. The key element 15 is designed to apply pulses from the generator output 14 pulses to the input of the time generator 17 in the time interval from the moment the error signal arrives at the first control input of the key element 15 from the second output of the syndrome analyzer 9 until it arrives at its second control input signal from the output element OR 16.

The time interval generator 17 is designed to set a time interval for analyzing information received at the input of the threshold counter 18. A time counter is used for the time interval generator 17, the trigger threshold of which is selected based on the message transformation tolerance, length of code constraints, corrective ability of quasi-10

20

thirty

35 805676

non-systematic tonal code.

The threshold counter 18 is designed to count the number of error signals from the second output of the analyzer 9 syndrome, and issue a signal about the presence of errors at the output 21 in case of overflow. The threshold for triggering the threshold counter 18 is selected based on the correction properties of the quasi-orthogonal non-systematic convolutional code and the requirements for the acceptable probability of message transformation.

During the operation of the codec, three cases are possible: the decoding of information occurs in the absence of errors; information decoding occurs when there are errors whose level is less than or equal to the correcting ability of the orthogonal non-systematic convolutional code; information decoding occurs when there are 25 errors, the level of which is more than the correcting ability of the orthogonal non-systematic convolutional code.

In the codec under consideration, by introducing an additional coder and decoders, slightly increasing the length of the code constraints of the obtained quasi-orthogonal non-systematic convolutional code, it is possible to obtain such a code whose minimum weight of code words is increased compared with the code words of an orthogonal non-systematic convolutional code, approaching the weight best non-orthogonal codes. And if in the first and second cases the codec works with identical to known codecs of an orthogonal unsystematic convolutional code, retaining the main advantage of an orthogonal non-systematic convolutional code - simplicity of technical implementation, in the third case, due to the increased minimum weight of code words of the quasi-orthogonal non-systematic convolutional code, error detection is possible , the weight of which is greater than the weight of errors that are capable of correcting non-systematic convolutional code.

The codec works as follows.

In the initial state, registers 22, 25, 28, 29, 31, 32, 37 shift encoders 4 and 5, calculator 6 syndrome, calculator 7 ratings, analyzer 40

45

50

55

pa 9 and decoders 12 and 13 syndrome recorded zero signals. The contents of the threshold counter 18 and the time interval generator 17 are also zero. The output of the generator 14 pulses of the key element 15 is disconnected from the input of the former 17 of the time interval. In the majority element 11 is set the threshold value

U O / IM H

2, where do min - minimum

the code distance of the orthogonal unsystematic convolutional code.

In the threshold counter 18 is set js)) pk (x) PQ; J (x) -, the threshold Y

do min

 allowable number

P2P0, (x) M (x) Pkg (x) P01 (x),

errors on the interval W cycles in received from channel 3 of sv. zi encoded sequences, the signal from the output of the threshold counter 18 is exhausted if the threshold value Y is exceeded. In the time interval imaging unit 17, the threshold of the number of clock cycles W is set, and the signal at its output is measured if the number W is exceeded.

At the input 19 of the codec (the input of the encoder 4) in a sequential code, the signals of the binary information sequence M (x) are received with a clock period UT. This sequence advances with a clock frequency according to the bits of the register 22 of the shift of the encoder 4 and further on the bits of the register of the 25 shift of the encoder 5. During this movement, the sequence M (x) with the help of adders 23, 24, 26, 27 modulo

20

75

P2P0, (x) M (x) Pkg (x) P01 (x), (3)

modulo two, which are component-wise compared with the sequence M (x) (P + P) supplied to the third input of the calculator 6 from the second output of the decoder 12, i.e.

P02 (x) + P2 RL1 (x) + M (x) x

 oi

thirty

& (x) P, {p (x) + P (x) M (x) PO, (x) -P02 (x)

e

P x + RL, (x)

PO, (x) + P0, (x) - P0, (x) -x +

02

01

+ PA2 (x) O1

(x) + P02 (x) PЈ, (x) +

+ P0 ((x) -Pa2 (x)} 0.

35

Thus, if no errors occurred in channel 3, then zero syndrome is recorded in the shift register 37 of the syndrome analyzer 9, i.e.

two, included in encoders 4 and 5, up to S (x) 0. At the same time, it is re-converted in the sequence of the form.

M (x) (x) -he + Rda (x) M (x) x

P „, (x);

M (X), (x) -x8 xR, (x), Pn, (x) M (x) x

(one)

which in the serial code are fed to the inputs of the communication channel 3. At the same time, the signals of the check sequences M (x) Pc (x) and M (x) -Rc2 (x) are distorted by the interference E, (x) and Eh (x), i.e., binary sequences of the form 3 come from the communication channel

M (x) M (x)

"one

kg

$ E, (x); (x) (x).

(2)

If there are no errors, i.e. E ((x) E2 (x) 0, the sequences of the form (1) enter the bits of the registers 28 and 29 of the calculator of the syndrome 6 syndrome and advance along them with a clock frequency. During this movement, these sequences are transformed into sequences of the form

)) pk (x) PQ; J (x) -,

P2P0, (x) M (x) Pkg (x) P01 (x), (3)

modulo two, which are component-wise compared with the sequence M (x) (P + P) supplied to the third input of the calculator 6 from the second output of the decoder 12, i.e.

P02 (x) + P2 RL1 (x) + M (x) x

 oi

0

& (x) P, {p (x) + P (x) M (x) PO, (x) -P02 (x)

e

P x + RL, (x)

PO, (x) + P0, (x) - P0, (x) -x +

02

01

+ PA2 (x) O1

(x) + P02 (x) PЈ, (x) +

+ P0 ((x) -Pa2 (x)} 0.

Thus, if no errors occurred in channel 3, then zero syndrome is recorded in the shift register 37 of the syndrome analyzer 9, i.e.

sequence of type (1) goes to the first and second inputs of block 8 (adders 35, 36 modulo two), the third and fourth inputs of which are followed by 5 sequences M (x) -Pj, (x) and M (x) - Pg (x) from the outputs of the decoder 13 (in FIG. 2, the first and the second, the swarm of the outputs of the decoder 13 are combined). At the first and second outputs of block 8, sequences of the orthogonal unsystematic convolutional code M (x) P0, (x) and M (x) - P02 (x) are formed, which are fed to the inputs of the evaluator 7 estimates. At the outputs of the evaluator 7, the estimates of E1, which are orthogonal with respect to each information symbol of the sequence M (x), are generated at each operation cycle. These estimates come on

0

the switch 10, where an exception from the consideration at the input of the majority element 11 of the estimate Ej, the number of which coincides with the non-zero function Tj, is if T0 1. If there are no errors in the communication channel 3, therefore, and the first outputs of the analyzer 9 syndromes are not connected to the first control inputs of the commutator 10. The evaluation system EJ arrives at the majority element 11 (the function T0 0 is formed here), where at each work cycle the decision about the reliability of each character of the information sequence M ( x). Next, the information sequence M (x) is fed to the inputs of the decoders 12 and 13, where it advances with a clock frequency of different shift registers 39 and 43 and goes to the information output 20 of the device. In the course of this movement, the sequence M (x) using adders 40 - 42 and 44 modulo 25 is two converted at the output of the first decoder 12 into a sequence of the form

thirty

M (x) P (X) + P (x) M (x) P02 (x)

Ra, (x)

+ Ro,

V

(x)

which arrives at the control input of the calculator 6 syndrome, at the outputs of the second decoder 13 - in the sequence of the form

);

P (x)

(five)

which arrive at the inputs of block 8. The zero syndrome recorded in the shift register 37 of the analyzer 9 syndrome does not affect the remaining blocks of receiving side 2: the contents of the threshold counter 18 and the conditioner 17 of the time interval and the key element 15 do not change. Except Moreover, the error Error at output 21 is absent, and also each symbol of the information sequence M (x) along the feedback circuit from the output of the majority element 11 is fed to the control input of the calculator 7 estimates and the corresponding bits of the registers 31 and 32 shift, simultaneously signal E, from the first output

The evaluator 7 receives the control input of the analyzer 9 syndrome.

If there is a one-time or two-time error in the communication channel 3 over the length of the code limit imposed on the system, the codec works as follows. As in the previous case, the sequences of the form (2) enter the bits of the registers 28 and 29 of the shift of the calculator 6 syndrome and advance along them with a clock frequency DT. In the process of movement, these sequences are converted to

P | p02 (x) n (x)) + E ((x));

PgP0,

(x)) Pk2 (x) + Ez (x) P0, (x),

(6)

Jq -jq 5

five

modulo two, which are componentwise compared with the sequence M (x) P (X) + P (x) | received at the third input of the transmitter 6 from the output of the first decoder 12, i.e.

Ј (x) P, (x) Horn (x) + P (x) P01 (x) + 0 + M (x) (x) + p (x) n (x) .Pk ((x) + + E , (x) P0, (x) (x) - P „2 (x) + E2 (x) Horn (x) + M (x) -P01 (x) -P. ((x) + M (x) x

P ", x)

P "ja (x) #).

0

five

0

five

Thus, a non-zero syndrome is recorded in the shift register 37 of the syndrome analyzer 9. At the same time, sequences of the form (2) arrive at the first and second inputs of block 8, and from its outputs, sequences of an orthogonal non-systematic convolutional code are struck, affected by errors of the first or second multiplicity, which enter the evaluator 7. Suppose, for definiteness, a double error occurred in channel 3 of a connection, then, in accordance with the codec operation algorithm in switch 10, an exception is excluded from the consideration at the input of the major element 1t of that estimate E J, the number of which coincides with the non-zero function Tj formed in the logic unit 38 analyzer 9 syndrome and connected to the first control inputs of the switch

10 at T 1. Corrected system, about

The EI estimates are sent to the major element 11, where a decision is made on the reliability of each information symbol of the sequence M (x). Since the multiplicity of the error that occurred in channel 3 does not exceed the correcting ability of the Code, then at the receiving side 2 pro 1 |, the double error correction disappears and there is no transformation of the message. Further, the information sequence arrives at decoders 12 and 13, and the engine rods are by register registers.

39 and 43 shift, and in the sequential code goes to the output 20 of the codec. During this movement, the sequence M (x) with the help of adders

40-42 modulo two is converted at the output of the first decoder 12 into a sequence of the form (4), which goes to the third input of the syndrome calculator 6, at the outputs of the second decoder 13 into a sequence of the form (5), which goes to the inputs of block 8.

At the outputs of block 8, sequences of an orthogonal unsystematic convolutional code are formed, affected by a double error of the form (7)

M (x) P K ((x) + E, (x) + M (x) P j, (x); m (x) P kg (x) + Er (x) + M (x). Pa2 ( x).

(7)

 Since the syndrome recorded in the shift register 37 of the analyzer 9 syndrome is non-zero, the threshold counter 18 counts the number of errors received from the second output of the syndrome analyzer 9 at its counting input. At the same time, the signals oshi-, side from the second output of the analyzer 9 syndrome arrive at the first input of the key element 15, connecting the output of the generator 14-pulses to the input of the imaging device 17 time interval. Since, by assumption, the number of errors is small, the time interval former 17 counts the W ticks and generates a signal at its output before the threshold counter 18 overflows. The output of the time interval former 17 through the OR 16 element resets the former 17

five

0

five

0

five

0

five

0

five

the time interval, the threshold counter 18, and returns to the initial state the key element 15, having turned off the generator output 14 pulses from the input of the time generator 17. At the output of the threshold counter 18, the error signal does not appear, and reliable information from output 20 goes to the receiver. The next signal causes the repetition of these operations.

If a three or more fold error has occurred in the communication channel 3 on the length of the code limit, the correcting ability of the orthogonal non-systematic convolutional code included in the quasi-orthogonal non-systematic convolutional code is not sufficient for error correction. The work of the codec is similar. The threshold counter counts the number of error pulses from the second output of the analyzer 9 syndrome and generates an Error signal if it overflows in the time interval W. The Error signal is not generated only if the errors E, (x) and ЕЈ (Х) form a sequence , a code word of a quasi-orthogonal unsystematic convolutional code (i.e., errors translate the signals of one code word to another, transform it), or a sequence that differs from the code word of a quasi-orthogonal non-systematic convolutional code among the V 6 t0 bits. In the first case, the threshold counter 18 does not receive a single signal, while the second receives V signals that do not cause the counter to overflow.

Thus, in the proposed codec, noise immunity is improved.

Claims (1)

Invention Formula The codec of a non-systematic convolutional code containing a communication channel, on the transmitting side, is the first encoder whose input is the input of the codec; on the receiving side, the syndrome calculator, the first and second inputs of which are connected to the communication output of the same name, the output of the syndrome calculator is connected to the information the input of the syndrome analyzer, the first outputs of which are connected to the first control inputs of the switch, the outputs of which are connected to the inputs of the majority element, the first output of which is connected to the control input ohm rating evaluator, whose outputs are connected to the corresponding information inputs of the switch, the second output of the majority element is connected to the second control input of the switch, the control input of the syndrome analyzer is connected to the first output of the assessment calculator, characterized in that, in order to improve noise immunity, on the transmitting side the second encoder is entered, the first - the third outputs of the first encoder are connected to the inputs of the second encoder of the same name, the first and second outputs of which are connected to the channels of the same name At the receiving side, the first and second decoders, a pulse generator, a time interval generator, a key element, an OR element, a threshold counter and a unit of modulators two, the first and second inputs of which are connected to the same outputs of the first channel, the inputs of the first and second decoders are combined and -% ЗШ1СШ tҐ 0 five 0 five They are connected to the first output of the ME-AOJI element, the output of the first decoder is connected to the third input of the syndrome calculator, the second output of the second decoder is connected to the third and fourth inputs of the adder mode two, the second outputs of which are proud with the same reform circuit The inputs of the evaluator, the third BJIXOT, the second decoder is the first output of the codec, in the second output of the syndrome lizator is connected to the counting input of the threshold counter and the first control input of the echo S a, w - the generator a nmols.v is connected to the information input of the key element, the output of which is connected to the information input of the time interval formers, the output of which is connected to the first go of the OR, the output of which is connected to the control key input of the key element and the zeroing inputs of the time interval and a threshold counter, the output of which is connected to the second input of the OR element and is the second output of the codec. thirty