CN100388824C - Decoding Method for Adaptive Multi-Rate Speech Coding in Discontinuous Transmission System - Google Patents

Abstract

A decoding method for adaptive multi-rate speech coding under the discontinuous transmission system, including A1) according to the signal-to-noise ratio SNR output by the demodulator at the receiving end; A2) calculating the threshold value T and outputting the demodulator The soft-decision data mode bit IM_D(n) is limited to IM_D(n)'; A3) negatively polarizes one IM in the type identification mode of the frame to be decoded into IM'; A4) contains N A part or all of the IM_D(n)' of the bit is correlated with a part or all of the data of the pattern IM' corresponding to the frame, and the correlation R is calculated; A5) if R＞=T, then determine the output of the demodulator The frame has the same type as the frame to be decoded; otherwise, use another negatively polarized identification mode IM' to return to step A4), until all identification modes have been calculated and the correlation R is less than the threshold value T, then determine the demodulation The frame output by the converter is independent of the frame type corresponding to the mode IM used. By using the decoding method of the invention, the frame type of the adaptive multi-rate speech coding under the continuous transmission system can be effectively identified from the interfered data in the GSM system.

Description

Decoding Method for Adaptive Multi-Rate Speech Coding in Discontinuous Transmission System

technical field

The invention relates to communication technology, in particular to a decoding method for adaptive multi-rate speech coding under the discontinuous transmission system in the GSM system.

Background technique

After the Global system for Mobile communication (hereinafter referred to as "GSM") developed to PHASE II, it provided a new voice codec method, that is, Adaptive Multi-Rate Speech Coding (Adaptive Muti-Rate speech coding) codec, hereinafter referred to as "AMR"), this type of speech coding service can dynamically change the speech coding rate and channel coding rate according to the wireless environment. One of the code rates, so as to achieve the optimal combination of speech coding and channel coding to meet the instantaneous wireless channel conditions and local capacity requirements. AMR provides 8 bit rate options from 4.75kbits/s to 12.2kbits/s. For general speech frames, there is convolutional coding to implement FEC (Forward Error Correction Code). In discontinuous transmission (hereinafter referred to as DTX), some special frames do not have FEC. How to decode these frames is not described in the protocol. Therefore, it is impossible to implement AMR in DTX. decoding.

According to the GSM protocol, there are 9 special frames that need to be identified without FEC in AMR: full-rate SID_UPDATE, full-rate SID_FIRST frame, full-rate ON_SET frame, half-rate SID_UPDATE frame, half-rate SID_UPDATE_INH frame, half-rate SID_FIRST_P1, half-rate SID_FIRST_P2 frame, half-rate SID_FIRST_INH frame, half-rate ONSET frame. When sending these DTX frames of the above 9 kinds of full rate and half rate, an IM (identification mode) is included as shown in the table below.

frame type Identification Mode (IM) Full rate SID_UPDATE 0, 1, 0, 0, 1, 1, 1, 1, 0 Repeat to 212 bits Full rate SID_FIRST frame 0, 1, 0, 0, 1, 1, 1, 1, 0 Repeat to 212 bits Full rate ON_SET frame ic1 Repeat to 228bit Half-rate SID_UPDATE frames 1, 0, 1, 1, 0, 0, 0, 0, 1 Repeat to 212 bits Half-rate SID_UPDATE_INH frames 0, 1, 0, 0, 1, 1, 1, 1, 0 repeat to 212bit Half rate SID_FIRST_P1 0, 1, 0, 0, 1, 1, 1, 1, 0 Repeat to 212 bits Half-rate SID_FIRST_P2 frames ic repeated to 114 bits Half-rate SID_FIRST_INH frames 1, 0, 1, 1, 0, 0, 0, 0, 1 Repeat to 212 bits Half-rate ONSET frame ic1 Repeat to 114bit

Due to encoding, when actually decoding, for SID_UPDATE_INH and SID_FIRST_INH, its recognition mode has become 1, 1, 1, 0, 0, 1, 0, 1, 1, 0, 0, 0, 1 , 1, 0, 1, 0, 0 and 0, 0, 0, 1, 1, 0, 1, 0, 0, 1, 1, 1, 0, 0, 1, 0, 1, 1; in the table ic and ic1 refer to one of the following four 16-bit modes. The four 16-bit identification modes are expressed in hexadecimal: 0x530f, 0x3eb8, 0x8863, 0xe5d4. For example, 0x530f indicates that the identification mode is 0, 1, 0, 1, 0, 0, 1, 1, 0, 0, 0, 0, 1, 1, 1, 1, for the frames represented by ic and ic1 in the table, it needs to perform 4 matches to find the maximum value. No matter which mode is adopted, the technical problem faced is the same, and the original data format needs to be identified from the data disturbed by noise. The prior art does not provide a solution for the identification or decoding of these frame types.

Contents of the invention

The technical problem to be solved by the present invention is how to decode the data output from the demodulator at the receiving end to find out which of the following frame types the frame type actually sent by the sender is: full rate SID_UPDATE, full rate SID_FIRST frame, full rate ON_SET frame, half-rate SID_UPDATE frame, half-rate SID_UPDATE_INH frame, half-rate SID_FIRST_P1, half-rate SID_FIRST_P2 frame, half-rate SID_FIRST_INH frame, half-rate ONSET frame. In other words, the object of the present invention is to provide a decoding method for adaptive multi-rate speech coding under the discontinuous transmission system.

The above-mentioned technical problem of the present invention is solved like this, constructs a kind of decoding method to the self-adaptive multi-rate speech coding under discontinuous transmission system, comprises the following steps:

A1) Calculate the threshold value T according to the signal-to-noise ratio SNR output by the demodulator at the receiving end;

A2) according to the signal-to-noise ratio SNR, the soft-decision data mode bit IM_D(n) limit of the demodulator output is IM_D(n)';

A3) Negatively polarize one of the identification modes IM of the frame type to be decoded into IM';

A4) A part or all of the IM_D(n)' containing N bits after the clipping is correlated with a part or all of the data of IM', and the correlation degree R is calculated;

A5) If R＞=T, then determine that the frame type that the demodulator outputs has the same type as the frame to be decoded; Otherwise, return to step A4) with another identification mode IM' of negative polarization processing, until all identification modes If both are calculated and the correlation degree R is smaller than the threshold value T, it is determined that the frame output by the demodulator has nothing to do with the frame type corresponding to the mode IM used.

Wherein, the negative polarity of the identification mode IM of the frame type to be decoded is changed to IM', which is to convert the bit "0" into "1" and "1" into "-1" for each bit of the IM, The negative polarity recognition pattern IM' is obtained.

Wherein, calculating the threshold value T according to the signal-to-noise ratio SNR output by the demodulator at the receiving end includes the following steps:

1) Calculate the reliability p corresponding to the SNR

p=10 ^SNR/20 /17

2) Limit the calculated reliability p, that is, take 0.25 when it exceeds 0.25;

3) Calculate the threshold T=N*p*127 according to the reliability p and the number of bits N of IM_D(n).

Wherein, according to the signal-to-noise ratio SNR, the soft-decision data mode bit IM_D(n) clipping of the demodulator output is IM_D(n)', comprising the following steps:

1) Set the highest value jitter of the allowable range Δ=p*128;

2) If IM_D(n)>128+Δ, adjust IM_D(n)'=128+Δ;

3) If IM_D(n)<128-Δ, adjust IM_D(n)'=128-Δ, where p is the reliability corresponding to SNR.

Wherein, the correlation operation is a multiplication operation.

In the above method, the corresponding relationship between the signal-to-noise ratio SNR and the threshold value T is calculated and stored in a table in advance, and the threshold value T is obtained by looking up the table through the SNR each time.

In the above method, the type identification modes include the following nine types:

1) The identification mode of the full-rate SID_UPDATE frame is 0, 1, 0, 0, 1, 1, 1, 1, 0 repeated to 212 bits;

2) The identification mode of the full-rate SID_FIRST frame is 0, 1, 0, 0, 1, 1, 1, 1, 0 repeated to 212 bits;

3) The full-rate ON_SET frame has four identification modes, which are formed by repeating 0x530f, 0x3eb8, 0x8863 or 0xe5d4 to 228 bits;

4) The identification pattern of the half-rate SID_UPDATE frame is 1, 0, 1, 1, 0, 0, 0, 0, 1 repeated to 212 bits;

5) The identification mode of the half-rate SID_UPDATE_INH frame is 0, 1, 0, 0, 1, 1, 1, 1, 0 repeated to 212 bits;

6) The identification pattern of the half-rate SID_FIRST_P1 frame is 0, 1, 0, 0, 1, 1, 1, 1, 0 repeated to 212 bits;

7) The half-rate SID_FIRST_P2 frame has four identification modes, which are formed by repeating 0x530f, 0x3eb8, 0x8863 or 0xe5d4 to 114 bits;

8) The identification pattern of the half-rate SID_FIRST_INH frame is 1, 0, 1, 1, 0, 0, 0, 0, 1 repeated to 212 bits;

9) The half-rate ONSET frame has four identification modes, which are formed by repeating 0x530f, 0x3eb8, 0x8863 or 0xe5d4 to 114 bits.

Implementing the method for decoding AMR DTX frames in the GSM system provided by the present invention can effectively identify the type of AMR DTX frames used in GSM from the data polluted by noise.

Detailed ways

In order to illustrate the object of identification or decoding of the method of the present invention, the special frames and identification modes that need decoding/recognition of the present invention are described as follows:

1) Full rate SID_UPDATE frame, the identification pattern is 0, 1, 0, 0, 1, 1, 1, 1, 0 repeated to 212 bits;

2) Full rate SID_FIRST frame, the identification pattern is 0, 1, 0, 0, 1, 1, 1, 1, 0 repeated to 212 bits;

3) Full-rate ON_SET frame, one of four modes, consisting of 0x530f or 0x3eb8 or 0x8863 or 0xe5d4 repeated to 228 bits;

4) Half-rate SID_UPDATE frame, the identification pattern is 1, 0, 1, 1, 0, 0, 0, 0, 1 repeated to 212 bits;

5) Half-rate SID_UPDATE_INH frame, the identification pattern is 0, 1, 0, 0, 1, 1, 1, 1, 0 repeated to 212 bits;

6) Half-rate SID_FIRST_P1 frame, the identification pattern is 0, 1, 0, 0, 1, 1, 1, 1, 0 repeated to 212 bits;

7) Half-rate SID_FIRST_P2 frame, one of four modes, consisting of 0x530f or 0x3eb8 or 0x8863 or 0xe5d4 repeating all 114 bits;

8) Half-rate SID_FIRST_INH frame, the identification pattern is 1, 0, 1, 1, 0, 0, 0, 0, 1 repeated to 212 bits;

9) Half-rate ONSET frame, one of four modes, consisting of 0x530f or 0x3eb8 or 0x8863 or 0xe5d4 repeated to 114 bits.

It should be noted that frames with the same recognition pattern are placed in different positions, and it is assumed that the data has been extracted from the correct position.

Below, each step of identifying or decoding the above-mentioned frames of the discontinuous transmission according to the method of the present invention is described in detail as follows:

Step 1: Calculate the threshold value T according to the signal-to-noise ratio (SNR) output by the demodulator at the receiving end.

The judgment threshold of the present invention is associated with the noise. In other words, the present invention determines different thresholds based on different noises, that is, under the premise that the threshold is associated with the noise, at the receiving end of the above nine special frames, the demodulator can output a The signal-to-noise ratio (SNR) expressed in decibels, the method of the present invention to determine the threshold is exactly to calculate the threshold value T under the signal-to-noise ratio (SNR) according to the signal-to-noise ratio (SNR) obtained by demodulation. The process is as follows:

1) Calculate the reliability corresponding to SNR (this reliability is considered to be about 0.25 when SNR13dB):

p=10 ^SNR/20 /17

2) Limit the calculated reliability p, that is, take 0.25 when p exceeds 0.25;

3) Calculate the threshold T according to the reliability p:

T=N*p*127, N is the total number of bits calculated

In the specific implementation, the calculation of the threshold can also pre-calculate the relationship between the SNR and the threshold, and set it into a table. Every time it is needed, as long as the SNR of the demodulation output is used as an index to check this table, the corresponding Threshold T. This reduces the time overhead of computing inventions.

Step 2: Calculate Relevance

The demodulator outputs the soft decision data of the frame, which is generally a string of 8-bit signed integers with negative polarity. -128 means that the transmitted symbol is closest to "1", 127 means that the transmitted symbol is closest to "0", and 0 means "1" and "0" are equally probable; the details of calculating the correlation are as follows:

2-1) Limiting the soft decision data mode IMD(n), that is, limiting the soft decision data mode bit IM_D(n) output by the demodulator according to the signal-to-noise ratio SNR to IM_D(n)'; specifically Assume that the highest value jitter of the allowable amplitude is: Δ=p*127

If IM_D(n)＞128+Δ IM_D(n)’＝128+Δ

If IM_D(n)＜128-Δ IM_D(n)’＝128-Δ

Otherwise, it remains unchanged, that is, IM_D(n)'=IM_D(n).

2-2) Negatively polarize the identification mode IM of the frame type to be decoded into IM', that is, the "1" in the identification mode IM of the frame type to be decoded is changed to "-1", and "0" is changed to "1" ";

2-3) Multiply part or all of the IM_D(n)' containing N bits after clipping with part or all of the data of the identification mode IM' corresponding to the frame (<=N) for correlation Calculate the degree of correlation R=IM_D(n)'*IM';

Step 3: Determine the DTX frame type according to the threshold and correlation

Compare and judge according to the correlation calculation result R obtained above and the threshold T obtained by calculation or table lookup. The judgment method is as follows: if R>=T, it is judged that the data matches the compared pattern, so that the frame type can be determined as The frame corresponding to the mode; otherwise, the judgment is negative, that is, the frame type is unlikely to be the frame corresponding to the mode IM used.

The following illustrates the processing flow of a general decoding based on the method of the present invention

1) Negatively polarize the IM to be identified in advance, and the identification modes include:

0, 1, 0, 0, 1, 1, 1, 1, 0; 1, 0, 1, 1, 0, 0, 0, 0, 1; 1, 1, 1, 0, 0, 1, 0, 1,1,0,0,0,1,1,0,1,0,0; 0,0,0,1,1,0,1,0,0,1,1,1,0,0, 1, 0, 1, 1; and 0x530f, 0x3eb8, 0x8863, 0xe5d4 in hexadecimal notation.

2) Look up the table according to the signal-to-noise ratio SNR to obtain the threshold value T;

3) According to the signal-to-noise ratio SNR, limit the soft decision data mode bit IM_D(n) output from the demodulator, and the limit is IM_D(n)';

4) Calculate correlation R=IM_D(n)'*IM';

5) If R≥T, judge that the frame type corresponding to the current IM is the frame type of the output data, and end the decoding procedure; otherwise, use another IM' to return to step 4), until all IM's have been calculated, and at this time the current It cannot be decoded or recognized under the threshold value.

Taking the sent data frame as a full-rate SID_UPDATE as an example, an example of the decoding process of a specific code in the above processing flow is described below. The assumed data frame is a full-rate SID_UPDATE, and its corresponding identification mode IM is 0, 1, 0, 0, 1, 1, 1, 1, 0 repeated to 212 bits; it is assumed that all 9 identification modes IM have been performed in advance Negatively polarized and stored; assuming that the received part of the data is 9 bit patterns IM_D (9), each bit pattern is a binary number with 8 bits of signed bits, and the following decimal representation of IM_D (9) is: 39 - 28 49 26 -46 -43 -42 -31 36;

Assuming that the signal-to-noise ratio SNR of the demodulator output is 10dB, calculate p=0.186 according to the formula, N=9 (assuming calculation of 9 bits), calculate the threshold T=213 (integer);

The mode bit IM_D(9) is limited as follows:

39->32 49->32 46->32 43->32 42->32 36->32, the rest remain unchanged;

IM_D(9)'32 -28 32 26 -32-32-32-31 32 after getting the limit

Negatively polarize the recognition mode IM to IM'=1, -1, 1, 1, -1, -1, -1, -1, 1

Calculate the correlation R

=IM_D(9)'*IM(9)'

＝32*1-28*(-1)+32*1+26*1-32*(-1)-32*(-1)-32*(-1)-31*(-1)+32* 1

=277

Since R>T, it is judged that the received frame type is a full-rate SID_UPDATE frame.

Claims (7)

1. a decoding method to adaptive multi-rate speech coding under the discontinuous transmission system, it is characterized in that, comprises the following steps: A1) Calculate the threshold value T according to the signal-to-noise ratio SNR output by the demodulator at the receiving end; A2) According to the signal-to-noise ratio SNR, the soft-decision data mode bit IM_D(n) limit output by the demodulator is IM_D(n)'; A3) Negatively polarize one of the type identification modes IM of the frame to be decoded into IM'; A4) A part or all of the IM_D(n)' containing N bits after the clipping is correlated with a part or all of the data of IM', and the correlation degree R is calculated; A5) If the correlation degree R is greater than or equal to the threshold value T, then determine that the frame type output by the demodulator has the same type as the frame to be decoded, otherwise use another negatively polarized identification mode IM' to return to step A4) , until all the identification modes are calculated and the correlation R is less than the threshold value T, then it is determined that the frame output by the demodulator has nothing to do with the frame type corresponding to the mode IM used. 2. The method according to claim 1, characterized in that, the step of depolarizing the identification mode IM of the frame type to be decoded into IM' is to change the bit "0" to "1" for each bit forming IM ", "1" into "-1", and the negative polarity identification pattern IM' is obtained. 3. according to the described method of claim 1, it is characterized in that, the step of calculating threshold value T according to the signal-to-noise ratio SNR that receiver demodulator outputs comprises the following steps: 1) Calculate the reliability p corresponding to the SNR p=10 ^SNR/20 /17 2) Limit the calculated reliability p, that is, take 0.25 when it exceeds 0.25; 3) Calculate the threshold according to the reliability p and the number of digits N of IM_D(n) T=N*p*127. 4. according to the described method of claim 1, it is characterized in that, the soft decision data mode bit IM_D (n) clipping of demodulator output is the step of IM_D (n) ' according to said signal-to-noise ratio SNR, comprising the following steps : 1) Set the highest value jitter of the allowable range Δ=p*128; 2) If IM_D(n)>128+Δ, adjust IM_D(n)'=128+Δ; 3) If IM_D(n)<128-Δ, adjust IM_D(n)'=128-Δ, where p is the reliability corresponding to SNR. 5. The method according to claim 1, wherein the correlation operation is a multiplication operation. 6. The method according to claim 3, characterized in that the corresponding relationship between the signal-to-noise ratio (SNR) and the threshold value T is calculated and stored in a table in advance, and the threshold value T is obtained by looking up the table through the SNR each time. 7. The method according to any one of claims 1-6, wherein the type identification modes include the following nine types: 1) The identification mode of the full-rate SID_UPDATE frame is 0, 1, 0, 0, 1, 1, 1, 1, 0 repeated to 212 bits; 2) The identification mode of the full-rate SID_FIRST frame is 0, 1, 0, 0, 1, 1, 1, 1, 0 repeated to 212 bits; 3) The full-rate ON_SET frame has four identification modes, which are formed by repeating 0x530f, 0x3eb8, 0x8863 or 0xe5d4 to 228 bits; 4) The identification pattern of the half-rate SID_UPDATE frame is 1, 0, 1, 1, 0, 0, 0, 0, 1 repeated to 212 bits; 5) The identification mode of the half-rate SID_UPDATE_INH frame is 0, 1, 0, 0, 1, 1, 1, 1, 0 repeated to 212 bits; 6) The identification pattern of the half-rate SID_FIRST_P1 frame is 0, 1, 0, 0, 1, 1, 1, 1, 0 repeated to 212 bits; 7) The half-rate SID_FIRST_P2 frame has four identification modes, which are formed by repeating 0x530f, 0x3eb8, 0x8863 or 0xe5d4 to 114 bits; 8) The identification pattern of the half-rate SID_FIRST_INH frame is 1, 0, 1, 1, 0, 0, 0, 0, 1 repeated to 212 bits; 9) The half-rate ONSET frame has four identification modes, which are formed by repeating 0x530f, 0x3eb8, 0x8863 or 0xe5d4 to 114 bits.