WO2000035138A1

WO2000035138A1 - Adaptive concatenated channel coding method

Info

Publication number: WO2000035138A1
Application number: PCT/DE1999/003845
Authority: WO
Inventors: Egon Schulz
Original assignee: Siemens Aktiengesellschaft
Priority date: 1998-12-04
Filing date: 1999-12-01
Publication date: 2000-06-15
Also published as: DE19856085A1

Abstract

The inventive channel coding method which has at least two interleaved codes is characterized in that the concatenated channel coding method is adaptive. According to the invention, the concatenated channel coding method uses a block code as the outer code and a convolutional code as the inner code. Preferably, a code word is subjected to block coding in m blocks B1,...,Bm with m ≥ 1 and the resulting blocks are encoded with generator polynominals g1,...,gn, n ≥ 1, in convolutionally coded blocks g1(B1),..g1(Bm),...,gn(B1),...,gn(Bm), with n or m > 1.

Description

description

Adaptive chained channel coding method

The invention relates to a concatenated channel coding method in which the information to be transmitted is protected against interference by executing at least two codes in series. Mobile phones are the preferred area of application.

In mobile radio systems, error correction methods for increasing the transmission security are used on the transmission interface between base station and mobile station for the transmission of data and signaling. The methods for channel coding are described, for example, by a

Standard, as is the case with GSM (global system for mobile communication) or DECT (digital enhanced cordless telephony). ARQ (Automatic Repeat Request) or Forward Error Correction with a fixed number of redundancy symbols are mainly used for the transmission of data. With the ARQ method, the information to be transmitted is requested again from the transmitter in the event of a faulty transmission.

In order to improve the transmission quality, for example in the case of mobile radio channels, but also in the case of satellite channels, interlinked channel coding methods are used, among other things. In the case of chained channel coding methods, the information to be transmitted is protected against interference by the implementation of at least two codes behind one another. Coding adds redundancy to the information. This redundancy is formed by means of a coding rule from the information to be coded.

In the case of chained channel coding methods, the information is first protected, for example by a block code. The information sequence coded in this way by means of a block code finally, their redundancy is then fed to a further encoder, most of the time a convolutional encoder. The first coding level is referred to as the outer code (outer code) and the following second coding level is referred to as the inner code (inner code). In the example given, the block code is the outer code and the convolutional code is the inner code.

A data transfer was therefore carried out according to the following scheme:

Data Source

Outer code 1st coding level as block coding

Inner code 2nd coding level like convolution coding

channel

Decoding decoder αer 2nd coding level of the inner code (e.g. Viterbi algorithm)

Decoding Decoder of the 1st coding level of the outer code (e.g. Berlekamp-Massey-

Algorithm)

Data sink

Adaptive channel coding methods have been developed for both block codes and convolutional codes. With such adaptive channel coding methods, the data throughput can be increased significantly. The information on the transmission side is encoded with a corresponding block or convolutional code. The information sequence coded in this way, hereinafter referred to as the code word, consists of the information to be transmitted and the redundancy formed from the information by means of the coding regulation of the code. The principle of the adaptive channel coding method mentioned above is described below. With an adaptive channel coding method, a code word is not transmitted all at once, but rather, for example, in blocks. As an example, the transmission of a code word is carried out in three blocks, ie block B1, block B2 and block B3. In this example this means:

Codeword = Block B1 Block B2 Block B3

The first block B1 can contain the information to be transmitted along with a few CRC bits (redundancy). These CRC bits serve on the one hand to correct a few errors and on the other hand to recognize whether errors have occurred during the transmission. For example, blocks B2 and B3 can only contain redundancy. Depending on the transmission strategy, information can also be present in blocks B2 and / or B3. In a first transmission attempt, the first block B1 of the code word is transmitted. The receiver tries to decode the first block B1. If the decoding is successful, the receiver informs the transmitter that the first block B1 has been successfully transmitted and the transmitter therefore no longer has to transmit, for example, the remaining two blocks B2 and B3 if, as in the example above, there is no information contain. The transmitter can therefore continue with the next code word. As a result, the data throughput is increased since, with a good transmission quality of the channel, it is not necessary to transmit the entire redundancy of the code word, since with good transmission quality there are little or no transmission errors.

If, on the other hand, the decoding of the first block B1 of the first transmission attempt was unsuccessful, the receiver requests the second block B2. With the help of the first and second

Blocks B1, B2 attempts the receiver to decode the information. In the event of successful decoding of the information tion from the first and second blocks B1, B2, the receiver notifies the transmitter of the successful decoding and the transmitter no longer sends the third block B3.

If the decoding of the second transmission attempt was unsuccessful, the receiver requests the third block B3 and tries to decode information by means of the three received blocks B1, B2 and B3.

Is the decoding also with the completely received code word, i.e. If the three blocks B1, B2 and B3 are still unsuccessful, then either the transmission process is started again from the beginning or the receiver indicates which block should be sent again due to the transmission quality and / or reception field strength.

Such adaptive channel coding methods have the advantage that the entire code word does not have to be sent if the transmission quality is good. In the event of poorer transmission quality, only portions of the code word which, for example, were severely disturbed, have to be sent again in order to reconstruct the information at the receiver. As a result, the data throughput is increased, both with good transmission properties and partly with severely disturbed channels, since the transmitter is selectively informed of which parts of the code word may have to be sent again.

Furthermore, there is the possibility of receiving the portions received by the recipient, for example by means of a maximum ratio

Combinmg to combine if a portion was requested twice or more by the broadcaster.

Another strategy of the adaptive channel coding method is a transmission of the respective first block from a multiplicity of code words. The decoder on the receiver side tries to use the first block of codewords to decode the information it carries. If the decoder is not able to do this, it requests the respective second and later the third block from the non-decoded code words.

Other transmission strategies are also possible. In the first step, the first block can be transmitted from all code words for the first time, and if the sender does not receive a stop signal for certain blocks from the receiver, the transmitter sends the second block of all code words.

The invention is therefore based on the object of providing a method of channel coding in which the throughput rate is further increased and the transmission quality is improved.

The object is achieved by the features of the method according to claim 1 and the mobile radio system with the features of claim 15. Preferred embodiments of the invention are the subject of the dependent claims.

According to the invention, the chained channel coding method, which consists of at least two codings chained together, is adaptive. With such an adaptive concatenated channel coding method, the redundancy is variable. The concatenated channel coding method preferably uses block coding as the outer code and convolutional coding as the inner code.

Furthermore, the convolutional coding of the channel coding method according to the invention has a number n> 1 generator polynomials gl, ..., gn. A code word is subjected to block coding mm blocks Bl, ..., Bm with m> 1 and the resulting blocks are convolutionally coded blocks gl (Bl), ..gl (Bm), ..., gn (Bj ), ..., gn (Bm) coded, where either m or n> 1 must apply. In a preferred embodiment, both m and n are> 2. A transmission strategy could be such that the first convolutionally coded block gl (Bl) is transmitted and, in the event of an unsuccessful decoding, the next block is requested until either the existing blocks have been successfully decoded or the last block gn (Bm) is requested has been. In the case of a complete transmission of all blocks and a failure of the decoding, the method starts the transmission again with the first convolutionally coded block gl (B1) of the code word.

Furthermore, for decoding, those convolutionally coded blocks gk (Bj) of the same block index j e {l, .., m} can be combined with k e {l, ..., n} and decoded.

In another transmission strategy for a

Quantity of code words to be transmitted first transmit all the first convolutionally coded blocks, and the receiver requests further blocks for those code words that could not be decoded. A quality measure can be defined for "bad" blocks, for example via the reception level, so that in the event of a failure of the decoding only the blocks whose quality measure is below a predetermined threshold are retransmitted.

The invention is described below with the aid of an example, the coding and decoding being carried out both on the network side and in mobile stations of a mobile radio system, depending on the transmission direction. For the construction of various mobile radio systems, reference is made to P. Jung, "Analysis and design of digital mobile radio systems"; 1997; ISBN 3-519-06190-2.

In order to further increase the throughput rate or to further improve the transmission quality, an adaptive concatenated channel coding scheme is used in accordance with the invention. An example consisting of a reed _S olomon code and a convolution code considered. A derarti ^¬ ges example is not intended to be limiting.

The symbols of the Reed-Solomon code consist of several bits, and therefore Reed-Solomon codes have the property that burst errors are also corrected, which frequently occur, for example, in mobile radio. The adaptive concatenated channel coding method according to the invention is illustrated in the example below.

An exemplary Reed-Solomon code with the parameters (85, 81) has 81 information points and 4 redundancy symbols, each symbol consisting of 8 bits. The code has a code rate of approximately 1. The code word (85, 81) is referred to below as block B1. If more than one error occurs in the first transmission attempt (SV), a further 85 redundancy symbols (block B2) are transmitted in the second transmission attempt. It is now possible to correct a further 43 errors. The code rate is then about 1/2. In a third attempt to send, a further 85 redundancy symbols (block B3) can be sent, so that a total of 86 errors can be corrected. This results in a code rate of approx. 1/3. It is also possible to send not only redundancy in further transmission attempts, but also information symbols, as was also described, for example, in the introduction of other block codes.

According to the invention, the block-coded symbols are additionally coded by means of convolutional code generator polynomials gk with ke {l, ..., n}.

In the example, a convolutional code with the code rate R = 1/4 is used as the convolutional code. The convolutional code consists of the 4 code generator polynomials gl, g2, g3 and g4. The transfer strategy could look like this:

The blocks listed above are defined by the following assignment:

Block 1 gl (Bl) Block 2 gl (B2) Block 3 gl (B3) Block 4 g2 (Bl) Block 5 g2 (B2) Block 6 g2 (B3) Block 7 g3 (Bl) Block 8 g3 (B2) Block 9 g3 (B3) block 10 g4 (Bl) block 11 g (B2) block 12 g4 (B3) After each transmission attempt, an attempt is made to decode the transmitted information on the basis of the block received. A

Decoding strategy could proceed according to the following scheme

(VA = Viterbi algorithm, BMA = Berlekamp-Massey algorithm):

Step 1: Decode block 1 with gl using VA. Result: Block lf Block lf decode with the BMA. If successful: End the decoding procedure for the code word received. If unsuccessful, continue with step 2.

Step 2: request block 2.

Decode block 2 with gl using VA.

Result: Block 2f.

Decode block lf and block 2f with the BMA.

If successful: End the decoding procedure for the code word received.

If unsuccessful, continue the

Procedure with step 3.

Step 3: request block 3. Decode block 3 with gl using VA.

Result: Block 3f.

Decode block lf, block 2f, block 3f with the BMA.

If successful: End the decoding procedure for the code word received.

If unsuccessful, continue the

Procedure with step 4.

Step 4: request block 4. Decode block 1 and block 4 with gl and g2 using VA. Result: Block 14f. Decode block 14f, block 2f and block 3f with the BMA.

If successful: End the decoding procedure for the code word received. If unsuccessful, continue the

Procedure with step 5.

Step 5: request block 5.

Decode block 2 and block 5 with gl and g2 using VA.

Result: Block 25f

Decode block 14f, block 25f and block 3f with the BMA.

If successful, end the decoding procedure for the received code word.

If unsuccessful, continue the

Procedure with step 6.

Step 6: request block 6. Decode block 3 and block 6 with gl and g2 using VA.

Result: Block 36f

Decode block 14f, block 25f and block 36f with the BMA. If successful, end the decoding procedure for the received code word.

If unsuccessful, continue the

Procedure with step 7.

Step 7: request block 7.

Decode block 1, block 4 and block 7 with gl, g2 and g3 using VA.

Result: block 147f

Decode block 147f, block 25f and block 36f with the BMA.

If successful, end the decoding procedure for the received code word. If unsuccessful, continue with step 8.

Step 8: request block 8. Decode block 2, block 5 and block 8 with gl, g2 and g3 using VA. Result: block 258f

Decode block 147f, block 258f and block 36f with the BMA. If successful, end the decoding procedure for the received code word. If unsuccessful, continue with step 9.

Step 9: Request block 9.

Decode block 3, block 6 and block 9 with gl, g2 and g3 using VA.

Result: block 369f

If successful, end the decoding procedure for the received code word.

If unsuccessful, continue the

Procedure with step 10.

Step 10: Request block 10. Decode block 1, block 4, block 7 and block 10 with gl, g2, g3 and g4 using VA.

Result: block 14710f

Block 14710, Block 258f and Block 369f with the

Decode BMA. If successful, end the decoding procedure for the received code word.

If unsuccessful, continue the

Procedure with step 11.

Step 11: request block 11.

Decode block 2, block 5, block 8 and block 11 with gl, g2, g3 and g4 using VA. Result: block 25811f

Decode block 14710f, block 25811f and block 36f with the BMA.

If successful, end the decoding procedure for the received code word. If unsuccessful, continue with step 12.

Step 12 request block 12.

Decode block 3, block 6, block 9 and block 12 with gl, g2, g3 and g4 using VA.

Result: block 36912f

Decode block 14710, block 25811f and block 36912f with the BMA.

If successful, end the decoding procedure for the received code word.

If unsuccessful, request the code word again.

(FURTHER)

The table below shows which fold-decoded blocks are available after the individual transmission attempts:

Block Bl 1. SV: Block lf 4. SV: Block 4f, Block 14f 7. SV: Block 7f, Block 17f, Block 47f, Block 141 f

10. SV: Block lOf, Block HOf, Block 410f, Block 710f,

Block 1410f, Block 1710f, Block 4710f, Block 14710f

Block B2

2nd SV Block 2f

5. SV Block 5f, Block 25f

8th . SV Block 8f, Block 28f, Block 58f, Block 258f

11. SV Block llf, Block 211f, Block 511f, Block 811f,

Block 2511f, Block 2811f, Block 5811f, Block 25811f Block B3:

3rd SV: Block 3 f

6. SV: block βf, block 36f

9. SV: Block 9f, Block 39f, Block 69f, Block 369f

12th SV: block 12f, block 312f, block 612f, block 912f,

Block 3612f, Block 3912f, Block 6912f, Block 36912f

Using the convolutionally decoded blocks from the list above, another strategy for decoding the block code can be used, which looks as follows:

It is determined from the convolutionally decoded blocks of the first and fourth transmission attempts by comparison which received symbols have a high probability of being correctly received. One compares e.g. the 85 symbols from block 1f with block 4f and block 14f. The symbols that do not have the same value in all three blocks are rated as erasure. You get a new block lf *. The blocks Block lf *, Block 2f and Block 3f are then decoded with the BMA. If this is not successful, then a block 2f * is determined analogously to the method explained in the preceding and then decoded accordingly with the BMA.

1 shows a comparison between the transmission rate DR as a function of the channel error rate KFR. Curve I shows the result of the known adaptive channel coding method with block coding after the first and second transmission attempts, while curve II shows the result of the adaptive concatenated channel coding method according to the invention also after the first and second transmission attempts. It can be clearly seen that the adaptive chained channel coding method according to the invention has better performance than the simple adaptive channel coding method. Another strategy for adaptive channel coding is the assignment of a quality measure for each block received. Such a quality measure can be derived, for example, in a known manner from the received level received. For example, if the code word cannot be decoded after the first three transmission attempts in the example above, then if block i (i = 1, 2, 3) has been assigned a poor quality measure, the block (i + k-3) with ke ( 1, 2, ...) requested. Step 4 would then look like this:

- Block 2 is of poor quality - Step 4: Request block 5 (k = 1).

Decode block 1 and block 5 with gl and g2 with VA. Result: Block 15f.

Decode block lf, block 3f, block 15f with BMA.

If successful: finish decoding

If unsuccessful: request block (i + k-1), where block i is the worst quality block.

Furthermore, the invention is not limited to the combination of block code and convolutional code. The following cases can also be treated using the adaptive procedure described above:

Outer code = turbo code Inner code = convolutional code

Outer code = turbo code Inner code = turbo code

Outer code = block code Inner code = turbo code

Outer code = block code Inner code = block code

Claims

claims

1. Channel coding method with at least two codings linked together, characterized in that the linked channel coding method is adaptive.

2. The method according to claim 1, characterized in that the concatenated channel coding method uses a block coding as the outer code and a convolutional coding as the inner code.

3. The method according to claim 2, characterized in that the convolutional coding has a number n> 1 generator polynomials gl, ..., gn, a code word being subjected to block coding in m blocks Bl, ..., Bm with m> 1 and the resulting blocks are encoded with the generator employnoms in convolutionally coded blocks gl (Bl), .. gl (Bm), ..., gn (Bl), ..., gn (Bm), whereby either m or n> 1 must apply.

4. The method according to claim 4, characterized in that applies to both m and n> 2.

5. The method according to claim 3 or 4, characterized in that the first convolutionally coded block gl (Bl) is transmitted and in the event of unsuccessful decoding the next

Block is requested until either a successful decoding has taken place or the last block gn (Bm) has been requested.

6. The method according to claim 5, characterized in that in the case of a complete transmission of all blocks and a failure of the decoding, the transmission starts again with the first convolutionally coded block gl (B1) of the code word.

7. The method according to any one of claims 5 or 6, characterized in that for decoding convolutionally coded blocks gk (Bj) of the same block index each {l, .., m} with ke {l, ..., n} are combined and used .

8. The method according to claim 4, characterized in that for a set of code words to be transmitted first all convolutionally coded blocks are transmitted, and the receiver requests further blocks for those code words that could not be decoded.

9. The method according to claim 1, characterized in that a turbo code is used as the outer code and a convolutional code is used as the inner code.

10. The method according to claim 1, characterized in that a turbo code is used as the outer code and a turbo code as the inner code.

11. The method according to claim 1, characterized in that a block code is used as the outer code and a turbo code as the inner code.

12. The method according to claim 1, characterized in that a block code is used as the outer code and a block code is used as the inner code.

13. The method according to any one of the preceding claims, characterized in that a quality measure is assigned to the blocks received, and a retransmission of a block is then requested if its quality measure is below a predetermined measure.

14. The method according to claim 13, characterized in that the reception level of a block is used to derive the quality measure.

15. Mobile radio system for performing the method according to claim 1.