CN1758579A

CN1758579A - Method and device of HARQ

Info

Publication number: CN1758579A
Application number: CN 200510105204
Authority: CN
Inventors: 曹一卿; 古建; 奇琳; 杨大成
Original assignee: Beijing University of Posts and Telecommunications
Current assignee: Beijing University of Posts and Telecommunications
Priority date: 2005-09-26
Filing date: 2005-09-26
Publication date: 2006-04-12

Abstract

The present invention provides a Hybrid Automatic Repeat Request (HARQ) method. The invention is applied in a wireless communication environment, provides a flexible, efficient, and low-complexity communication mode, can significantly improve system transmission efficiency without obviously increasing system complexity, is easy to implement, and has a wide application range. Based on the difference in the degree distribution of Low Density Parity Check (LDPC) code nodes, the present invention fully considers the different degrees of bit protection corresponding to different degree distributions, and provides a novel and practical hybrid automatic request retransmission scheme, And named degree distribution based hybrid automatic request retransmission scheme. The scheme can significantly improve the system performance without significantly increasing the system complexity by using a small bandwidth.

Description

A kind of method and apparatus of mixed automatic repeat request

Technical field

What the present invention relates to is h-arq (HARQ) method and apparatus in a kind of wireless way system, particularly relate to and adopt low-density checksum (Low Density Parity Check, LDPC) the mixed automatic repeat request technology of coded data transmission in the wireless communication system.

Background technology

Current radio communication development is rapid, and wireless communication user improves day by day to the needs of data business, and various high-speed data service schemes occur one after another, and nearly all scheme all requires system to have the characteristic of low-residual error rate and high-throughput.Wherein, adopting low-density checksum (LDPC) sign indicating number is exactly a kind of scheme that is widely used as chnnel coding and in conjunction with mixing request repeat (HARQ).

At present, the combination of these two kinds of technology mainly contains two class schemes: the one, and low density parity check code and second class (type-II) mix the combination of request repeat technology, this scheme requires the whole bits in the frame all retransmitted, and channel utilization is not high, and throughput is lower; The 2nd, low density parity check code and the 3rd class (type-III) mix the combination of request repeat technology, it is the steadily increase redundancy technology of low density parity check code, though this technology can guarantee good throughput characteristic, corresponding system complexity is increased greatly.

Summary of the invention

The present invention proposes a kind of new mixed automatic repeat request method based on low density parity check code, combine the advantage of two methods, can be under the prerequisite that does not increase the system coding complexity, the throughput of significant increase system, we are referred to as the mixing request repeat technology based on the degree distribution.This new repeating transmission mode is a kind of mode of retransmitting selected, and the content that promptly at every turn resends all is a part of transmitting data for the first time, does not need extra coding.

Low density parity check code is to be subjected to the outstanding chnnel coding of a kind of performance extensively paid close attention at present.Its performance has wide application prospect very near shannon limit.Low density parity check code is a kind of linear block codes, its corresponding sparse parity matrix.The number (being the column weight amount) of " 1 " of each row is called as the degree distribution of row in the check matrix.Low density parity check code is divided into regular code and two kinds of non-regular codes according to the difference of each column weight amount of check matrix: the column weight amount of regular code is all the same, but not the column weight amount of regular code then is a function about row.Because the degree of non-regular code row distributes and does not impose unanimity, has greater flexibility, so performance is better than regular code.Simultaneously, for non-regular code, the degree of protection difference that the difference that degree distributes has caused each bit to be subjected to when decoding, the useful information that the big more node of degree comprises is many more, and is also big more to the contribution of decoding.

The theory that above-mentioned degree distributes is exactly a theoretical starting point of the present invention.

Basic thought of the present invention is that the height that transmitting terminal distributes according to degree selects bit to resend, and spends higher node and is preferentially resend in the time of receiving terminal decoding failure.The present invention has utilized and has spent the design feature that higher node comprises more this low density parity check code of useful information, select retransmit bit according to the degree distribution, system neither needs extra cataloged procedure, and receiving terminal does not need to change decode structures yet, so the present invention is a kind of system efficiently.

Description of drawings

Fig. 1 is transmission of the present invention, receiving system schematic diagram;

Fig. 2 is the retransmit bit selector structure schematic diagram based on the mixed automatic repeat request method of degree distribution that the present invention proposes;

Fig. 3 the present invention is based on the accumulation degree distribution map that the embodiment of non-rule low density parity check code obtains by Computer Simulation;

Fig. 4 the present invention is based on the throughput result that the embodiment of non-rule low density parity check code obtains by Computer Simulation.

Embodiment

Fig. 1 is transmission of the present invention, receiving system schematic diagram.

Concrete workflow is as follows:

1. encoder receives information sequence, adopts non-rule low density parity check code method coding then.

2. coding result is exported by encoder, enters memory stores.

3. after the information in the memory sent out, the noise that superposeed in channel arrived receiver then.

4. receiver end received signal, and the signal storage of the noise that superposeed in memory.

5. signal enters decoder decode.

6. if decode successfully, receiving terminal sends by reverse link determines signal (ACK) to transmitting terminal, and the receiving terminal memory is cleared simultaneously.

7. after transmitting terminal received to determine signal, the content of deletion memory entered memory behind the information coding of next frame.

Otherwise, if decoding failure, receiving terminal sends recurrent signal (NACK), the retransmission counter of receiving terminal adds one simultaneously, the retransmit bit selector is selected the position of retransmit bit according to current Counter Value.

9. transmitting terminal resends bit according to the position of retransmit bit selector indication, and the bit that needs to retransmit is formed a retransmission subpacket bag.

10. the child bag of retransmitting is through channel, and receiving terminal receives the recurrent signal of the noise that superposeed.Under the guidance of retransmit bit selector, the frame of preserving in retransmission subpacket bag that receives and the memory is merged, enter decoder decode then.

11. if decoding is correct, then export the result, jump to step 6.

12. if step 8 is jumped in the decoding failure.

13. all will relatively retransmit Counter Value and the initial maximum number of retransmissions of setting when retransmitting at every turn.When the retransmission counter value greater than maximum number of retransmissions: then transmitting terminal is no longer retransmitted; Receiving terminal is announced the decoding failure, and current decoded result is output; The memory of transmission, receiving terminal all empties; The retransmission counter zero clearing; Enter the transmitting terminal memory behind the information coding of next frame.System begins the transmission of next Frame.

When step 10 receiving terminal merges recurrent signal, can select modes such as equal gain combining, high specific merging for use according to predefined merging mode.If the mode that adopts high specific to merge then needs to obtain channel signal to noise ratio information according to channel estimation module and determines weight coefficient; If adopt the mode of equal gain combining, then not effect of channel estimation module.

Fig. 2 is the structural representation of retransmit bit selector among the present invention.The weight of each row of calculation check matrix determines that the degree of each variable node correspondence distributes.Determine to retransmit putting in order of sequence according to the rule that retransmits then, a content that obtains is the vector of each row sequence number, the order of retransmit bit that Here it is.This sequence is divided into n son bag uniformly, comprises κ bit in each son bag.When resending, transmitting terminal just selects a son bag to send at every turn, and receiving terminal also merges according to the indication of bit selector.

During re-transmission, transmitter of the present invention can adopt a kind of in the multiple choices mode to determine the bit position information that retransmits:

1. the order that distributes according to degree sends a son from high to low at every turn and wraps the degree distribution that the degree of each any bit that sends of assurance distributes and all is not less than remaining bits;

2. repeat k maximum bit of degree of transmission distribution at every turn;

3. the bit of selecting various degree to distribute according to specific ratio retransmits.

Fig. 3, the 4th the present invention is based on the figure as a result that the embodiment of non-rule low density parity check code obtains by Computer Simulation.

The degree distribution function of low density parity check code is among the figure

λ(x)＝2.621×10 ^-4+0.2647x+0.2406x ²

ρ(x)＝0.6351x ⁶+0.3649x ⁷

+0.1588x ⁵+0.0862x ⁶+0.033x ¹³+0.2163x ¹⁴

Simultaneously, adopted the low density parity check code of (1024,506) in the present embodiment.

Fig. 3 the present invention is based on the accumulation degree scatter chart that the embodiment of non-rule low density parity check code obtains by Computer Simulation.

Abscissa is positional information of each row after distribution is rearranged on earth by height according to degree among the figure; Ordinate is an accumulation degree distribution curve, is that current accumulation degree distribution accounts for the ratio that total degree distributes.

Can be observed by Fig. 3, the degree over half of this low density parity check code distributes and is included in preceding 25% the bit (256 bit), and preceding field has comprised about 70% degree, and promptly the degree information of half is included in preceding 1/4th frames in the decoding.

Can be seen that by Fig. 3 most degree distributes and all concentrates on a part of variable node, promptly most of useful information is included in the preceding field.Repeating transmission comprises the node of more degree distributed intelligence, and most useful information has also just obtained utilization.

Fig. 4 the present invention is based on the throughput comparison diagram that the embodiment of non-rule low density parity check code obtains by Computer Simulation.

Wherein first kind of selection mode mentioned above adopted in the system of selection of retransmit bit, i.e. transmission and before different child bags at every turn, and the degree distribution of any bit correspondence in the child bag that guarantees to send all is not less than the degree distribution of remaining bits.Having compared among the figure does not have to retransmit, distributes based on degree and retransmit, select at random to retransmit, all retransmit and five kinds of schemes of node that degree of repeating transmission is less.Wherein selection is retransmitted at random, distribution is retransmitted and the child bag size of these three kinds of methods of bit that the repeating transmission degree is less all is 256 bits based on spending, and maximum number of retransmissions is set at 3, and maximum number of retransmissions was 1 when entire frame was all retransmitted.

Can see by Fig. 4, be far longer than other scheme based on the throughput of system of spending the automatic repeat request scheme that distributes; Repeating transmission is more or less the same with whole retransmission capabilities at random, just selects the efficient of repeating transmission to omit height at random; Repeating transmission degree little this scheme of the node performance in four kinds of automatic repeat request schemes retransmitting that distributes is the poorest, and this is because the useful information that decoder obtains is minimum.

The present invention has made full use of the high bit of non-rule low density parity check code moderate distribution and has carried these characteristics of more useful informations, provides a kind of practical and effective to mix the automatic repeat request scheme---based on the mixed automatic repeat request method of degree distribution.This method band efficiency is higher, realizes flexibly, and the complexity of not obvious increase system, can obtain bigger performance gain simultaneously.

Claims

1. A method for automatically requesting retransmission of information applicable to a low density parity check (Low Density Parity Check, LDPC) code encoding device. The method includes the following steps: sending packet data from the sender to the receiver, and when the receiver receives the data, if the receiver cannot decode correctly, the sender uses an automatic request retransmission mechanism to select part of the data for retransmission.

It is characterized in that,

(1). When the transmitter sends data for the first time, the transmitter sends data, and the retransmission counter is cleared.

(2). The receiver receives the data. If the receiver can decode correctly, the receiver sends an acknowledgment signal to the transmitter, returns to step (1), and the transmitter sends the next data packet; otherwise, skips to step (3).

(3). If the receiver fails to decode, add one to the retransmission counter, and the retransmission bit selector calculates the position of the retransmission bit according to the degree of the variable node and the value of the retransmission counter. Each retransmitted subpackage contains k bits. The sender selects subpackets for retransmission according to the instruction of the retransmission bit selector.

(4). The receiver receives the retransmitted subpacket and merges it with the frame before the last decoding. Decode the merged frame, if successful, the receiver will send a confirmation signal to the transmitter, and skip to step (1) to send the next frame of data; otherwise, skip to step (5).

(5). If the decoding fails, compare the retransmission counter with the maximum number of retransmissions. If the retransmission count is less than the maximum number of retransmissions, skip to step (3); otherwise, skip to step (6).

(6). If the retransmission counter value is greater than the maximum number of retransmissions, then the decoding fails, skip to step (1), output the current decoding result, and continue to send the next data frame.

2. The method for automatically requesting retransmission as claimed in claim 1, wherein the retransmission bits are selected according to the degree of each column in the parity check matrix, and the bits with higher degrees are transmitted preferentially.

3. The degree as claimed in claim 2, wherein the degree is the column weight of each column in the check matrix, ie the sum of non-zero elements in each column.

4. subpackage as claimed in claim 1, it is characterized in that, the size k of subpackage can be set according to parity check matrix, in order to guarantee the effectiveness of transmission, usually guarantee the heavy summation of bit position in the first subpackage Not less than half of the total column weight.

5. transmission effectiveness as described in claim 4, it is characterized in that, ensure that the weight summation of columns in the transmission sub-package is as large as possible, contains more reliable information, reduces unnecessary bit transmission as far as possible again simultaneously, avoids the loss of transmitter power waste.

6. as the selection rule of the sub-package in claim 1, it is characterized in that, sub-package selection comprises the following three ways:

(1). Send a subpacket each time from high to low according to the order of degrees, ensuring that the degree of any bit sent each time is not less than the degree of the remaining bits;

(2). The k bits with the highest degree of sending are repeated each time;

(3). Select various degrees of bit retransmission according to a specific ratio.

7. The frame combining in the receiver as claimed in claim 1, wherein multiple methods can be used for combining, including equal gain combining, maximum ratio combining and so on.

8. The combining method according to claim 7, characterized in that, for different combining methods, the receiver needs to calculate the weighted values corresponding to the signals to be combined.

9. The combining method according to claim 7, wherein when the maximum ratio combining method is selected, channel information needs to be obtained through a channel estimation module.