DE102005036295A1 - Data e.g. data packets, transmission method for e.g. universal mobile telecommunication system- mobile radio system, involves selecting parity bits of single class in repetition packet to transmit all parity bits after repetition packets - Google Patents

Abstract

The method involves converting data into systematic bit and two classes of parity bit by turbo-coding, and transmitting one of data packets from a transmitter to a receiver (2), where the packets contain a systematic bit and a part of a parity bit. One of repetition data packets is transmitted to the receiver during existence of corresponding request of the receiver. A number of parity bits of single class are so selected in the repetition packet that all parity bits are transmitted after the repetition data packets, where the repetition data packets are not selected same for all classes. Independent claims are also included for the following: (1) a transmitter for transmitting of data in form of data packets (2) a receiver for receiving of data in form of data packets.

Description

The The present invention relates to a method for data transmission as well as a suitably designed transmitter and receiver. The data transfer takes place in particular according to a ARQ method or a hybrid ARQ method, in a communication system, in particular a mobile radio system.

Especially in connection with mobile radio systems is often the use of so-called Packet access method or packet-oriented data connections proposed because the emerging message types often a have very high burst factor, so only short activity periods exist that are interrupted by long breaks. packet-oriented Data connections can in this case the efficiency compared to other data transmission methods, in which a continuous data stream is present, considerably increase, as in data transfer procedures a once allocated resource with a continuous data stream, such as. a carrier frequency or a timeslot, while the entire communication relationship remains assigned, i. a Resource remains occupied even if there are no data transfers at the moment abut, so this resource does not work for other network participants to disposal stands. this leads to to a non-optimal use of the scarce frequency spectrum for mobile radio systems.

Future mobile radio systems, such as according to the UMTS mobile standard ("Universal Mobile Telecommunications System "), will offer a variety of different services, besides the pure voice transmission Multimedia applications become increasingly important. The accompanying Requires service diversity with different transmission rates a very flexible access protocol on the air interface in the future Mobile systems. Have packet-oriented data transmission methods proved to be very suitable here.

in the Connection with UMTS mobile radio systems has been involved in packet-oriented data connections so-called ARQ method ("Automatic Repeat Request ") proposed. In this case, the transmitted from a transmitter to a receiver Data packets on the receiver side checked for their quality after decoding. Is a received data packet faulty, the recipient demands a retransmission this data packet from the sender, i. it will be a repeat data packet from the transmitter to the receiver sent, which with the previously sent and received incorrectly Data packet is identical or partially identical (depending on whether the retry data packet less or as much data as the original one Contains data packet, is spoken of a full or a partial repetition). Regarding this for the UMTS mobile standard proposed ARQ method, which also termed hybrid ARQ type I method is both transmission of data as well as so-called header information in one Data packet provided, the header information also information for error checking, how for example, CRC bits ("Cyclic redundancy Check ") and can also be coded for error correction (so-called "Forward Error Correction", FEC).

In UMTS systems are used in particular for the realization of the HSDPA (High Speed Downlink Packet Access) and for the realization of the so-called EDCH (Enhanced Uplink) HARQ method used. For coding the one to be transferred Data is used in particular turbo codes, whereby the to be transmitted Data or bits are converted into systematic bit and parity bit.

Becomes doing the transfer a first data packet not correctly decoded on the receiving side (For example, determined by the checksum), so is a retransmission requested. Upon receipt of the corresponding repeat data packet then a new decode using both data packets carried out.

If over the Air interface transmitted less bit can be must be generated as part of the coding, an appropriate subset for the first transmission (first data packet) and also for the repetition (s) (repetition data packet (s)) are selected. This selection is made by a so-called rate matching algorithm, in particular a puncturing algorithm realized, wherein the total amount of bits appropriate bits are removed (dotted), which then not transferred become.

A punctuation algorithm has already been specified for the UMTS Release 99. This algorithm removes a given number of bits from a block of bits and selects these bits as evenly as possible, ie with as even spacing as possible. When using turbo codes, it was further specified that the so-called systematic bits are not punctured and first parity bits (parity 1 (bit)) and second Parity bit (parity 2 (bit)) are each punctured separately, both parity bit streams are preferably equally punctured.

• – Es gibt Übertragungen, die systematische Bit bevorzugen. Dabei werden die Systematischen Bits nicht punktiert, die ersten und zweiten Paritätsbit werden beide gleich stark punktiert. Eine solche Übertragung wird bevorzugt für die Erstübertragung verwendet.
• – Es gibt Übertragungen, die Paritätsbit bevorzugen. Dabei werden die Paritätsbit nicht punktiert, statt dessen die systematischen Bit. Eine solche Übertragung wird bevorzugt für eine Wiederholungsübertragung verwendet. Dadurch soll erreicht werden, dass bei der Wiederholungsübertragung die Paritätsbit, welche ja bei der Erstübertragung teils punktiert wurden, nun möglichst vollständig übertragen werden. Stehen für die Wiederholungsübertragung, also im Wiederholungsdatenpaket, nicht so viele Bitstellen zur Verfügung, dass alle Paritätsbit übertragen werden können, so werden im Widerholungsdatenpaket überhaupt keine systematischen Bit übertragen und die ersten und zweiten Paritätsbit werden beide gleich stark punktiert.

An important problem to be solved by the rate matching algorithm used is to make an appropriate choice of bits for puncturing for the first transmission and for the repeat transmission (s). The problem arises in particular when a relatively high number of bits must be punctured, especially if at least 1/2 of the bit must be punctured. This was already evident in the introduction of HSDPA. The following was determined:

• - There are transfers that prefer systematic bits. The systematic bits are not punctured, the first and second parity bits are both equally punctured. Such a transmission is preferably used for the first transmission.
• - There are transfers that prefer parity bits. The parity bits are not punctured, instead the systematic bits are punctured. Such a transmission is preferably used for a retransmission. This is intended to ensure that in the retransmission, the parity bits, which were partially dotted during the initial transmission, are now transmitted as completely as possible. If there are not so many bit locations available for the retransmission, ie in the repetition data packet, that all parity bits can be transmitted, then no systematic bits are transmitted in the repeat data packet and the first and second parity bits are both equally punctured.

The individual to be transferred Bit or the bit to be punctured by a defined Rate matching algorithm (Ratematchingalgorithmus) selected.

wobei I die TrCH (Verkehrskanal) Nummer ist und die Folge in 4.2.7.3 für den Uplink und in 4.2.7.4 für den downlink definiert ist. Die Parameter X_i, e_ini, e_plus, und e_minus sind in 4.2.7.1 für den Uplink und in 4.2.7.2 für den downlink angegeben.Here is a quote from the 3GPP Ratematching Algorithm TS 25.212 V6.3.0 (2004-12), Chapter 4.2.7.5 Rate matching pattern de termination (Only the portion relevant for puncturing is shown):
Label the bits before rate matching with:

where I is the TrCH (traffic channel) number and the sequence is defined in 4.2.7.3 for the uplink and 4.2.7.4 for the downlink. The parameters X _i , e _ini , e _plus , and e _minus are given in 4.2.7.1 for the uplink and in 4.2.7.2 for the downlink.

The rate adjustment procedure is carried out according to the following regulation:

The parameters e _plus , e _minus and X _i are set slightly differently for the first and second parity bits in order to achieve different puncturing patterns here as well. Also for different transmissions, the puncturing pattern can be varied by varying the parameter to avoid it that identical transfers are made. A variation of one shifts the puncturing pattern; this is controlled by the parameter r.

The definition of these parameters for the case of EDCH (E-DCH) is contained in the specification in the chapters of 3GPP TS 25.212 V6.3.0 (2004-12) cited below:
"4.9.2.2 Information field mapping of retransmission sequence number"
"4.8.4.3 HARQ Rate Matching Stage"
"4.5.4.3 HARQ Second Rate Matching Stage"

in the This rate adaptation will be briefly explained again below, wherein the presentation is limited to a range in which 1/3 to 1/2 of Bits are dotted.

In 4.9.2.2 it is determined that for the first transmission the E-DCH RV (redundancy version) index "0" used will, and for the first retransmission the E-DCH RV index "3" is used.

In 4.8.4.3 it is specified that the first transmission of the E-DCH RV index "0" is the parameter s = 1 and r = 0 are used. s = 1 means that the systematic bits are preferred. Furthermore, it is determined that the E-DCH RV Index "3" of the retransmission the parameters s = 0 and r = 1 are used. s = 0 means that the parity bits are preferred.

Chapter 4.5.4.3 defines how to calculate the parameters e _plus , e _minus and X _i for the systematic bits, the parity 1 bit (first parity bit) and the parity 2 bit (second parity bit) for the rate matching algorithm. The rate matching algorithm itself is then described in chapter 4.2.7.5 (see above).

This algorithm is shown in pseudo-code. A bit is punctured by assigning it the value delta, since in a following step all bits with the value delta are removed. Translated into German and presented in a slightly more understandable way, the algorithm can also be represented as follows:

Elaborate analyzes underlying the invention have revealed the following. The prior art method works satisfactorily when the coding rate is at most 1/2, that is, when at most one third of the coded bits are punctured. In this case, all the parity bits can be transmitted during the retransmission, which of course ensures that all the bits are available during the subsequent decoding: the systematic bits are completely available from the first transmission, the parity bit is completely from the retransmission, from the In each case, other bits are additionally available in a redundant manner. Due to the fact that all coded bits are available, the full coding rate of the turbo-code is achieved Achieve efficiency of the code.

If However, the coding rate is in the range between 2/3 and 1/2, the is the case when between 1/2 and 1/3 of the bits are punctured, then the procedure of the prior art for the following reasons is disadvantageous or not optimal: In this case, at the first transmission between 1/4 and 1/2 of the parity bits. In the retransmission can transfer more parity bit be, in this case between 3/4 and all. In any case, be in first transmission and retransmission together a larger number transmitted by parity bits, than after coding. It would be theoretically possible, too transfer all parity bits at least once. However will that by the given algorithm of the prior art not reached.

The reason for not transmitting individual parity bits lies in the uniform selection of parity bits according to the prior art rate matching algorithm. This is briefly explained by the following example: It is assumed that there are 20 systematic and 20 parity 1 and parity 2 bits each after coding; During the first transmission, 7 parity 1 bit and 7 parity 2 bits are transmitted in addition to the 20 systematic bits. In total, 20 + 2 · 7 = 34 bits are transmitted. The Ratematching algorithm selects these bits as evenly as possible. This selection is shown in the following table. In the table, the first column indicates the value of the loop variable m of the rate matching algorithm. In the second column, the value of the error variable e is plotted at the time of the if query (the value 20 in the line denoted by ini corresponds to the initial value of e, that is, e _ini ). In the third column, the bits are selected, which are selected by the Ratematchingalgorithmus. In this case, the algorithm chooses the distances between the parity bits to 3, 3, 2, 3,3,3 (the last 3 corresponds to the distance with wrap around from m = 19 to m = 2).

In Columns 4 (e) and 5 (bits) are the values of e and the selected bits for the 2nd transmission (column heading "2.") in the same way, as for the first transmission plotted. At the 2nd transmission (the retransmission) 17 parity 1 and parity 2 bit are transmitted, in total they are again 2 · 17 = 34 bits; the same number as in the first transmission. This is important, because then the same formats for the transfer can be used. The rate matching algorithm chooses the not transferred Bit also equally, namely at a distance of 7, 6, 7 (last distance again wrap around of m = 20 to m = 7).

Column 6 (1 + 2) lists the bits transmitted in either the first or second transmission (first data packet or repeat data packet). As you can see, in this case, this column is identical to the 5th column, ie the first transfer could not stuff any "holes" in the second transfer, which is the least favorable case, you could try changing the starting value for e _In this way, one could achieve that at least some of the bits transmitted during the first transmission encounter gaps in the second transmission, but one can not achieve this in the first _{instance by} shifting the pattern (puncturing pattern) for the second (or the first) transmission the gap between gaps in the second transmission is 7 or 6, that between transmitted bits of the first transmission but 3, 3, 2, 3, 3, 3 Distance of 6 is achievable by the combination of two distances of length 3, but a distance of 7 is not achievable (reachable are 3 + 3 = 6, 2 + 3 + 3 = Even with any shift of the puncturing pattern can not be achieved so that both non-transmitted bit of the second transmission transmitted during the first transmission who the. Thus, according to the prior art, in each case after the second transmission, there is at least one bit which has not been transmitted in any transmission.

Of the The invention is based on the object of specifying a technical teaching, which is a reliable one and preferably simple transmission of data.

Especially The invention is based on the object, a technical teaching specify a simple and reliable transmission of data according to a ARQ method allows that a turbo coding with a rate between 2/3 and 5/7 (under consideration rate matching) and / or a rate matching method with a puncturing rate between 1/2 and 4/5 of the parity bit.

These Task is solved by the characteristics of the independent Claims. Advantageous and expedient developments are the dependent claims refer to. In the context of the invention are also developments the independent one Device claims, the dependent claims correspond to the method claim.

The Invention is based first on the knowledge that it is state of the art, in particular also at a puncture rate between 1/2 and 4/5, is not always possible all parity bits at least once.

One The further step of the invention lies in recognizing that, in particular in the mentioned range of the puncturing rate, a transmission of all parity bits after all possible is. If necessary, three transmissions are received at the receiving end (Data packets) combined.

Elaborate investigations then showed that, especially in the mentioned range of puncturing rates, a transmission all parity bits then possible is when a first data packet contains the systematic bits and one Part of the parity bit contains and the number of parity bits a class in the retry data packet is chosen so that after a number of repetition data packets all parity bits of this class are transmitted where this number of retry data is not pacts for all classes the same So, the number of retry data packets will be at least a class different than for chosen another class becomes.

Become Data for example by a turbo-coding in systematic Bit, a first class of parity bit and a second class of parity bit implemented, so the number of parity bit of the first class in a first repetition data packet selected so that after a repetition data packet all parity bits transferred to the first class and the number of parity bits of the second class in a first and a second repeat data packet selected so that after two retry data packets all parity bits of the second class are transmitted have been.

The parity So the first class will be replaced by a repeat data packet and possibly transmitted by the first data packet.

The parity So the second class will be replaced by two repetition data packets and possibly transmitted by the first data packet.

If all parity bits transmit exactly once in the first data packet or in the retry data packet be, this results in a particularly reliable transmission the data. Same if all parity bit in the first data packet or in the first repeating data packet or in the second repeating data packet transferred exactly once become.

• a) Setzen der Fehlervariablen auf den Initialwert
• b) Setzen des Bitindex auf das erste Bit
• c) Subtraktion des Erniedrigungswertes der Fehlervariablen von der Fehlervariablen.
• d) Falls die Fehlervariable kleiner oder gleich 0 ist durchführen der Schritte e) bis f)
• e) Punktieren des durch den Bitindex angezeigten Bit
• f) Addition des Erhöhungs-Wert zur Fehlervariablen
• g) Erhöhen des Bitindex
• h) wiederholen der Schritte c) bis g), bis der Bitindex die Anzahl der zu verarbeitenden Bit überschreitet.

In particular, in the application of the following rate matching algorithm for selecting the bit contained in a first data packet and / or repetition data packet is achieved in the invention that all parity bits in the first data packet or in the retry data packet and if necessary in the second retry data packet are transmitted exactly once. The (bit) rate matching algorithm contains the following method steps:

• a) Setting the error variable to the initial value
• b) Setting the bit index to the first bit
• c) Subtracting the decrement value of the error variable from the error variable.
• d) If the error variable is less than or equal to 0, perform steps e) to f)
• e) puncturing the bit indicated by the bit index
• f) Addition of the increase value to the error variable
• g) increasing the bit index
• h) repeating steps c) to g) until the bit index exceeds the number of bits to be processed.

These Process steps lead to the same result as the execution of the following process steps or the following rate matching algorithm, its use alternatively to the just specified rate matching algorithm therefore also within the scope of the invention.

e_int

den Initialwert der Fehlervariablen

e

den aktuellen Wert der Fehlervariablen

e_minus

den Erniedrigungs-Wert der Fehlervariablen

e_plus

den Erhöhungs-Wert der Fehlervariablen

X_i

die Anzahl der zu verarbeitenden Bit

m

den Schleifenindex

x_i,m

das Bit Nummer m

δ

die Bezeichnung für ein punktiertes d.h. entferntes Bit.

Where:

e _int

the initial value of the error variable

e

the current value of the error variable

e _minus

the decrement value of the error variable

e _plus

the increase value of the error variable

X _i

the number of bits to be processed

m

the loop index

x _{i, m}

the bit number m

δ

the name for a dotted, ie remote, bit.

Preferably Thus, in the context of the invention, the same rate matching algorithm is used used in a UMTS system for others Purposes already used anyway. This allows the rate matching algorithm be implemented with less effort.

Preferably, the initial value e _ini (r) for the error variable of the rate matching algorithm is chosen such that the first repeat data packet does not contain any parity bits contained in the first data packet and preferably contains those parity bits which are contained neither in the first data packet nor in a further retry data packet ,

• – für den Einsatz des Ratenanpassungsalgorithmus zur Auswahl der Paritätsbit im ersten Datenpaket gilt: eini(r) = {(Xi – ⎣r·eplus/rmax⎦ – 1)modeplus} + 1,
• – für den Einsatz des Ratenanpassungsalgorithmus zur Auswahl der Paritätsbit der einen Klasse im ersten Wiederholungsdatenpaket gilt: eini(r) = {(2Xi + 1 – ⎡(r + 1)·eplus/rmax⎤)modeplus} + 1
• – für den Einsatz des Ratenanpassungsalgorithmus zur Auswahl der Paritätsbit der anderen Klasse im ersten Wiederholungsdatenpaket gilt: eini(r) = {(Xi – ⎡(r + 1)·eplus/rmax⎤)modeplus} + 1,wobei gilt:

  e_ini(r)

  = der Initialwert für die Fehlervariable;

  r

  = 0, falls der Ratenanpassungsalgorithmus in Zusammenhang mit dem ersten Datenpaket angewendet wird;

  r

  = 1, falls der Ratenanpassungsalgorithmus in Zusammenhang mit dem Wiederholungsdatenpaket angewendet wird;

  rmax

  = 2.

For the use of the rate matching algorithm for selecting the parity bit contained or to be transmitted in the first data packet or the first or second repeat data packet, the following preferably applies:

• The following applies to the use of the rate matching algorithm for selecting the parity bits in the first data packet: e ini (r) = {(X i - ·r · e plus / r Max ⎦ - 1) mode plus } + 1,
• For the use of the rate matching algorithm for selecting the parity bit of the one class in the first repeat data packet: e ini (r) = {(2X i + 1 - ⎡ (r + 1) · e plus / r Max ⎤) mode plus } + 1
• For the use of the rate matching algorithm for selecting the parity bit of the other class in the first repeat data packet, the following applies: e ini (r) = {(X i - ⎡ (r + 1) · e plus / r Max ⎤) mode plus } + 1, where:

  e _ini (r)

  = the initial value for the error variable;

  r

  = 0 if the rate matching algorithm is applied in conjunction with the first data packet;

  r

  = 1 if the rate matching algorithm is applied in conjunction with the retry data packet;

  rmax

  = 2.

Nt,sys = Ndata – Nt,p1 – Nt,p2;dabei bezeichnen

Ndata:

Verfügbare Bit für die Übertragung

Nsys:

Anzahl der systematischen Bit vor der Ratenanpassung

Nt,sys:

Anzahl der systematischen Bit nach der Ratenanpassung

Nt,p1:

Anzahl der Parity 1 Bit nach der Ratenanpassung

Nt,p2:

Anzahl der Parity 2 Bit nach der Ratenanpassung.

In addition, the following may preferably apply:

N t, sys = N data - N t, p1 - N t p2 ; thereby designate

N data:

Available bits for transmission

Nsys:

Number of systematic bits before rate matching

Nt, sys:

Number of systematic bits after rate matching

Nt, p1:

Number of parity 1 bit after the rate adaptation

Nt, p2:

Number of parity 2 bits after rate matching.

The following extensive investigations have shown that, in particular for the specified choice of the initial value for the error variable e _ini for the rate adaptation in the case of the first data packet and / or the first repeat data packet, all parity bits are transmitted once in total:
The description starts with a treatment of the case that the coding rate is between ½ and 2/3, later also the case is treated that the coding rate lies between 2/3 and ¾.

Especially It is an object of the present invention, a technical To provide teaching for many cases ensures that all parity bits (parity bits) are transmitted, i. each Bit is played at least once in the first transmission (in the first data packet) or the second transfer (in the repeat data packet). This goal is achieved by taking less than the maximum possible number from parity bit in the second transmission (in the repeat data packet) (see above). In this case, the number of transmitted in the second transmission Bit and the shift of the puncturing pattern so cleverly chosen that every parity bit exactly once, either in the first transfer or the second transmission becomes.

For the derivation of the optimal procedure, the known rate matching algorithm - in a slightly different form - is considered again: Algorithm for first transmission analysis, called A1:

For notation, another index is introduced which distinguishes between first and second transmission, ie between A1 and A2. The variables are each denoted by:
instead of e with e ₁ and e ₂
instead of e _ini with e _ini1 and e _ini2
instead of e _minus with e _minus1 and e _minus2
instead of e _plus with e _plus1 and e _plus2

in this connection the decrementation of e is not done at the beginning of the loop, but for the first loop pass in the initialization and for the others Loop Count in the if or else branch of the previous loop pass. These We use rewording of the rate matching algorithm for the analysis the first transmission. (It should be mentioned here, that for an implementation of course uses any variant of the implementation of the algorithm can be, we consider special variants easier here the optimal parameters for both transmissions to apply.)

For the analysis the second transmission formulate the algorithm a bit more by using the swap the if and else branch, thereby achieving the condition e <= 0 vice versa becomes e> = 1. This are mutually inverse conditions, i. the second is just then Fulfills, if the first is not fulfilled and vice versa.

Algorithm for analysis of the second transmission, called A2:

aim is to achieve that at every bit, so for all m, then at A2 the bit used or transmitted becomes when is punctured at A1, and vice versa.

Both algorithms A1 and A2 should therefore both run into the if branch or both into the else branch for a given m. This can be realized if, for each iteration, ie for every m: e 2 = 1 - e 1

For then the condition from A2, namely e ₂ > = 1, is equivalent to 1 - e ₁ > = 1, because it follows that
-E ₁ > = 0 and e ₁ <= 0 are synonymous with the condition in A1.

The equality e ₂ = 1 - e ₁ is given if, firstly, this applies to the initial value when entering the loop, and, secondly, if the update operations in the two loops in the If and Else branches are corresponding.

"first" means:
e ₂ = 1 - e ₁ at the loop _entrance , After inserting the dependence of e _ini1 and e _ini2 one obtains: e ini2 - e MINUS2 = 1 - e INI1 - e minus 1

"second" means that the changes of e in both the if branch and the else branch must be the same: + e plus1 - e minus 1 = -E MINUS2 and e minus 1 = + e plus2 - e MINUS2

By substituting the definitions of e _minus and e _plus from the citation of the specification, one can verify that this is the case when, in the second transmission, the difference is the number of parity bits present minus the parity bits transmitted in the first transmission , This equation should apply to parity 1 bits and parity 2 bits. Thus, the core of the invention, namely to send exactly as many parity bits in the second transmission as were missing in the first transmission, is also derived via the rate matching algorithm.

So that both puncturing patterns interlock, the first condition should also be satisfied, namely e ini2 - e MINUS2 = 1 - e INI1 - e minus 1

For this there are many solutions, for each given e _ini1 one can determine a suitable e _ini2 . It is, for practical reasons, advantageous to maintain the definition of e _ini1 since this definition has already _yielded good results in many simulations (but of course the other solutions are also included in the invention). Then one can calculate that for e _ini2 : e ini2 (r) = {(X i - ⎡ (r + 1) · e plus / r Max ⎤) mode plus } + 1

It has already been taken into account that the parameter r is set to r = 0 for the first transmission, but to r = 1 for the second transmission, and that the transmission was specified for the first transmission e INI1 (r) = {(X i - ·r · e plus / r Max ⎦ - 1) mode plus } + 1

With this choice of the number of parity bits and this choice of e _ini for the second transmission, one can achieve that the patterns interact optimally.

This is in the above table in columns 7 and 8 (both overwritten with 2 opt.). Again, the value of e is dependent of m and the selected Bits are shown. In the last column (1 + 2 opt) are the Bit listed, which are now sent either in the first or second transmission. As you can see now actually transmit every bit. At the second transmission exactly those bits are transmitted, which in the first transmission not transferred were. This will significantly improve the transmission characteristics reached.

Surprisingly, with the use of an algorithm already used for other purposes, it is possible with the invention to achieve a significant improvement in the transmission characteristics if the parameters for calculating the number of parity 1 and parity 2 bit and the calculation of e _{ini are} adapted only during the retransmission , The formulas for this are no more complex than the formulas according to the prior art.

It Preferably, during the second transmission, less parity bits are transmitted, as provided in the prior art. The remaining bits are then preferred filled with systematic bit. This has another advantage: Should a receiver the Erstübertragung completely for some reason do not receive e.g. because an associated signaling information could not be received correctly, it will only be the second transmission receive. Because this second transmission now contains more systematic bit than According to the prior art, it is easier to decode, which in turn improves the system properties. Further lie also with the detection of the first and second transmission better information over the systematic bit before, since of these now some double (in the first and second transmission) become. This also improves, as in the study of HSDPA was shown, the performance, so the decoding probability. With HSDPA an improvement of the quality by so-called SMP (Signal Mapping Priority), while doing systematic bit comparison and in contrast to parity bit mapped to modulation symbols, which had a better decoding property became thereby a better reception quality the systematic bits and thereby better decoding property reached. However, this method is only at higher modulations (e.g. QAM), and therefore not applicable to E-DCH. By the Optimized puncturing in the retransmission can be but a similar one Achieve effect, making a similar Improvement of system performance can be achieved.

These Explanations the elaborate - the invention underlying investigations and considerations should only be the background of the complex invention and its Embodiments illuminate, but the invention in no way to certain restrict preferred embodiments.

Preferably become the puncturing patterns for the case of the first data packet and the case of the retry data packet nested together.

Preferably becomes the method according to the invention for data transmission in the context of a hybrid ARQ procedure for the realization of the HSDPA or EDCH (E-DCH).

As already explained several times, according to a preferred embodiment of the invention, the parity bits contained in a data packet or repetition data packet are selected from the turbo-coding-resultant parity bits by a rate matching algorithm. Are selected by the rate matching algorithm certain bits from a set of bits, because not all bits from the set of Bits are transmitted or contained in a data packet, the rate matching algorithm acts as a puncturing algorithm. The punctured by the puncturing algorithm bit are not transmitted and are not included in the corresponding data packet. The remaining bits are contained in the corresponding data packet.

Preferably the punctuation rate of the rate matching algorithm is in the range between 2/3 and 3/4; that is the case when between 1/2 and 5/9 of the bits are punctured. In this case, at the first transmission (and the third) between 1/4 and 1/6 of the parity bits and in the second transmission at least 4/6 of the parity bits, in total, can So in Erst- second and third transfer transmit all parity bits become. In the retransmission could while theoretically more parity bits are transmitted, namely between 4/6 and 3/4. According to the invention but only the first transmission and the second retransmission still missing parity bit transferred.

Preferably the coding rate of the turbo coder is taken into account the rate adjustment between 2/3 and 3/4.

The In particular, the invention contemplates turbo-coding with a Rate between 2/3 and 5/7 (taking into account the rate adjustment) and / or a rate matching method with a puncturing rate between 1/2 and 4/5 of the parity bit. In the context of this invention, the influence of the so-called. Termination (Tail Bits) are not considered closer to the bits generated by turbo code. As in UMTS, one can e.g. one third of the termination bits the Systematic bit, the parity 1 and the parity 2 bits strike. Strictly speaking, the coding rate is due to the influence of the termination bits a little lower than stated, because you can calculate the coding rate Strictly speaking, the termination bits would also have to be considered. These But fineness is for the invention irrelevant, which is why both types of calculation by the invention is included. For the invention is the total number of parity 1 and parity 2 bits relevant, regardless whether it is "real" parity bit or termination bit.

Of course it includes the invention also includes data transmissions, where besides the stipulated puncturing rates or code rates any other puncturing rates or code rates are used, wherein at the other puncturing rates or code rates also arbitrary other rate adaptation methods or initial values for the error variable can be used. In particular, data transfers are in the Frame of the invention, in which at a puncture rate between 1/2 and 5/9 apply rate matching or bit selection according to the method claims and at other puncturing rates, rate matching or bit selection according to the state the technique, in particular according to the rate matching algorithm according to the currently current UMTS standard (see above) is used.

in the The scope of the invention is also a hardware technology and / or software technology appropriately arranged transmitter, in particular a mobile station or a base station, and a hardware and / or according to software technically configured receiver, the in particular a base station or a mobile station.

in the Frame of the invention are also a transmitter and a receiver, the to carry out of the method according to claim 2 are set up accordingly.

The The present invention will be explained in more detail below with reference to FIGS attached Drawings based on preferred embodiments of a packet-oriented data transfer explained in a mobile radio system, Of course, the present invention does not apply to mobile radio systems is limited but generally used in any kind of communication systems in particular an ARQ method for data transmission is provided.

1 shows a representation to clarify the signal processing according to a packet-oriented ARQ method,

2 shows a representation to illustrate the communication in a mobile radio system;

3 shows a diagram;

4 shows a diagram.

As has already been explained above, it is assumed below that, with the aid of the present invention, a packet-oriented data transmission in a mobile radio system, as shown schematically in FIG 2 is shown to be realized. It is in 2 exemplifies the communication between a base station 1 and a mobile station 2 a mobile radio system, for example a UMTS mobile radio system represented. The transmission of information from the base station 1 to the mobile station 2 takes place via the so-called "downlink" channel DL, while transmitting the information from the mobile station 2 to the base station 1 via the so-called "uplink" channel UL.

The present invention will now be described by way of example with reference to a packet-oriented data transmission from the base station 1 to the mobile station 2 , ie, explained by packet-oriented data transmission over the "downlink" channel, but the present invention is analogously applicable to data transmission over the "uplink" channel. Furthermore, the present invention is explained below with reference to the signal processing measures to be carried out in the respective transmitter, wherein, however, it is to be noted that in the respective receiver for evaluating the data processed in this way on the transmitter side a corresponding signal processing in reverse order is required, so that the present invention, not only the transmitter side, but also the receiver side is concerned and encompassed. Although no puncturing is made in the receiver in the sense that bits are selected from a bit stream which are then not transmitted. However, the receiver preferably implements a corresponding algorithm that takes into account which bits have been punctured, that is, not transmitted, to correctly determine the relationship of the actual transmitted bits to the bits of the original coded bitstream. Of course, the punctured bits can not be recovered, but at the receiving end, the identity of the transmitted bits is correctly detected in order to use them correctly for decoding.

In 1 the signal processing of the data and header information to be transmitted in the data packets is shown according to a hybrid ARQ method.

On the header side are those of a function block 3 generated header information a function block 12 which ensures that all headers of all data packets to be sent in one and the same radio packet are combined into a single header (so-called header concatenation). A functional block 13 adds CRC bits for header recognition to the resulting header information. Subsequently, by a function block 14 a channel coding and a functional block 15 rate matching of the resulting bitstream. An interleaver 16 causes the symbols or bits supplied to it to be reordered and spread in a specific manner. The one from the interleaver 16 output data blocks are from a function block 17 assigned to the individual transmission or radio frames (so-called "radio frame segmentation"). However, the coding of the header information is only of secondary importance for the invention.

On the data page is also a function block 4 provided for adding CRC bits. A functional block 5 serves for the splitting of a channel encoder 6 supplied data such that from the channel coder 6 always a limited to a certain number of bits encoding can be performed.

By the channel coder 6 Channel coding performed adds redundant information to the data actually being sent. This has the consequence that several data packets transmitted one after the other have bits with the same information origin.

Those of the channel coder 6 output bits become a function block 19 fed, which adjusts the bit rate of the bit stream by hiding or omitting individual bits (so-called puncturing) or by repeating individual bits (so-called repetition). From a subsequent function block 9 For example, so-called DTX bits ("discontinuous transmission") can be added to the data stream. Furthermore, function blocks are also on the data page 10 and 11 provided the same functions as the function blocks provided on the header side 16 and 17 perceive.

Finally, the bits output on the data and header side are from a function block 18 mapped or multiplexed on the respectively present physical transmission or transmission channel (so-called "multiplexing") and with the aid of a suitable modulation, for example a QAM modulation, transmitted to the receiver.

at The Hybrid ARQ Type I method will fail if received or a faulty decoding of a data packet by the receiver a repeat data packet requested, which with the previously sent and incorrectly received data packet completely or partially identical is. Depending on whether the retry data packet is less or equal to how much data the original one Data packet is characterized by a full or partial repetition spoken. The data packet and the respective repeat data packet thus have bits with an at least partially identical information origin on. The recipient can thus by common evaluation of the originally sent data packet and the requested subsequent retry data packets originally regain sent information with better quality.

The present invention relates essentially to the in 1 shown function section 19 , This function section 19 includes a functional block 20 , which depends on a control by the function block 3 that from the upstream channel encoder 6 output coded bits to at least two parallel partial bit streams, which are each separately, ie independently, subjected to a rate adaptation. In 1 In this regard, three partial bit streams A-C are shown, with one functional block for each partial bit stream 21 - 23 to perform a corresponding rate adaptation, ie for puncturing or repetition of individual bits, is provided. In this way, several differently coded parallel partial bit streams, which form a further functional block 24 be supplied. This further functional block 24 The task is to separate the individual bits of the parallel bit streams in the same order as that of the function block 20 for the bit separation, ie for the division on individual parallel partial bit streams, has been used to collect (bit collection). This ensures that overall the order of bits left after rate matching does not change.

As has already been explained above, the rate adaptation provided for the individual partial bit streams A-C by the function blocks 21 - 23 completely independent of each other. In particular, the bits of one or more sub-bit streams can not undergo puncturing or repetition at all. Overall, the rate adaptation of the individual parallel partial bit streams A-C is to be selected such that of the entire functional section 19 on the one from the function block 6 output channel-coded bit stream per data packet or repetition data packet, a desired rate matching pattern is applied. With the in 1 shown realization of the functional section 19 with multiple parallel rate adjustments, extremely high coding flexibility can be achieved.

The functional section 19 is configured such that it depends on the control by the function block 3 apply a different rate matching pattern to the bits of a repetition data packet than to the bits of the corresponding originally transmitted data packet. That is the functional section 19 is from the function block 3 informed whether a repeat data packet has been requested by the respective recipient, the functional section 19 in this case those of the individual function blocks 21 - 23 realizes rate matching pattern such that, overall, the bits of the repeat data packet are processed with a different rate matching pattern than the bits of the underlying originally transmitted data packet.

The total of the functional section 19 For example, realized rate matching may be performed according to the rate matching algorithm, which is known per se from the prior art (see above).

• – Zur Datenübertragung in Form von Datenpaketen werden die Daten durch eine Kanalcodierung, insbesondere eine Turbo-Codierung, in systematische Bit und Paritätsbit umgesetzt. Von dem Sender wird an den Empfänger ein erstes Datenpaket gesendet, das die systematischen Bit und einen Teil der Pari tätsbit enthält. Bei Vorliegen einer entsprechenden Aufforderung des Empfängers (2) wird mindestens ein Wiederholungsdatenpaket an den Empfänger (2) gesendet, das die Paritätsbit enthält, die im ersten Datenpaket nicht enthalten sind.
• – Das Wiederholungsdatenpaket enthält keine Paritätsbit, die im ersten Datenpaket enthalten sind.
• – Das Wiederholungsdatenpaket enthält neben den Paritätsbit, die im ersten Datenpaket nicht enthalten sind, systematische Bit.
• – Die Kodierungsrate des Turbocoders liegt unter Berücksichtigung der Ratenanpassung zwischen 2/3 und ½.
• – Die in einem Datenpaket oder Wiederholungsdatenpaket enthaltenen Paritätsbit werden durch einen Ratenanpassungsalgorithmus aus den aus der Turbocodierung resultierenden Paritätsbit ausgewählt.
• – Die Punktierungsrate des Ratenanpassungsalgorithmus liegt zwischen ½ und 1/3.
• – Der Initialwert der Fehlervariablen des Ratenanpassungsalgorithmus wird derart gewählt, dass das Wiederholungsdatenpaket keine Paritätsbits enthält, die im ersten Datenpaket enthalten sind.
• – Der Ratenanpassungsalgorithmus enthält folgende Verfahrensschritte: a) Setzen der Fehlervariablen auf den Initialwert b) Setzen des Bitindex auf das erste Bit c) Subtraktion des Erniedrigungswertes der Fehlervariablen von der Fehlervariablen. d) Falls die Fehlervariable kleiner oder gleich 0 ist durchführen der Schritte e) bis f) e) Punktieren des durch den Bitindex angezeigten Bit f) Addition des Erhöhungs-Wert zur Fehlervariablen g) Erhöhen des Bitindex h) wiederholen der Schritte c) bis g), bis der Bitindex die Anzahl der zu verarbeitenden Bit überschreitet.
• – Für den Einsatz des Ratenanpassungsalgorithmus zur Auswahl der im ersten Datenpaket enthaltenen Paritätsbit gilt: eini(r) = {(Xi – ⎣r·eplus/rmax⎦ – 1)modeplus} + 1und im Wiederholungsdatenpaket enthaltenen Paritätsbit gilt: eini(r) = {(Xi – ⎡(r + 1)·eplus/rmax⎤)modeplus} + 1, wobei gilt:

  e_ini(r)

  = der Initialwert für die Fehlervariable;

  r

  = 0, falls der Ratenanpassungsalgorithmus in Zusammenhang mit dem ersten Datenpaket angewendet wird;

  r

  = 1, falls der Ratenanpassungsalgorithmus in Zusammenhang mit dem Wiederholungsdatenpaket angewendet wird.

The just explained functional sections, in particular the functional section 19 , may be set up so that one or more of the following configurations are put into practice:

• - For data transmission in the form of data packets, the data is converted by a channel coding, in particular a turbo-coding, into systematic bit and parity bit. The transmitter sends to the receiver a first data packet containing the systematic bits and a part of the parity bit. If there is a corresponding request from the beneficiary ( 2 ) at least one repeat data packet is sent to the receiver ( 2 ) containing the parity bits not included in the first data packet.
• The retry data packet contains no parity bits contained in the first data packet.
• The repetition data packet contains systematic bits in addition to the parity bits which are not contained in the first data packet.
• - The coding rate of the turbo coder is between 2/3 and ½, taking into account the rate adaptation.
• The parity bits contained in a data packet or repetition data packet are selected by a rate matching algorithm from the turbo-coding resulting parity bits.
• The puncturing rate of the rate matching algorithm is between ½ and 1/3.
• The initial value of the error variable of the rate matching algorithm is chosen such that the repeat data packet does not contain any parity bits contained in the first data packet.
• The rate matching algorithm contains the following method steps: a) setting the error variable to the initial value b) setting the bit index to the first bit c) subtracting the reduction value of the error variable from the error variable. d) If the error variable is less than or equal to 0, perform steps e) to f) e) puncturing the bit index indicated by the bit index f) adding the increment value to the error variable g) incrementing the bit index h) repeating steps c) through g ) until the bit index exceeds the number of bits to be processed.
• The following applies to the use of the rate matching algorithm for selecting the parity bits contained in the first data packet: e ini (r) = {(X i - ·r · e plus / r Max ⎦ - 1) mode plus } + 1 and the parity bit contained in the retry data packet: e ini (r) = {(X i - ⎡ (r + 1) · e plus / r Max ⎤) mode plus } + 1, where:

  e _ini (r)

  = the initial value for the error variable;

  r

  = 0 if the rate matching algorithm is applied in conjunction with the first data packet;

  r

  = 1 if the rate matching algorithm is applied in conjunction with the retry data packet.

At the receiving end the first data packet and the repeat data packet are combined, in particular by so-called combining soft combining (summarizing the soft-decission values, so summarizing the probabilities that individual bits have the value 1 or 0), decoded.

The embodiments just explained are particularly suitable for the case where the coding rate is in the range between 2/3 and 1/2; this is the case when puncturing between ½ and 1/3 of the bits. In this case, between 1/4 and 1/2 of the parity bits are transmitted during the first transmission and at least ¾ of the parity bits in the second transmission, so in total, all parity bits can be transmitted in the first and second transmission. At higher coding rates than 2/3, this is no longer the case, so the formulas given above are then preferably not applied directly. In particular, it is then no longer possible to select for the second transmission exactly the parity bits which were not transmitted in the first transmission since not all the missing ones can be transmitted in the secondary transmission. However, the starting value e _{ini may} preferably be selected such that the least possible overlap results between the parity bits which are transmitted on the first and second times.

In the following, however, solutions are also given for this case:
In this case, it is impossible to transmit all the parity bits after the second transmission. But this can be possible again after the third transmission. Conveniently, one uses at a coding rate that is not too much over 2/3, for the third transmission again a transmission that prioritizes the systematic bit; this is already provided in the standard. The puncturing patterns for the 1st and 3rd transmission are shifted against each other. It is possible, by suitable choice of the parameter e _ini for the 3rd transmission, to have the amount of parity bits transmitted in the 1st and 3rd transmission taken together having a similar distribution (a similar pattern) a single transmission, but twice as many bits are transmitted. The second transmission can now be optimized to transmit exactly those bits that are transmitted neither in the first transmission nor in the third transmission. In this case, the above-mentioned formulas can be used in an analogous manner, whereby, of course, it is taken into account that twice as many parity bits are transmitted in the first and third transmission as in the first transmission alone. The formulas for the optimal choice of the parity 1 bit and the parity 2 bit to be transmitted in the second transmission can then be derived as follows: Determine an equivalent rate matching which selects the parity 1 and parity 2 bits which be transmitted in the first or third transmission. The number of transmitted bits corresponds to the sum, the value of eini should be determined accordingly. Possibly. if one chooses the value of one for the third transmission appropriately (in relation to the value of one for the first transmission) that these two transmissions can be grouped together to give a total transmission pattern, as in a single transmission. The value for the number of parity 1 and parity 2 bits to be transmitted and the value of one for the second transmission can then be determined on the basis of the pattern combined for this, as described above.

In Chapter 4.5.4.3, Table 10 defines how to calculate the parameters e _plus , e _minus and X _i for the systematic bits, parity 1 bit (first parity bit) and parity 2 bit (second parity bit) for the rate matching algorithm become:

To simplify the nomenclature, we introduce the parameter a, which takes the value 2 for the parity 1 bit and the value 1 for the parity 2 bits. Then, for both types of parity bits, for the case of puncturing considered here: eplus = aN _p and eminus = a abs (N _p -N _{t, p} ) = a (N _p -N _{t, p} )

N _p1 and N _p2 or generalize N _p the number of parity 1 or parity 2 bits before the rate adaptation, N _{t, p1} and N _{t, p2} or generalizes N _{t, p} the number of transmitted parity 1 or Parity 2 bits. The task now is to parameterize the third transmission so that in the first and third transmission taken together the same bits are transmitted as in a transmission that transmits twice as many parity bits as well as to determine the parameters for this hypothetical transmission. If necessary, the parameters used in this hypothetical transmission are indicated by the index h. In such a transfer, the value eminus would be set to a ( _Np -2N _{t, p} ) since twice as many bits are transferred, but the eplus value would be unchanged.

We formulate the algorithm further to the algorithm A3: Algorithm for analysis of combined transmission, called A3:

The value of e at each non-transmitted, so punctured bits to the value eplus as is easily seen when viewing the algorithm A2 is - eminus = a N _p - a (N _p - N _{t, p)} = a N _{t, p} elevated. For each transmitted bit e is decremented by the value eminus = a N _p . The reduction value is the same for the first, third and hypothetical transfer, but the increase value for the hypothetical transfer is twice as large. The result is, if one plots the value of e over m each a sawtooth-like curve, the height of the spikes is the same, but the slope of the spikes for the hypothetical curve twice as large and thus the number of spikes. Each spike corresponds to a transmitted bit. The aim is to find a parameterization for the hypothetical transmission in which exactly the bits transmitted in the first and second transmission are transmitted together. For this purpose, the third and first transmission must be well nested, which can be achieved if their values differ from one half the value of the increase value, ie if the difference is (a N _{t, p} / 2). This is already the case according to the current state of the standard. If one chooses the value of e at the first if query for both the first and the hypothetical transmission to be equal to 1, then the "saw teeth" fit appropriately, but this is not the case by default, but one can do a proper alignment If one considers the double slope of the sawtooth curve, then the distance of 1 (the comparison value in the query) for the hypothetical transmission must be twice as large as for the first transmission: 2 (1 - e 1 (m = 1)) = (1 - e H (m = 1))

From this one can also calculate the corresponding value of one for the hypothetical transmission, since e (m = 1) = e _ini - e _plus . Again, from the value of e _ini for the hypothetical transmission, analogously, as already shown, the necessary value of e _ini for the second transmission can be calculated, so that the bits of the second transmission are exactly the bits that are not transmitted in the hypothetical transmission and thus exactly those which are not transmitted in either the first or the third transmission.

und analog:

The number of systematic or parity 1 and parity 2 bits to be transmitted in the second transmission is then in the range of the coding rate discussed here, ie if (4/3) N _sys ≦ N _data <1.5 N _sys : N t, sys = 3N daty - 4N sys

and analogously:

The result for the second transmission is the formula e ini2 = (X i Fashion plus · 2 + 2.

This Value hangs not that in the first transmission used value of r, which is also logical: if in the first transmission the value r = 0 is used, so in the third transmission the value r = 1 is used. If the value r = 1 is used in the first transfer would, so would have to in the third transmission the value r = 0 can be used. Both times will be in the first you third transfer taken together transmit the same bits, so that both times for the second transmission the same parameterization results.

Of course, it is quite possible, using the parameter r, to define another transmission which contains bits other than the second transmission, and which could then be used, for example, in a fourth transmission. As already stated in the specification, this would be done by modifying the value of e _ini by r * e _plus / r _max . In order to maintain the convention that the second transmission is transmitted with r = 1 and that r _max = 2, the additive term results: (1 - r) e plus / r Max

The formula for e _ini for the parity bits of the retry _{data packet} would thus be: e ini2 (r) = {(2X i - ⎡ (1 - r) · e plus / r Max ⎤) mode plus } + 2 this formula does not yet take into account that the value of e _{ini should} always be between 1 and e _plus , one can achieve this by the Umformumg e ini2 (r) = {(2X i + 1 - ⎡ (1 - r) · e plus / r Max ⎤) mode plus } + 1 and further, since r _max = 2 and considering the modulo function e ini2 (r) = {(2X i + 1 - ⎡ (r + 1) · e plus / r Max ⎤) mode plus } + 1

Note: Unlike in the case of the code rate being between ½ and 2/3, in this case it is not particularly relevant whether the above calculation rounds up or down, since now for the second transmission, r = 1, then If (r + 1) e _plus / r _max = e _plus and e _plus is already an integer, the rounding does nothing in this case and thus it is irrelevant whether it is rounded up or down.

This embodiment is again illustrated by a table. Again, consider the case where there are 20 systematic (and 20 parity 1 or 2 parity) after encoding. It is now assumed that 28 bits of this can be transmitted with each transmission. Then the following values result for e and the following transmitted bits:

The Designation is according to the above table.

• 1: erste Übertragung,
• 3: dritte Übertragung,
• 2: Zweite Übertragung nach dem Stand der Technik,
• hyp 1 + 3: Hypothetische Übertragung, die die Parity Bit der ersten und dritten Übertragung umfasst,
• 1 + 3: zum Vergleich dazu die in der ersten und dritten Übertragung übertragenen Bit
• 1 – 3: in der ersten bis dritten Übertragung übertragene Bit
• 2 opt: zweite Übertragung nach dem Ausführungsbeispiel
• 1 – 3 opt: in der ersten bis dritten Übertragung übertragene Bit nach dem Ausführungsbeispiel
• 1 + 2 opt: in der ersten und zweiten Übertragung übertragene Bit nach dem Ausführungsbeispiel
• 1 + 2: in der ersten und zweiten Übertragung übertragene Bit nach dem Stand der Technik

The columns indicate:

• 1: first transmission,
• 3: third transmission,
• 2: second transmission according to the prior art,
• hyp 1 + 3: Hypothetical transmission comprising the parity bits of the first and third transmission,
• 1 + 3: for comparison, the bits transmitted in the first and third transmission
• 1 - 3: bits transmitted in the first to third transmission
• 2 opt: second transmission according to the embodiment
• 1 - 3 opt: bits transmitted in the first to third transmission according to the embodiment
• 1 + 2 opt: bits transmitted in the first and second transmission according to the embodiment
• 1 + 2: prior art bits transmitted in the first and second transmission

In the columns below there are some parameters listed:
Transmitted parity 1 bits
Transmitted Systematic bits:
X _i
e _plus
e _minus
r
e _ini

Then follow the values of parameter e as a function of m and there are the transferred ones Bit marked.

you sees that the parameters of the hypothetical transfer actually so chosen are that same bits selected become, as in the first and third transmission together. Furthermore it can be seen that in the first to third transmission according to the embodiment transmit all parity bits be, unlike the prior art. In the last two Columns are shown that after the invention not only after the third transmission gives an advantage, but already after the second transmission. Although in this example in the second transmission according to the embodiment only 12 instead of 14 as in the prior art, the total number is in the first and second transmission transmitted according to the embodiment Parity bit equal to 16 and thus higher as in the prior art (14).

Alternatively, the following procedure for determining suitable values for e _{ini is} possible: As in the case where the coding rate is between ½ and 2/3, formulas are written down for the three rate matching algorithms, except that in this case they are written instead of three are. Then it is determined that for each value of m, one bit is always used for exactly one algorithm. Analogously, as shown above, this leads to corresponding conditions for the values of e in the three algorithms and thus also for the values of e _ini .

This embodiment is explained by a numerical example:
Suppose that a transmission consists of 100 systematic bits each, parity 1 bit and parity 3 bits, ie a total of 300 bits. Suppose further that 140 bits are transmitted in a transmission. In the first (and also in the third) transmission, the 100 systematic bits and 20 each parity 1 and parity 2 bits are transmitted. Together in 1st and 3rd transmission so each 40 parity 1 and parity 2 bits. thus missing 60 parity 1 and parity 2 bits each. Exactly these bits are now transmitted in the second transmission. In order to get the total number of 140 bits again, in the second transmission also 140-60-60 = 20 systematic bits are transmitted. This embodiment can be applied if at least 1/6 of the parity 1 (or parity 2) bits are transmitted in the first transmission. Then 1/6 can also be transmitted in the third transmission, in the second transmission at least ½ + 1/6 = 2/3 of the parity bits can be transmitted, that is, together at least 2/3 + 1/6 + 1/6 = 1/1 all parity bits. This embodiment is therefore particularly applicable for coding rates between 2/3 and ¾ = 0.75.

This embodiment is described below in further formulas: Let N be the number of systematic bits; this is the number of parity 1 as well as the parity 2 bit (a generalization to the case that these numbers are different is easily possible but will be omitted here for the sake of clarity) Let Ns1 = Ns3, N11 = N13 and N21 = N23 the number of systematic, parity 1 and parity 2 bits transmitted in the first (and third) transmissions. Let Ns2, N12 and N22 be the number of systematic, parity 1 and parity 2 bits to be transmitted in the second transmission. Then: N12 = N - N11 - N13 N22 = N - N21 - N23 Ns2 = Ns1 + N11 - N12 + N21 - N22 = Ns1 + N11 + N21 - N + N11 + N13 - N + N21 + N23 = Ns1 + N13 + N23 + 2 · (N11 + N21 - N)

The last line is simply the expression that the remaining bits filled with systematic bit be, so bring for the practice no further computational simplification but is more one alternative presentation.

This generalization of the method according to the invention thus allows the application up to quite high coding rates for the first transmission. At even higher rates, ie a rate greater than ¾, two retransmissions are required, which prefer parity bits to transfer all parity bits in total in three transmissions. In this case, all the parity bits are already transmitted by the two retransmissions, so that it is not decisive for the performance after receipt of the third transmission which of the parity bits were additionally sent in the first transmission. However, in order to improve the performance after the second transmission, it is still desirable that in the second transmission (first retransmission) those parity bits are transmitted which were not transmitted in the first transmission. In this respect, the invention can be applied in a modified form also in this case. In particular, the value of e _{ini should} again be chosen such that there are as few overlaps as possible, and it may again be advantageous to send less parity bit in a transmission than would be possible in order to achieve a better intermeshing of the patterns.

alternative you can at a higher Encoding rate as ¾ too more than two transfers with s = 1 with a transmission combine with s = 0. Again, one sets the shift of Puncturing pattern of parity bits of multiple transmissions with s = 1 so, that in combination the pattern results as in a single one hypothetical transmission and designed the transmission with s = 0 so that just the missing bits are transferred become. This is a generalization of the coding rate method given above between 2/3 and 3/4. As a further alternative, one can apply the invention even then if fewer transfers performed with s = 1 are needed, as actually needed Then one also determines a hypothetical transmission the all parity bit both of the carried out transfers with s = 1 so also the non-sent and designed the transmission with s = 0 such that in this case just those not transferred in the hypothetical transfer Transmit bit become.

• – Zur Datenübertragung in Form von Datenpaketen werden die Daten durch eine Kanalcodierung, insbesondere eine Turbo-Codierung, in systematische Bit und Paritätsbit umgesetzt. Von dem Sender wird an den Empfänger ein erstes Datenpaket gesendet, das die systematischen Bit und einen Teil der Paritätsbit enthält. Bei Vorliegen einer entsprechenden Aufforderung des Empfängers (2) wird mindestens ein Wiederholungsdatenpaket an den Empfänger (2) gesendet, das keine Paritätsbit enthält, die im ersten Datenpaket enthalten sind.
• – Ein erstes Wiederholungsdatenpaket enthält die Paritätsbit, die weder im ersten Datenpaket noch in einem weiteren Wiederholungsdatenpaket enthalten sind.
• – Das erste Wiederholungsdatenpaket enthält neben Paritätsbit, die im ersten Datenpaket nicht enthalten sind, systematische Bit.
• – Das zweite Wiederholungsdatenpaket enthält die Paritätsbit, die weder im ersten Datenpaket noch im ersten Wiederholungsdatenpaket enthalten sind.
• – Die Kodierungsrate des Turbocoders liegt unter Berücksichtigung der Ratenanpassung zwischen 2/3 und ¾.
• – Die in einem Datenpaket oder Wiederholungsdatenpaket enthaltenen Paritätsbit werden durch einen Ratenanpassungsalgorithmus aus den aus der Turbocodierung resultierenden Paritätsbit ausgewählt.
• – Die Punktierungsrate des Ratenanpassungsalgorithmus liegt zwischen ½ und 5/9.
• – Der Initialwert der Fehlervariablen des Ratenanpassungsalgorithmus wird derart gewählt, dass das erste Wiederholungsdatenpaket keine Paritätsbits enthält, die im ersten Datenpaket enthalten sind.
• – Der Initialwert der Fehlervariablen des Ratenanpassungsalgorithmus wird derart gewählt, dass das erste Wiederholungsdatenpaket solche Paritätsbits enthält, die weder im ersten Datenpaket noch in einem weiteren Wiederholungsdatenpaket enthalten sind.
• – Der Ratenanpassungsalgorithmus enthält folgende Verfahrensschritte: a) Setzen der Fehlervariablen auf den Initialwert, b) Setzen des Bitindex auf das erste Bit, c) Subtraktion des Erniedrigungswertes der Fehlervariablen von der Fehlervariablen, d) Falls die Fehlervariable kleiner oder gleich 0 ist durchführen der Schritte e) bis f), e) Punktieren des durch den Bitindex angezeigten Bit, f) Addition des Erhöhungs-Wert zur Fehlervariablen, g) Erhöhen des Bitindex, h) wiederholen der Schritte c) bis g), bis der Bitindex die Anzahl der zu verarbeitenden Bit überschreitet.
• – Für den Einsatz des Ratenanpassungsalgorithmus zur Auswahl der Paritätsbit im ersten Datenpaket gilt: eini(r) = {(Xi – ⎣r·eplus/rmax⎦ – 1)modeplus} + 1,
• – bei dem für den Einsatz des Ratenanpassungsalgorithmus zur Auswahl der Paritätsbit im ersten Wiederholungsdatenpaket gilt: eini(r) = {(2Xi + 1 – ⎡(r + 1)·eplus/rmax⎤)modeplus} + 1,wobei gilt:

  e_ini(r)

  = der Initialwert für die Fehlervariable;

  r

  = 0, falls der Ratenanpassungsalgorithmus in Zusammenhang mit dem ersten Datenpaket angewendet wird;

  r

  = 1, falls der Ratenanpassungsalgorithmus in Zusammenhang mit dem Wiederholungsdatenpaket angewendet wird;

  rmax

  = 2.

• – Es gilt. Nt,sys = 3Ndata – 4Nsys;
  
  dabei bezeichnet

  Ndata:

  Verfügbare Bit für die Übertragung

  Nsys:

  Anzahl der systematischen Bit vor der Ratenanpassung

  Nt,sys:

  Anzahl der systematischen Bit nach der Ratenanpassung

  Nt,p1:

  Anzahl der Parity 1 Bit nach der Ratenanpassung

  Nt,p2:

  Anzahl der Parity 2 Bit nach der Ratenanpassung.

The functional section explained above, in particular the functional section 19 , may be set up so that one or more of the following configurations are put into practice:

• - For data transmission in the form of data packets, the data is converted by a channel coding, in particular a turbo-coding, into systematic bit and parity bit. The sender sends to the receiver a first data packet containing the systematic bits and part of the parity bits. If there is a corresponding request from the beneficiary ( 2 ) at least one repeat data packet is sent to the receiver ( 2 ) containing no parity bits contained in the first data packet.
• A first retry data packet contains the parity bits which are neither contained in the first data packet nor in a further retry data packet.
• The first retry data packet contains systematic bits in addition to parity bits that are not contained in the first data packet.
• The second repetition data packet contains the parity bits which are contained neither in the first data packet nor in the first repetition data packet.
• - The coding rate of the turbo coder is between 2/3 and ¾, taking into account the rate adaptation.
• The parity bits contained in a data packet or repetition data packet are selected by a rate matching algorithm from the turbo-coding resulting parity bits.
• The puncturing rate of the rate matching algorithm is between ½ and 5/9.
• The initial value of the error variable of the rate matching algorithm is chosen such that the first repeat data packet does not contain any parity bits contained in the first data packet.
• The initial value of the error variable of the rate matching algorithm is chosen such that the first repetition data packet contains such parity bits which are contained neither in the first data packet nor in a further repetition data packet.
• The rate matching algorithm contains the following method steps: a) setting the error variable to the initial value, b) setting the bit index to the first bit, c) subtracting the reduction value of the error variable from the error variable, d) if the error variable is less than or equal to 0, performing steps e) to f), e F) addition of the increment value to the error variable, g) incrementing the bit index, h) repeating steps c) through g) until the bit index exceeds the number of bits to be processed.
• The following applies to the use of the rate matching algorithm for selecting the parity bits in the first data packet: e ini (r) = {(X i - ·r · e plus / r Max ⎦ - 1) mode plus } + 1,
• In which the following applies to the use of the rate matching algorithm for selecting the parity bits in the first repeat data packet: e ini (r) = {(2X i + 1 - ⎡ (r + 1) · e plus / r Max ⎤) mode plus } + 1, where:

  e _ini (r)

  = the initial value for the error variable;

  r

  = 0 if the rate matching algorithm is applied in conjunction with the first data packet;

  r

  = 1 if the rate matching algorithm is applied in conjunction with the retry data packet;

  rmax

  = 2.

• - It applies. N t, sys = 3N data - 4N sys ;
  
  thereby designated

  N data:

  Available bits for transmission

  Nsys:

  Number of systematic bits before rate matching

  Nt, sys:

  Number of systematic bits after rate matching

  Nt, p1:

  Number of parity 1 bit after the rate adaptation

  Nt, p2:

  Number of parity 2 bits after rate matching.

Above were standing for two areas indicated procedures to ensure optimal selection of Parity bit of retransmissions to reach. Depending on the encoding rate, the parity bit in the first retransmission selected so that they are just the gaps in the first transmission fill, or the gaps, taken together in first transmission and second retransmission consist. Depending on the code rate, they are listed in the following table used in column 1 or 2 parameters.

This method is also described in the standardization documents:
3GPP TSG-RAN 1 Meeting # 41 R1-050398
Athens, Greece, 9-13 May, 2005
3GPP TSG-RAN 1 Meeting # 41 R1-050399
Athens, Greece, 9-13 May, 2005.

The latter describes this selection and the former shows how the formulas be introduced in the already mentioned specification 25.212 can, in particular in chapter 4.8.4.3 "HARQ Rate Matching Stage"

In the diagram 3 are the numbers of bits transmitted in the (first) retransmission, differentiated according to parity 1, parity 2 and systematic bit.

P1 TX2:

Anzahl der in der Wiederholungsübertragung übertragenen Parity 1 bit (durchgezogen)

P2 TX2:

Anzahl der in der Wiederholungsübertragung übertragenen Parity 2 bit (durchgezogen)

S TX2:

Anzahl der in der Wiederholungsübertragung übertragenen systematischen bit (strich-punkt Linie)

Where:

P1 TX2:

Number of parity 1 bit transmitted in retransmission (solid)

P2 TX2:

Number of parity 2 bits transmitted in retransmission (solid)

S TX2:

Number of systematic bit (dash-dot line) transferred in retransmission

The V marked curves denote the corresponding magnitudes for a further developed embodiment.
V P1 TX2 (dotted)
V P2 TX2 (dotted)
VS TX2 (dash-dot-dot line)

The diagram looks at the case where N _sys = 100. The number of bits transmitted via the air interface is plotted on the X axis (Ndata). It can be seen that in the range of Ndata from 199 down to 150, the number of parity bits to be transmitted increases and thus fewer and fewer systematic bits are transmitted. The number of parity 1 and parity 2 bit differs only very slightly, mainly due to the different rounding, so the two lines look like a "twisted" line: starting from 149 (and less bit) in two transmissions, not all Instead of only three transmissions, this has the disadvantage that only relatively few parity bits can be transmitted at this limit.

It is now or another object of this invention to overcome this drawback, ie to provide that at (something) is less than 149 bits, more specifically in the range of 149 bits down to 140 bits, or more generally, at a coding rate between 2/3 and 5/7 to ensure that in the retransmission as many parity bits can be sent that have not already been sent in the first transmission. This object is achieved by the claims, the method is described in more detail below:
In another embodiment, one can mitigate this disadvantage by treating the two classes of parity bits differently: one of the classes continues to transmit so many bits that, along with the first transmission, all of the parity bits of that class have been transmitted, while one of the So many others send that after three transmissions all parity bits of that other class are transmitted. In other words, if the coding rate becomes so high that it is no longer possible to select both parity 1 and parity 2 bits according to column 1 of the above table, then both classes according to column 2 are not selected immediately, but only one class at first. The other class can still be transmitted even with less transmitted bit (smaller Ndata) still so that even after the second transmission all bits of this class have been transmitted. By this individual selection, that is for each class of parity bit individually determined whether already after 2 or only after 3 transmissions all bits of this class are to be transmitted, one gains a further flexibility and can thereby exploit the available number of bits even better ,

This can be done so that the condition for the application of column 2 (instead of column 1) is initially applied only for parity 1 bit. For parity 2 bit, you only switch from column 1 to column 2 if this is necessary because there are not enough bits left. This can be determined as follows:

It is possible to transmit the parity 2 bits according to column 1 and the parity 1 bits according to column 2, as long as N _{t, p} 1 + N _{t, p} 2 ≤ N _data follows from this

7Nsys ≤ 5Ndata Assuming that Ndata and Nsys are both even (or both odd), one can dispense with the rounding brackets and get:

7N sys ≤ 5N data

This means that the coding rate is Coding rate = N sys / N data ≤ 5/7 ≈ 0.714

This can be used to rewrite the table:

By comparing the parameters used for parity 1 bit and parity 2 bits in the columns V with the values in the columns 1 and 2, one sees that the parity 1 bits used are calculated in the newly inserted column V as in column 2, the parity 2 bit but as in column 1. The systematic bits to be transmitted are then always the remaining ones and are calculated as N t, sys = N data - N t, p1 - N t, p2

undOf course, this table can also be written individually for parity 1 and 2 bits:

and

The effect of this optimization can also be seen in the upper diagram, in the range of 140 = <N data <150

There the numbers of parity 1 and parity 2 bits are shown in dashed lines, one sees that in this area then significantly more Parity 2 than Parity 1 bit to be sent.

As a further embodiment, it is of course also possible to use the parity 1 bit (instead of the parity 2 bit as shown above). This then gives the following formulas: For the parity 1 bit, it only switches from column 1 to column 2, if this is necessary because there are not enough bits left. This can be determined as follows:

It is possible to transfer both the parity 1 bit according to column 1 and also the parity 2 bit according to column 2, as long as N _{t, p 1} + N _{t, p} 2 _≦ N _data follows from this

daraus folgt wiederum
7N_sys ≤ 5N_data und die Kodierrate ist
Kodierrate = N_Sys/N_data ≤ 5/7 ≈ 0,714Assuming that Ndata and Nsys are both even (or both odd), you can omit the rounding brackets, otherwise both rounding brackets cause the stapled value to increase by 0.5 each. But since the first bracket is multiplied by -2, the second by +1, the rounding causes the left side to be decremented by 0.5. If the inequality is thus fulfilled, neglecting the rounding clips, then it is even more satisfied with the rounding clips (since the left side is lowered by 0.5 by the rounding clips). Thus, a sufficient (but not necessarily necessary) condition for the above inequality is:

it follows again
7N _sys ≤ 5N _data and the coding rate is
Coding rate = N _Sys / N _data ≤ 5/7 ≈ 0.714

There the rounding in this case leads to a sufficient condition is It is altogether advantageous to use this variant.

This allows the table to be written like this:

undBy comparing the parameters used for parity 1 bit and parity 2 bits in the columns V with the values in the columns 1 and 2, one sees that the used parity 1 bits are calculated in the newly inserted column V as in column 1, but the parity 2 bits are the same as in column 2. The systematic bits to be transmitted are then the remaining ones and are calculated as N _{t, sys} = N _{data -N t, p 1 -N t, p 2} . Of course, this table can also be written individually for parity 1 and 2 bits:

and

Similar optimizations can be applied in other areas as well: For N _data > 166, for example, all parity 2 bits can be sent, whereas for the parity 1 bit only those that were missing in the first transmission can be sent. However, this is probably not as advantageous as the previously described optimization in the range of the coding rate between 2/3 and 5/7 and is therefore not further elaborated here.

at Rates over ¾ can to optimize the parity 1 bit so that in the retransmission and n1 transmissions with s = 1 all parity 1 bit transmitted and optimize the parity 2 bit so that in the retransmission and n2 transfers with s = 1 all parity 2 bits are transmitted become. In particular, it is possible to select n1 not equal to n2. This then results in the curve shown.

Of another one can in a further embodiment, the parity 1 Optimize bit so that in n01 transmissions with s = 0 (belongs to the retransmission) and n11 transmissions with s = 1 all parity 1 bit transmitted and optimize the parity 2 bit so that in n02 transfers with s = 0 (including the retransmission) and n12 transmissions with s = 1 all parity 2 bits are transmitted become. It can in particular n01, n11, n02 and n12 are chosen unequal.

As a further embodiment, at a coding rate greater than 0.75, the following procedure may also be used: Of the parity bits of the one class, as many bits as indicated above in column 2 are transmitted. The remaining bits are then padded by the other class. Thus, at least for one class, it is achieved that the bits are sent well nested with those of the first and third transmissions. The formulas are pretty simple for that. From a certain coding rate (in comparison to the second transmission) so few bits are transmitted in the first and third transmission that it is no longer worthwhile to synchronize the bits of the second transmission with it. Then just half parity 1 and parity 2 bit should be sent. At most, by optimizing the parameter eini still some optimization could be made, so that the smallest possible overlap of the transmitted in the retransmission bit with the first transmission (and possibly the second retransmission transmitted bit) results. The resulting curve is in the diagram 4 given for an example. It can be seen that the number of bits transmitted in the range from 133 down to (in this example) 126 still increases the number of parity 2 bits, whereas the number of parity 1 bits accordingly decreases. From 125 (and less) transmitted bits on the other hand (except for rounding) the same amount of parity 1 and parity 2 bits are transmitted.

The Invention was for some embodiments described. It should be noted, however, that they are not up these examples are limited is. Those who are versed in matter become according to the principle of the invention can implement further variants, which also by this Invention are covered. In particular, one can see the invention for the case Expand that more than two different classes of parity bits exist.

Claims (17)

Method for data transmission in the form of data packets, - in which data is converted into systematic bits and at least two classes of parity bits by a turbo-coding, - in which a first data packet, preferably containing systematic bits, is sent by the transmitter to the receiver contains part of the parity bit, and - if there is a corresponding request from the recipient ( 2 ) at least one retry data packet to the recipient ( 2 ), characterized in that the number of parity bits of a class in the retry data packet is chosen so that after a number of retry data packets all parity bits of that class are transmitted, this number of retry data packets not being equal for all classes. Method for data transmission in the form of data packets, in which data are converted into systematic bits by turbo-coding, a first class of parity bits and a second class of parity bits, in which a first data packet is sent by the transmitter to the receiver, which preferably contains systematic bits and contains a part of the parity bit, and - if there is a corresponding request from the recipient ( 2 ) at least one retry data packet to the recipient ( 2 ), characterized in that the number of parity bits of the first class in a first repeating data packet is selected so that after a repeating data packet all parity bits of the first class are transmitted, and the number of parity bits of the second class in a first and a second repeating data packet it is selected that after two repetition data packets all parity bits of the second class are transmitted. Method according to one of the preceding claims, - in which the first repeating data packet next to the parity bit in the first data packet not included, contains systematic bit. Method according to one of the preceding claims, - in which the second retry packet contains the parity bits of a class that neither in the first data packet nor in the first retry data packet are included. Method according to one of the preceding claims, - in which the coding rate of the turbo coder taking into account the rate adaptation between 2/3 and 5/7 lies. Method according to one of the preceding claims, - in which those contained in a data packet or retry data packet parity by a rate matching algorithm from the turbo coding resulting parity bit selected become. Method according to one of the preceding claims, - in which the puncturing rate of the rate matching algorithm between ½ and 5/9 lies. Method according to one of the preceding claims, - in which the initial value of the error variable of the rate matching algorithm chosen like that is that the first repeat data packet no parity bits one Contains class, which are contained in the first data packet. Method according to one of the preceding claims, - in which the initial value of the error variable of the rate matching algorithm chosen like that is that the first repeat data packet such parity bits a Contains class, neither in the first data packet nor in another repeat data packet are included. Method according to one of Claims 6 to 9, - in which the rate matching algorithm comprises the following method steps: a) setting the error variable to the initial value, b) setting the bit index to the first bit, c) subtracting the error value of the error variable from the error variable, d) if the error variable is less than or equal to 0, performing steps e) to f), e) puncturing the bit indicated by the bit index, f) adding the increment value to the error variable, g) increasing the bit index, h) repeating steps c) to g) until the bit index exceeds the number of bits to be processed. Method according to one of the preceding claims, - in which the following applies to the use of the rate matching algorithm for selecting the parity bits in the first data packet: e ini (r) = {(X i - ·r · e plus / r Max ⎦ - 1) mode plus } + 1, In which the following applies to the use of the rate matching algorithm for selecting the parity bit of the one class in the first repeat data packet: e ini (r) = {(2x i + 1 - ⎡ (r + 1) e plus / r Max ⎤) mode plus } + 1 In which the following applies to the use of the rate matching algorithm for selecting the parity bit of the other class in the first repeat data packet: e ini (r) = {(X i - ⎡ (r + 1) · e plus / r Max ⎤) mode plus } + 1, where: e _ini (r) = the initial value for the error variable; r = 0 if the rate matching algorithm is applied in conjunction with the first data packet; r = 1 if the rate matching algorithm is applied in conjunction with the repeat data packet; rmax = 2. Method according to one of the preceding claims, in which:

N t, sys = N data - N t, P1 - N t, p2 ; where Ndata is: available bits for transmission Nsys: number of systematic bits before rate adaptation Nt, sys: number of systematic bits after rate adaptation Nt, p1: number of parity 1 bit after rate adaptation Nt, p2: number of parity 2 bits after the rate adjustment. Method according to one of the preceding claims, - in which a first retry data packet containing parity bits of a class that neither in the first data packet nor in another repeat data packet are included. Method according to one of the preceding claims, - in which at least one repetition data packet is sent to the recipient ( 2 ) containing no parity bits contained in the first data packet. Transmitter for transmitting data in the form of data packets with a control device, in particular processor device, which is set up in such a way that data is converted into systematic bits and at least two classes of parity bits by a turbo-coding, that from the transmitter to the receiver first data packet is sent, preferably systematic Contains bit and contains a part of the parity bit, and - that if there is a corresponding request from the recipient ( 2 ) at least one retry data packet to the recipient ( 2 ), characterized in that the number of parity bits of a class in the retry data packet is chosen so that after a number of retry data packets all parity bits of that class are transmitted, this number of retry data packets not being equal for all classes. receiver for receiving data in the form of data packets, wherein the data by turbo coding into systematic bit and at least two Classes Parity bit are implemented, with a control device, in particular processor device, which is set up in such a way that - a first data packet, preferably systematic bit and one Part of the parity bit contains Will be received, - that a retry data packet is requested from the sender if that first data packet is not decoded correctly on the receiving side, - thereby marked that the number of parity bits of a class in the retry data packet so chosen will do that after a number of retry data packets all parity transferred to this class where this number of retry data is not pacts for all classes the same becomes. Method of data transmission in the form of data packets, in which data is converted by coding into bits of at least two classes, in which a first data packet is sent by the transmitter to the receiver, and in the case of a corresponding request from the receiver ( 2 ) at least one further data packet, a repeat data packet to the receiver ( 2 ), characterized in that the number of bits of a class in the retry data packet is chosen such that after a number of retry data packets all bits of that class are transmitted, this number of retry data packets not being equal for all classes.