Hybrid ARQ method for packet data transmission

The present invention relates to a transmission device and a transmission method, and more particularly to a bit sequence modification method in ARQ retransmission in a communication system.

Techniques often used in communication systems with unreliable and time-varying line states are error correction based on automatic repeat request (ARQ) and forward error correction (FEC) techniques. It is called hybrid ARQ (HARQ). When a commonly used Cyclic Redundancy Check (CRC) detects an error, the receiver of the communication system requires the transmitter to retransmit the erroneously received data packet.

Non-Patent Document 1 and Non-Patent Document 2 define three different types of ARQ schemes.

-Type I: Received packets containing errors are discarded, and a new copy of the same packet is resent and decrypted separately. Both old and new packets received are not combined.

-Type II: Received packets containing errors are not discarded, but are combined with additional retransmission packets for continued decoding. The retransmission packet has a relatively high coding rate (coding gain), and may be combined with the soft-information obtained from the previously stored transmission at the receiving unit.

-Type III: Same as Type II, but with the restriction that each retransmitted packet can be automatically decrypted. This means that the transmitted packet can be decrypted without being combined with the previous packet. This is useful when some packets are damaged and the information can hardly be reused. If all transmissions contain the same data, this can be considered a special case called a single redundancy version of HARQ Type III.

Type II and Type III methods are clearly intelligent and superior in performance to Type I because they can reuse information from packets containing previously received errors. There are basically the following three methods for reusing the redundancy of previously transmitted packets. Soft-Combining -Code-Combining Combination of soft synthesis and code synthesis There is.

<u style="single">Soft synthesis</u> Using soft synthesis, the retransmission packet carries the same information that was previously received. In this case, for example, as disclosed in Non-Patent Document 3 or Non-Patent Document 4, a plurality of received packets are either symbol-by-symbol basis or bit-by-bit basis. Synthesize with. By synthesizing the soft judgment values from all the received packets, the reliability of the transmission bit increases in proportion to the number and power of the received packets. From the perspective of the decoder, the same FEC method (with a constant code rate) is used for all transmissions. Therefore, since the decoder only looks at the synthesized soft judgment value, it is not necessary to know the number of times the retransmission is executed. This method requires that all outgoing packets carry the same number of symbols.

<u style="single">Code synthesis</u> Code synthesis concatenates received packets to generate new codewords (the coding rate decreases as the number of transmissions increases). Therefore, the decoder needs to know the FEC method to be applied for each retransmission. Since the length of the retransmitted packet can be changed according to the line condition, code synthesis is more flexible than soft synthesis. However, code synthesis requires more transmission signal data than soft synthesis.

<u style="single">Combination of soft synthesis and code synthesis</u> If the retransmission packet contains the same symbol as the previously transmitted symbol and a different sign symbol, the same sign symbol is synthesized using the soft synthesis described in the "Soft Synthesis" section, and the remaining sign symbols are It is synthesized using code synthesis. The signal requirements here are similar to the signal requirements for code synthesis.

For example, as shown in Non-Patent Document 5, HARQ performance for Trellis Coded Modulation (TCM) can be enhanced by changing the symbol constellation for retransmission. In that case, the changes are made on a symbol-by-symbol basis, so performance gains can be gained by maximizing the Euclidean distance between the mapped symbols through retransmissions.

<p><nplcit num="1"><text>S. Kallel, "Analysis of a type II hybrid ARQ scheme with code combining", IEEE Transactions on Communications, Vol.38, No.8, August 1990</text></nplcit><nplcit num="2"><text>S. Kallel, R. Link, S. Bakhtiyari, "Throughput performance of Memory ARQ schemes", IEEE Transactions on Vehicular Technology, Vol.48, No.3, May 1999</text></nplcit><nplcit num="3"><text>D. Chase, "Code combining: A maximum-likelihood decoding approach for combining an arbitrary number of noisy packets", IEEE Trans. Commun ., Vol. COM-33, pp. 385-393, May 1985</text></nplcit><nplcit num="4"><text>BA Harvey, S. Wicker, "Packet Combining Systems based on the Viterbi Decoder", IEEE Transactions on Communications, Vol.42, No. 2/3/4, April 1994</text></nplcit><nplcit num="5"><text>MP Schmitt, "Hybrid ARQ Schemes TCM and Packet Combining", Electronics Letters, Vol. 34, No. 18, September 1998</text></nplcit></p>

<p> Considering the higher order modulation method (when the number of bits carried by the modulation symbol exceeds 2 bits), the synthesis method using soft synthesis has a big drawback. That is, the reliability of the bits in the soft-synthesized symbol is a constant percentage for all retransmissions. In other words, previously received transmission-based bits that are unreliable are unreliable even after receiving further transmissions, as are previously received transmission-based bits that are highly reliable. Is reliable even after receiving further transmissions.</p><p> The change in bit reliability is due to the constraints of two-dimensional signal constellation mapping, assuming that modulation schemes that carry more than 2 bits per symbol have equal transmission likelihood for all symbols. , Not all bits can have the same average reliability. The term average reliability ultimately means the reliability of a particular bit for every symbol of the signal constellation.</p><p> A bit mapping sequence i₁q₁i₂q₂Using the signal constellation for the 16QAM modulation scheme according to Figure 1, which shows the Gray coded signal constellation of, the bits mapped to the symbol differ from each other in the average reliability of the first transmission of the packet. Specifically, bit i₁And q₁Has high average reliability because it is mapped to half the space of the signal constellation diagram. Therefore, their reliability is irrelevant to the fact that the bit sends a "1" or a "0".</p><p> On the other hand, bit i₂And q₂Has a low average reliability because its reliability depends on the fact that the bit sends a "1" or a "0". For example, bit i₂In the case of, "1" is mapped to the outer column and "0" is mapped to the inner column. Similarly, bit q₂In the case of, "1" is mapped to the outer row and "0" is mapped to the inner row.</p><p> Bit reliability remains constant with each other in the second and subsequent retransmissions, depending on the signal constellation used in the first retransmission. That is, bit i₁And q₁Bit i no matter how many times it is retransmitted₂And q₂Always has a higher average reliability than.</p><p> In the simultaneous pending PCT / EP01 / 01982, it was proposed that a method of equalizing or almost equalizing the average bit reliability after receiving each transmitted packet is extremely useful for improving the performance of the decoder. There is. Therefore, the bit reliability is adjusted through retransmissions so that the average bit reliability is averaged. This is achieved by selecting a predetermined first signal constellation and at least a second signal constellation for transmission so that the combined average bit reliability for each bit of all transmissions is approximately equal.</p><p> Therefore, a change bit mapping is obtained by changing the signal constellation (constellation rearrangement). Here, the Euclidean distance between the modulated symbols can be changed for each retransmission by moving the constellation point. As a result, the average bit reliability can be freely manipulated and averaged, thereby improving the performance of the FEC decoder in the receiving unit.</p><p> In the solution proposed above, the benefit of constellation rearrangement is a parameterized, bit-to-symbol mapping entity (hereinafter referred to as the "bit-to-symbol mapping entity"). Realized by. For complexity or efficient implementation, it is advantageous for the communication system to have a non-parameterized standard mapping device.</p><p> Therefore, an object of the present invention is to provide a transmission device and a transmission method capable of having higher error correction capability without having a parameterized bit-to-symbol mapping device.</p>

<p> The transmitter according to the first aspect of the present invention is a bit sequence (i).₁q₁i₂q₂), And based on the table, the bit sequence (i) prior to the modulation mapping.₁q₁i₂q₂) I₁And i₂Swap the position of, q₁And q₂Swap the positions of i₁And q₁The bit sequence changing unit that changes the bit sequence by logically inverting the above, and (a) the bit sequence (i) at the time of the first transmission.₁q₁i₂q₂) Is transmitted, and (b) a transmission unit that transmits the bit sequence changed by the bit sequence changing unit and the information indicating the change at the time of retransmission is adopted.</p><p> Further, the transmitter according to the second aspect of the present invention is a bit sequence (i).₁q₁i₂q₂), And based on the table, the bit sequence (i) prior to the modulation mapping.₁q₁i₂q₂) I₁And i₂Swap the position of, q₁And q₂Swap the positions of i₂And q₂The bit sequence changing unit that changes the bit sequence by logically inverting the above, and (a) the bit sequence (i) at the time of the first transmission.₁q₁i₂q₂) Is transmitted, and (b) a transmission unit that transmits the bit sequence changed by the bit sequence changing unit and the information indicating the change at the time of retransmission is adopted.</p><p> The receiving device according to the third aspect of the present invention is a bit sequence (i).₁q₁i₂q₂A table showing multiple change patterns of) and (a) the bit sequence (i) transmitted at the time of the first transmission.₁q₁i₂q₂) Is received, and (b) the bit sequence (i) transmitted at the time of retransmission, respectively.₁q₁i₂q₂) I₁And i₂Swap the position of, q₁And q₂Swap the positions of i₁And q₁The i of the modified bit sequence received by the receiver based on the table and a receiver that receives the modified bit sequence by logically inverting₁And q₁Swap the positions of i₂And q₂Swap the positions of i₁And q₁A configuration is adopted having a bit sequence changing unit that changes the changed bit sequence by inverting the logical value of.</p><p> The receiving device according to the fourth aspect of the present invention is a bit sequence (i).₁q₁i₂q₂A table showing multiple change patterns of) and (a) the bit sequence (i) transmitted at the time of the first transmission.₁q₁i₂q₂) Is received, and (b) the bit sequence (i) transmitted at the time of retransmission, respectively.₁q₁i₂q₂) I₁And i₂Swap the position of, q₁And q₂Swap the positions of i₁And q₁The i of the modified bit sequence received by the receiver based on the table and a receiver that receives the modified bit sequence by logically inverting₁And q₁Swap the positions of i₂And q₂Swap the positions of i₂And q₂A configuration is adopted having a bit sequence changing unit that changes the changed bit sequence by inverting the logical value of.</p><p> The idea of the present invention is to modify the bit sequence before inputting it to the mapping device. This modification to the signal constellation uses, for example, a logical reverser and / or a placement transducer that logically inverts the bits of the signal constellation and / or transforms its bit arrangement according to the transmission count parameter m. Can be achieved with. Therefore, the beneficial effects of constellation rearrangement are achieved without the need for a parameterized bit-to-symbol mapping device. As a result, the sequence output after processing by the placement transducer, logic reverser, and non-parameterized standard mapping device is a parameterized bit-to-symbol using various constellation rearrangement schemes. Indistinguishable from the output of the mapping device.</p>

<p> According to the present invention, it is possible to have a higher error correction capability without having a parameterized bit-to-symbol mapping device.</p>

<figref num="1">The figure which shows an example of the signal constellation which shows the 16QAM modulation method using a Gray coded bit symbol.</figref><figref num="2">Diagram showing the first two of the four examples of signal constellations for 16QAM modulation schemes with Gray coded bit symbols.</figref><figref num="3">Diagram showing the remaining two of the four examples of signal constellations for the 16QAM modulation scheme with Gray coded bit symbols.</figref><figref num="4">The figure which shows one Embodiment of the communication system using the method which is the basis of this invention.</figref>

In order to further deepen the understanding of the present invention, preferred embodiments will be described below with reference to the accompanying drawings.

Hereinafter, the concept of log-likelihood-ratio (LLR) will be described as a measure of bit reliability. First, a simple calculation of the bit LLR within the mapped symbol for one transmission is shown. Then, the LLR calculation is extended for multiple transmissions.

<u style="single">Single Transmission</u> Symbols in channel transmission with Additive White Gaussian Noise (AWGN) and symbols of equal likelihood_nThe i-th bit b under the constraint that_nⁱAverage LLR is<maths num="1"><img file="JP4451920B2_D0001.tif" /></maths>Obtained at. Where r_n= s_nIs the symbol s_nIndicates the average received symbol under the constraint of sending (in the case of AWGN), d_{n, m}²Is the received symbol r_nAnd symbols s_mIndicates the square of the Euclidean distance between, E_s/ N_{o o}Indicates the observed signal-to-noise ratio.

From equation (1), LLR is the signal-to-noise ratio E_s/ N_{o o}And the Euclidean distance between the signal constellation points d_{n, m}It turns out that it depends on.

<u style="single">Multiple Transmission</u> Given multiple transmissions, the symbols s with independent AWGN channels and symbols of equal likelihood_n^(j)The jth bit b under the constraint of sending_n^jThe average LLR after the kth transmission is<maths num="2"><img file="JP4451920B2_D0002.tif" /></maths>Obtained at. Here, j is the j-th transmission ((j-1) -th retransmission). As with a single transmission, the average LLR depends on the signal-to-noise ratio and the Euclidean distance at each transmission.

Euclidean distance d without constellation rearrangement_{n, m}^(j)= d_{n, m}⁽¹⁾Is constant for all transmissions, so the bit reliability (LLR) after k transmissions is determined by the observed signal-to-noise ratio at each transmission and the signal constellation point of the first transmission. For higher level modulation schemes (more than 2 bits per symbol), the result is a change in the average LLR for the bits, which in turn results in different average bit reliability. Differences in average bit reliability continue throughout all retransmissions, resulting in poor decoder performance.

The case of the 16QAM method, which provides two high-reliability bits and two low-reliability bits, will be described below as an example. Here, for low reliability bits, reliability depends on the transmission of "1" or "0" (see Figure 1). Therefore, as a whole, there are two levels of average reliability, and the second level is further subdivided.

Level 1 (high reliability, 2 bits): Bit mapping for "1" ("0") is half the positive (negative) real space for i bits and half the imaginary space for q bits. Is separated into. Here, it makes no difference whether "1" is mapped to half of the positive space or half of the negative space.

Level 2 (unreliable, 2 bits): "1" ("0") is mapped to the inner (outer) column for the i bits or the inner (outer) row for the q bits Will be done. Level 2 is further classified because the LLR depends on the mapping to the inner (outer) columns and rows.

Level 2 a: i_nIn the inner column, q_nAre mapped to the inner rows respectively.

Level 2b: Inverse mapping of level 2a. i_nTo the outer column q_nTo each of the outer rows.

That is, when 16QAM is used as the modulation method, the first bit and the second bit of the 4-bit output bit string representing one symbol have high reliability, respectively. This is a bit in which each bit of the first output bit position and the second output bit position in the signal constellation specifies the quadrant of the IQ plane of the signal constellation, and the third output bit position and the fourth output bit position. Each bit of is equivalent to being a bit that identifies the placement of the signal constellation in each quadrant of the IQ plane.

The level of reliability needs to be changed to ensure an optimal averaging process through transmission for all bits.

The order of the bit mappings is not fixed before the first transmission, but must be consistent throughout the retransmission. For example, bit mapping for the first transmission: i₁q₁i₂q₂ Bit mapping for all retransmissions: i₁q₁i₂q₂Is.

Some examples of possible constellations are shown in Figures 2 and 3. Table 1 shows the results of bit reliability according to FIGS. 2 and 3.<tables num="1"><img file="JP4451920B2_D0003.tif" /></tables>

Hereinafter, m indicates the parameter of the number of retransmissions, and m = 0 indicates the first transmission of the packet having the ARQ configuration. In addition, b indicates the number of bits forming the symbol of the mapping device. In general, b is an arbitrary integer, and the most frequently used value in communication systems is 2 to the power of an integer.

Without loss of generality, it is assumed that the number of bits n used as the input of the bit arrangement conversion process is divisible by b, that is, n is an integral multiple of b. In other cases, those skilled in the art will appreciate that dummy bits can be easily added to the sequence of input bits until the above conditions are met.

As mentioned above, some modulations can identify several reliability levels. Therefore, in the bit arrangement conversion process, it is necessary to average the reliability of the b bits through retransmission so that all the b bits have the same reliability on average. This means that the placement transducer repositions the b bits in the symbol so that each bit in the symbol (b bits) is mapped to all reliability levels at the same frequency as all other bits in the b bits ( It means that it has to be replaced). This means that the bit arrangement conversion is an intra-symbol bit arrangement conversion process.

Also, there may be some bit positions whose reliability depends on the logical bit value (low or high). When mapping a bit to such a position except for the first time, a logical inversion must also be performed on that bit.

According to such a rule, it is possible to configure a pattern for determining the bit arrangement conversion and the logical bit value inversion process for the number of retransmissions m.

Theoretically, a perfect averaging of reliability is only possible after an infinite number of retransmissions or a very large number of retransmissions. Therefore, in such a case, there are several options in which the sequence of the bit arrangement conversion or the logical bit value inversion pattern is different. Since there is no difference in performance between them, it is up to the system designer to decide which option to select.

To maintain the signal constellation shown in FIG. 1 and obtain constellation 2 from constellation 1 in FIG. 2, the following process, · Original bit i₁And i₂Position exchange · Original bit q₁And q₂Position exchange · Original bit i₁And q₁Inversion of the logical bit value of Need to be executed. However, their execution order is not important.

Bits ending at position 1 and position 2 can also be logically inverted.

Table 2 below shows an example that depends on the number of transmissions. Here, a bit always means the first transmission, and a long dash on the character indicates the logical inversion of that bit.<tables num="2"><img file="JP4451920B2_D0004.tif" /></tables>

The first example in each row of Table 2 corresponds to the constellations shown in Figures 2 and 3. As is readily apparent from Table 2, signal constellation 2 is from signal constellation 1 to bit i.₁And bit i₂Position and bit q₁And bit q₂Swap the positions of and bit pair i₁And i₂Or it is obtained by logically inverting any of all bits. Similarly, signal constellation 3 is an option from signal constellation 1 to bit i.₁And bit i₂Position and bit q₁And bit q₂Swap the positions of each other, and bit pair i₂And q₂Is obtained by logically reversing. Another option is to only exchange bit positions and not require inversion of logical bit values. Finally, the signal constellation 4 is a bit pair i from the signal constellation 1 without exchanging any bit positions.₂And q₂Or it is obtained by logically inverting any of all the bits of the symbol.

This allows you to choose from different strategies for the number of transmissions (not all).<tables num="3"><img file="JP4451920B2_D0005.tif" /></tables>

FIG. 4 shows an embodiment of a communication system using the method underlying the present invention.

At transmitter 100, the bit sequence is obtained from an error correction / decoding (FEC) encoder (not shown) and then input to the placement transducer 110 and the logic inversion device 120. The placement transducer 110 and the logic reverser 120 each modify the input bit sequence depending on the retransmission count parameter m. The bit sequence is then input to the mapper / modulator 130, which is a non-parameterized standard mapping device. Mappers typically use one of the signal constellations shown in FIGS. 2 and 3 to map bit b to a symbol transmitted through communication channel 200. The communication channel 200 is generally a wireless communication channel that is in an unreliable and time-varying channel state.

The bit arrangement conversion / logical bit value inversion pattern is stored in both the transmitter and the receiver, or is stored in the transmitter and transmitted to the receiver.

At receiver 300, the complex symbol is first input to the demapper / demodulator 330, where the received symbol is demodulated into the corresponding bit region sequence (eg, LLR sequence). This sequence is then input to the logical inversion device 320 and then to the placement inverse converter 310. The obtained bit region sequence is output from the placement inverse converter 310.

The placement converter 110 and the placement inverse converter 310 operate according to a well-known technique of applying a specific pseudo-random or random permutation of an input bit or symbol sequence to perform a placement / reverse placement transformation, i.e., a sequence. Swap the positions of bits or symbols within. In the present embodiment, the arrangement converter 110 is an intra-symbol bit arrangement converter that changes the position of the bits forming the symbol in the mapping device.

Logical reversers 120,320 operate according to the well-known technique of inverting the logical value of a bit, that is, converting a logical low to a logical high value and vice versa. As a practical example of a receiver using the log-likelihood ratio, this inversion operation is equivalent to the sign inversion of the log-likelihood ratio.

When an automatic retransmission request issued by an error detector (not shown) initiates a retransmission and, as a result, the same data packet is transmitted from the transmitter 100, the demapper / demodulator 330 contains data containing previously received errors. The packet is soft-synthesized with the retransmitted data packet. By modifying the bit sequence by the placement transducer 110 and the logic reversing device 120, the average bit reliability is averaged and the receiver performance is improved.

The above method has been described using a Gray coded signal and a QAM modulation scheme, but it is a plausible that other suitable coding and modulation schemes can be used as well to benefit from the present invention. It is obvious to the trader.

100 transmitter 110 Placement transducer 120, 320 Logical Inverter 130 Mapper / Modulator 200 channels 300 receiver 310 Placement inverse transducer 330 Demapper / Demodulator