JP5135603B2 - Retransmission method, transmitter and receiver using multilevel coded modulation - Google Patents

Description

  The present invention relates to a retransmission scheme using multilevel coded modulation.

  In order to achieve high reliability and a high data transfer rate in a limited frequency band, packet communication with high frequency utilization efficiency and high throughput is required. As one of means for realizing such communication, attention is paid to an HSDPA (High Speed Downlink Packet Access) technique that changes a modulation scheme and a code according to the state of a communication path. In HSDPA, a packet retransmission technique called HARQ (Hybrid Auto Repeat reQuest) that combines error correction code and retransmission is used (Non-patent Document 1).

  Also, in order to realize communication with high frequency utilization efficiency, application of coded modulation is essential. Coded modulation methods include trellis coded modulation method (TCM) (Non-Patent Document 2), multi-level code modulation method (Non-Patent Document 3), bit interleaved coded modulation method (BICM) (Non-Patent Document 4 and Three of 5) are known.

  In particular, the multi-level coded modulation method is a method proposed by Imai and Hirakawa in Non-Patent Document 3, and has recently attracted attention as a method for easily realizing high frequency utilization efficiency. This system has been studied for use in the fields of digital broadcasting and satellite broadcasting in Europe, for example. Further, Non-Patent Document 6 shows that the channel capacity can be approached by optimally designing the coding rate of each level of the multilevel coded modulation system.

S. Lin and D. J. Costello, Error Control Coding. Prentice Hall, second ed., 2004. G. Ungerboeck, “Channel coding with multilevel / phase signals,” IEEE Trans. Inform. Theory, vol. IT-28, pp. 5567, Jan. 1982. H. Imai and S. Hirakawa, “A new multilevel coding method using error correcting codes,” IEEE Trans. Inform. Theory, vol. 23, pp. 371377, May 1977. E. Zehavi, “8-PSK trellis codes for a Rayleigh channel,” IEEE Trans. Commun., Vol. 40, pp. 873884, May 1992. G. Caire, G. Taricco, and E. Biglieri, “Bit-interleaved coded modulation,” IEEE Trans. Inform. Theory, vol. 44, pp. 927946, May 1998. U. Wachsmann, R. F. Fischer, and J. B. Huber, “Multilevel codes: Theoretical concepts and practical design rules,” IEEE Trans. Inform. Theory, vol. 45, pp. 13611391, July 1999. Q. Luo and P. Sweeney, “Performance of bandwidth efficient multilevel harq schemes over wireless channels,” in Proc. PIMRC '04, vol. 4, pp. 25962600, Sept. 2004. Q. Luo and P. Sweeney, “Hybrid-ARQ protocols based on multilevel coded modulation,” Electron. Lett., Vol. 39, pp. 10631065, July 2003.

  However, when an error actually occurs in packet communication, a retransmission technique is essential. Therefore, an object of the present invention is to provide a retransmission scheme using multilevel coded modulation.

In order to achieve the above object, the present invention provides a retransmission method using multi-level coded modulation, wherein a multi-level coded modulation signal is transmitted; Decoding the signal at each level, performing error detection, requesting retransmission for the level in which the error was detected, and transmitting the code word of the level in which the error was detected between the signal points rather than the original transmission. and retransmitting a minimum distance is greater modulation scheme, the detected level of said error, looking contains a to decode a combination of the original signal and the retransmission signal, to the retransmission was detected in the error In addition to the level codeword, it is retransmitted in combination with a higher-level codeword .

The invention according to claim 2 is a transmitter that implements retransmission using multi-level coded modulation, and is a combination of an encoder that encodes bits of each level and a bit of a code word from each encoder A modulation unit that maps a modulation symbol to a modulation symbol, a receiver that receives a retransmission request at a level in which an error is detected from a receiver, and a codeword at a level at which the error is detected in response to the retransmission request. A retransmission control unit that retransmits from the modulation unit using a modulation scheme in which a minimum distance between signal points is larger than that of the transmission signal of the transmission signal , and the retransmission control unit includes, in addition to the codeword of the level where the error is detected, The modulator is retransmitted in combination with a higher level codeword .

According to a third aspect of the present invention, there is provided the transmitter according to the second aspect , wherein the encoder is an LDPC encoder.

The invention according to claim 4 is a receiver that implements retransmission using multi-level coded modulation, and receives a demodulating unit that receives and demodulates a multi-level coded modulation signal; and A decoder that decodes and detects an error at each level, a transmitter that transmits a retransmission request for the level in which the error is detected, and a minimum distance between signal points that is larger than the original transmission in response to the retransmission request A retransmission control unit that demodulates a signal retransmitted using a modulation scheme by the demodulation unit and decodes the signal in combination with the original signal by a corresponding decoder, in addition to the codeword of the level where the error is detected, It is characterized by being modulated in combination with a higher level codeword .

The invention according to claim 5 is the receiver according to claim 4 , wherein the decoder comprises:
It is an LDPC decoder.

  According to the present invention, a multi-level code modulation / multi-stage decoding (MLC / MSD: Multi-Level Coding / Multi-Stage Decoding) scheme is combined with a high-efficiency retransmission scheme, thereby providing excellent reliability and high throughput characteristics. Can be achieved. Non-Patent Documents 7 and 8 propose a method of retransmitting only an erroneous level by combining an MLC / MSD system and HARQ. In this document, a convolutional code is used for error correction, and CRC (Cyclic redundancy check) is used for error detection, and the overall throughput is improved by retransmitting only a level with an error.

  In the present invention, by using a low constellation at the time of retransmission, the reliability at the time of retransmission is improved and high throughput is realized. For example, if the original transmission uses 64-QAM, QPSK or 16-QAM is used for retransmission. The greatest advantage of this is that a codeword that has already been received and a retransmitted codeword can be easily combined as the sum of log likelihood ratio LLR (Log Likelihood Ratio), so that high reliability can be obtained efficiently. It is in. In addition, the reliability at the time of retransmission can be further improved by transmitting not only the erroneous level but also the higher level at the same time.

  In addition, in order to detect whether there is an error at each level, each code must be provided with an error detection capability. Generally, a CRC code that adds several parity bits is used. As an alternative, in the present invention, a low-density parity check code (LDPC) can be used as an element code for error correction. The LDPC code is one of codes that can approach the Shannon limit, and is itself a block code, and has a feature that error detection can be performed from the parity check matrix. In addition, since the coding rate can be selected flexibly, there is an advantage that it is suitable for MLC design.

(Outline of the present invention)
In the present invention, a high-efficiency retransmission scheme is combined with a multilevel coded modulation / multi-stage decoding (MLC / MSD) scheme, thereby achieving high throughput characteristics.

In multi-level code modulation (MLC), m binary information is encoded with codes having different coding rates and transmitted using 2 ^m multilevel symbols. Here, the m communication paths in the multi-valued symbol have different minimum distances. Here, a small minimum distance is referred to as a low level, and a large minimum distance is referred to as a high level communication path. In general, in MLC, a low-level binary channel has a low coding rate (and therefore high error correction capability), and a high-level binary channel has a high coding rate (and thus low error correction capability). ) Assign a code. Hereinafter, a case will be described in which the number of levels is m = 3 and one-dimensional 8-ASK (Amplitude Shift Keying) is used as multi-level modulation, but it should be noted that the present invention is not limited to this. For example, if 8-ASK is applied to each of the I and Q components, multilevel coding modulation using two-dimensional 64-QAM can be configured.

  FIG. 1 is a diagram illustrating an example of multilevel coded modulation. In this example, as shown in the figure, 3 bits are assigned to each signal point on the 8-ASK constellation using natural mapping. In this case, since the least significant bit of the low level changes for each adjacent signal point, the minimum distance is the smallest and an error is likely to occur. On the other hand, the most significant bit of the high level changes every 4 signal points, so that the minimum distance is the largest and an error is unlikely to occur. Therefore, for the low level encoder 0, the encoding rate is lowered to increase the error correction capability, and for the high level encoder 2, the encoding rate is increased to reduce the error correction capability. This enables efficient transmission while ensuring data reliability.

Here, if the code lengths of the codewords x ⁽⁰⁾ , x ⁽¹⁾ , and x ⁽²⁾ of the encoder are equal, all bits can be allocated to 8 signal points of 8-ASK without excess or deficiency. Therefore, the number of bits input to the encoder is adjusted in consideration of the coding rate of each encoder. Specifically, as shown in FIG. 1, the information bit string q is divided into three blocks q ^{(0 ) so that} the code lengths of the codewords x ⁽⁰⁾ , x ⁽¹⁾ , x ⁽²⁾ of each encoder are equal. ⁾ , Q ⁽¹⁾ and q ⁽²⁾ . Finally, the multilevel code-modulated signal is transmitted to the receiver via the transmission line.

  On the receiving side, the reception signal subjected to multilevel code modulation is demodulated and decoded by multistage decoding (MSD). In MSD, decoding performance can be improved by using a codeword error-corrected at a low level for decoding a high-level codeword.

FIG. 2 shows a configuration of a receiver that demodulates and multistage-decodes a multilevel code-modulated signal. The received signal X ′ is separated into each level for each bit and fed to the respective decoders. The decoder 0 decodes the codeword x ′ ⁽⁰⁾ received at level 0. When the codeword x ′ ⁽⁰⁾ is decoded, the least significant bit of the three bits on the 8-ASK constellation is determined. Therefore, decoding reliability can be improved by performing decoding at the next level using this decoding result. Specifically, when the least significant bit is determined, the 8-ASK signal points that can be taken at the next level are reduced to four. Therefore, the decoder 1 may decode the codeword x ′ ⁽¹⁾ received at level 1 for the remaining four signal points. Thereby, the reliability is improved as compared with the case where the received codeword is decoded for all eight signal points.

Similarly, when codewords x ′ ⁽⁰⁾ and x ′ ⁽¹⁾ are decoded, the least significant bit and the second bit among the three bits on the 8-ASK constellation are determined. Once these bits are established, the possible 8-ASK signal points at the next level are reduced to two. Therefore, the decoder 2 decodes the codeword x ′ ⁽²⁾ received at level 2 for the remaining two signal points.

  However, since the MSD uses the result of decoding at a low level for high level decoding, if a decoding error occurs at a low level, it cannot be correctly decoded at a high level. Therefore, in the present invention, an error after decoding is detected for each level, and if an error is detected at a certain level, only that level or a combination of that level and a higher level is assigned between signal points rather than 8-ASK. Retransmit with modulation with a large minimum distance. Thereby, it is excellent in reliability and high throughput can be achieved.

(System configuration of the present invention)
Hereinafter, a system configuration in the case of one-dimensional M-ASK modulation will be described, but the present invention is not limited to this. In an actual system, for example, two-dimensional M ² -QAM modulation can be employed. In multi-level coded modulation (MLC), the M-ASK constellation is composed of m = log ₂ M binary channels, and on the decoding side, ASK symbols are decomposed into m-level binary channels. . The system configuration of the transmitter and receiver in this case will be described below.

  FIG. 3 shows a configuration example of a transmitter according to an embodiment of the present invention. The transmitter 300 includes a partitioning unit 302 that divides the information bit string q into m blocks, and m encoders 0, 1,. . . , M−1, a modulation unit 304 that maps encoded bits from each encoder on the M-ASK constellation, an ARQ receiver 308 that receives a retransmission instruction from the receiver, and ARQ that performs control during retransmission And a control unit 306.

The partitioning unit 302 divides the information bit string q into a binary bit string q ^(l) of m blocks. Here, q ^(l) (l = 0, 1,..., M−1) indicates a bit transmitted on the l-th binary communication path. At level l, encoder l encodes information bit q ^(l) with a code having coding rate R _l and code length n _l and outputs codeword x ^(l) . Here, if the code lengths of the encoders are made equal, all codewords can be transmitted with the number of ASK transmission symbols n equal to the number of bits of the code length. That is, n ₀ = n ₁ =... = N _m−1 = n. At this time, the overall coding rate R is expressed by the following equation.
(1) R = R ₀ + R ₁ +... + R _m-1

The modulation unit 304 includes codewords x ⁽⁰⁾ , x ⁽¹⁾ ,. . . , X ^(m−1) bits are combined and mapped onto the M-ASK constellation. In the binary labeling of MLC, the minimum distance between signal points can be appropriately set by using the natural mapping described above instead of the Gray mapping used in bit interleave coded modulation (BICM) or the like. The ASK-modulated signal is sent from the transmitter to the receiver via a wireless transmission path.

  If the SNR (Signal to Noise Ratio) of the transmission path is sufficiently high and the influence of noise can be ignored, retransmission is not necessary if the code is ideal. However, in a real system, errors occur due to noise and other effects. Therefore, retransmission control (ARQ) is required. A combination of encoding capable of error correction and retransmission control is referred to as hybrid ARQ (HARQ). In HARQ, when a packet that cannot be corrected on the reception side and has an error is detected, the packet is notified to the transmission side, and the packet in which the error is detected is retransmitted from the transmission side. This HARQ generally has type I and type II. In type I, a packet in which an error is detected is retransmitted. In type II, the parity bit of each packet is retransmitted. In general, since a retransmission packet includes not only erroneous information bits but also correctly received information bits, it is inefficient if the number of erroneous information bits is small.

In the present invention, using multilevel coding modulation in which a packet is divided into a plurality of levels and transmitted, not all levels are retransmitted at the time of retransmission, but the level at which an error is detected is retransmitted with a lower constellation. . As a result, unnecessary retransmission can be suppressed, and more reliable retransmission can be performed. Specifically, the ARQ receiver 308 receives a retransmission instruction from the receiving side. In accordance with this retransmission instruction, ARQ control section 306 controls modulation section 304 to transmit a codeword at a level where an error is detected by 2-ASK (BPSK) modulation. Alternatively, the ARQ control unit 306 controls the modulation unit 304 so as to transmit with a modulation suitable for transmitting not only the codeword at the level where the error is detected but also the higher-level codeword. For example, if an error is detected in the level of the m-2, and m-2 levels codeword x ^(m-2), the codeword x level m-1 is the upper level ^{(m-1 )} Is transmitted by 4-ASK modulation.

  FIG. 4 shows a configuration example of a receiver according to an embodiment of the present invention. The receiver 400 includes a demodulator 402 that demodulates the received signal X 'and m decoders 0, 1,. . . , M−1, an error detection unit 404 that detects an error in the decoded bit string from each decoder and outputs an information bit string q ′, and an ARQ transmitter 408 that transmits a retransmission instruction to the transmission side when an error is detected And an ARQ control unit 406 that performs control at the time of retransmission.

  The demodulator 402 demodulates the reception signal X ′ that has been subjected to multilevel code modulation in accordance with M-ASK. The demodulated bits are stored in the decoders 0, 1,. . . , M−1. Each decoder 0, 1,. . . , M−1, encoders 0, 1,. . . , M−1. For example, an LDPC (Low Density Parity Check) code having characteristics close to the Shannon limit can be used for encoding of each encoder. Since the LDPC code can detect not only errors but also errors, the result can be used for retransmission control. When a turbo code is used, it is necessary to add a CRC (Cyclic Redundancy Check) bit as error detection for retransmission control.

The decoder l decodes the demodulated bits at level l. In the case of an LDPC code, the decoder 1 calculates the log likelihood ratio (LLR) of the received word x ′ ₀ ^(l) by the original transmission. This is represented as L (x ′ ₀ ^(l) ). Sum product decoding is performed using this LLR. When the error detection unit 404 finds that the decoding has not been correctly performed as a result of the parity check, the information is notified from the ARQ transmitter 408 to the transmitter 300, and the codeword x ^(l) is retransmitted. In this specification, ARQ that retransmits only codewords at a level where an error is detected is referred to as mode 1 in this specification.

At the time of retransmission, a suitable modulation method is selected by the ARQ control unit 306 of the transmitter 300, and therefore, in the receiver 400, the ARQ control unit 406 controls the demodulation unit 402 to select a corresponding demodulation method. For example, when an error is detected at one level, a 2-ASK modulation scheme is selected, and when an error is detected at two levels, a 4-ASK modulation scheme is selected. Then, the decoder l calculates the LLR from the retransmitted received word x ′ ₁ ^(l) and combines it with the previous LLR as in the following equation.
^{(2) L (x '(} l)) = L (x' 0 (l)) + L (x '1 (l))
Using this L (x ′ ^(l) ), x ^(l) is decoded.

Here, it should be noted that since the code word, that is, the packet length n is the same in all levels and retransmissions, the code combination is simply the sum of the LLRs and is easy to calculate. . This process is repeated until the information is obtained without error in the decoder l. If the required number of retransmissions is N _r , the LLR when the decoder gets out of this iteration (iteration) is expressed by the following equation.

  This ARQ mode 1 can improve the reliability by retransmission for the level in which the error is detected. However, at the time of retransmission, the level in which the error is detected and the level one level higher than that are modulated by 4-ASK. Can also be used for transmission. In this specification, this ARQ is referred to as mode 2.

In the ARQ mode 2, from the pair of received words {x ′ _tl ^(l) , x ′ _tl ^{(l + 1)} retransmitted by the decoder l at the t _l- th time, L (x ′ _tl ^(l) ) and L ( x ′ _tl ^{(l + 1)} ) is derived. In this mode, not only the level at which an error is detected but also the next higher level are retransmitted, so that when decoding is performed at the upper level, the reliability is improved and the number of retransmissions can be reduced. However, since 4-ASK has a smaller minimum distance between signal points of transmission symbols than 2-ASK, this mode is effective in a region where the SNR is high, but a region where the SNR is low, that is, the minimum distance is small. It is not necessarily efficient in a region that becomes dominant in decoding performance.

Furthermore, the ARQ mode m can be defined by generalizing the above idea. That is, this is a mode in which a level 1 where an error is detected and all levels above it are retransmitted by modulation of 2 ^ml- ASK. The mode selection depends on parameters such as the modulation standard, the code used, and the SNR of the transmission path.

(simulation result)
The performance of the hybrid ARQ according to an embodiment of the present invention was verified by simulation. In this simulation, the modulation method was 8-ASK, and bit interleave coded modulation (BICM) was used as a comparison target. Table 1 shows the specifications of this simulation.

MLC / MSD transmits m codewords with a length of n, where n is the number of symbols in each packet, but BICM can take a longer code length. In particular, when 2 ^m -ASK and the codeword is n × m, the code length of BICM is m times that of MLC / MSD. For this reason, BICM sometimes shows better performance than MLC / MSD optimally designed by applying the channel capacity law described in Non-Patent Document 6 (hereinafter referred to as “capacity rule”). It is also possible.

Each coding rate R _l when a transmission rate R is given is optimized by applying a capacity rule. However, when the coding rate obtained from this rule is 1 at a high level, the coding rate is slightly lowered in consideration of decoding at the time of retransmission in applying the present invention.

In this simulation, 8-ASK is used for modulation, and m = 3 bits are transmitted per symbol. At that time, the coding rate of each level is designed according to capacity rules, and R ₀ = 0.5, R ₁ = 0.98, R ₂ = 0.98, and the overall transmission rate is R = 2. .46. Similarly, the BICM parameters are set to have the same transmission rate.

  In this simulation, two retransmission schemes were evaluated. FIG. 5 applies ARQ mode 1 at all levels. This is referred to herein as Scheme 1. On the other hand, FIG. 6 applies ARQ mode 2 to the lowest level 0 and applies ARQ mode 1 to the other levels 1 and 2. This is referred to herein as Scheme 2.

  FIG. 7 shows the bit error rate (BER) characteristics. From this figure, it can be seen that MLC / MSD and BICM show almost the same BER characteristics when there is no retransmission. In practice, BICM has slightly better characteristics. This is because the BICM has a longer code length, and the MLC / MSD code rate setting is slightly lower than the capacity rule of Non-Patent Document 6 in consideration of retransmission. Conceivable.

  Next, if one retransmission is allowed, a large difference appears in the BER characteristics. In Scheme 1, the characteristics are worse than in BICM, but in Scheme 2, the improvement in characteristics is greater than in BICM. This can be explained from the viewpoint of capacity described later. Specifically, first, attention is focused on a region where the SNR in FIG. 11 is about 7 dB to 13 dB. In scheme 2, ARQ mode 1 is applied to all levels, and by allowing one retransmission, the capacity curve of level 0 moves to the 2-ASK curve. As a result, the coding rate falls below the capacity value at level 0, and errors are reduced. However, the level 1 coding rate, which is one level higher, exceeds the level 1 capacity value. Therefore, the error cannot be corrected at level 1, and there is an error as a whole.

  On the other hand, in scheme 2, ARQ mode 2 is applied to level 0 retransmission. With one retransmission, the level 0 capacity curve moves to the level 0 4-ASK curve and the level 1 capacity curve moves to the level 1 4-ASK curve. As a result, the coding rates of both level 0 and level 1 are lower than the capacity value. As a result, the error is reduced as a whole in this SNR region.

  FIG. 8 shows the decoding error rate (DER) characteristics. This also shows the same result as the BER characteristic. Next, the average number of retransmissions required until information could be decoded without error was evaluated.

Figure 9 shows the characteristics of the average required number of retransmissions N _r. From this figure, SNR is in Scheme 1 in a low region N _r is the lowest it can be seen that show the best results. However, _Nr is larger than 5 and seems to be impractical when considering an actual application. Conversely, scheme 2 is better for regions with moderate SNR. In this region, N _r is less than 3. From the above, it can be seen that scheme 1 is better in the region where the SNR is low, but scheme 2 is more efficient in the region where the SNR is high. Therefore, it can be said that HARQ according to the present invention has better performance than that using BICM by appropriately switching modes or schemes according to SNR.

Finally, throughput characteristics were evaluated. In transmission, throughput is defined as follows:
(4) Throughput = (1-DER) R [bit / dimension]
In this simulation, since retransmission is repeated until the entire information is received without error, the throughput is redefined as follows.
(5) Throughput = R / (1 + N _r ) [bit / dimension]

  FIG. 10 shows the characteristics depending on the throughput. From this figure, it can be seen that Scheme 1 is most excellent in the region where the SNR is low. However, it is very low, about 0.2 [bit / dimension]. In the region where the SNR has increased, Scheme 2 is the best. Further, the reason why the throughput rapidly increases near the SNR of 13.5 dB can be explained from the capacity analysis described later. Referring to FIG. 11, when the SNR exceeds about 13 dB, the coding rate is lower than the capacity value. In other words, it is considered that there are no errors and there is no need for retransmission. This is the reason why the throughput suddenly increases when the SNR exceeds about 13.5 dB in FIG. Further, a constant throughput value is taken when the SNR is about 8 dB to 13 dB. This is because the coding rate is lower than the capacity value in the scheme 1 with two retransmissions and with the scheme 2 in one retransmission, and therefore no further retransmission is required for this SNR. Thus, it can be seen that HARQ according to the present invention has better performance than that using BICM by appropriately switching the retransmission scheme or mode according to SNR.

(Capacity analysis)
A theoretical analysis of the retransmission method according to the present invention from the viewpoint of channel capacity (capacity) is as follows. The relationship between capacity and coding rate is first paraphrased as follows. If the coding rate is lower than the capacity value at a certain SNR, communication can be performed without error at that SNR (depending on the performance of the code itself, how close it is to the Shannon limit). In the retransmission method according to the present invention, the coding rate of each level is always constant, and the constellation is changed at the time of retransmission. At this time, since the constellation is lowered, considering the capacity, it can be considered that the capacity curve has moved to the left.

  As an example, assume that the original transmission uses an 8-ASK constellation, and the coding rate exceeds the capacity value at a certain SNR, resulting in an error that cannot be corrected. If retransmission is performed with 2-ASK at the time of retransmission, it can be considered that the capacity is shifted from the capacity curve of 8-ASK to the curve of 2-ASK. That is, the capacity value increases at a certain SNR, and the coding rate falls below (although the coding rate exceeded the capacity value at the time of original transmission). This allows communication without error. Further, since LLR before retransmission is added at the time of decoding, it is considered that errors can be almost corrected as long as the capacity is satisfied.

  FIG. 11 shows the capacity in the above simulation using an 8-ASK constellation for the original transmission. In this figure, the solid lines drawn in parallel indicate the coding rates of 0.5 and 0.98 used in the simulation. The above example is illustrated in this figure. The capacity of 8-ASK and level 0 is shown in level 0 of 8-ASK (Non-Patent Document 6), and the coding rate of level 0 is 0.5. When the SNR is about 13.5 dB or less, the coding rate 0.5 exceeds the capacity value, and it cannot be said that communication can be performed without error. If retransmission is performed using 2-ASK in mode 1, the 8-ASK capacity curve at level 0 shifts to the 2-ASK curve. Then, the coding rate 0.5 is below the capacity curve up to about −3 dB, and it can be said that level 0 can be communicated without error in one retransmission.

  Next, consider a case where level 0 and level 1 are retransmitted using 4-ASK constellation in mode 2 at the time of retransmission. At this time, the coding rate of level 1 is 0.98. The capacity of 4-ASK and level 0 is indicated by 4-ASK of level 0, and the capacity of level 1 is indicated by 4-ASK of level 1 (Non-patent Document 6). As in the previous case, when the SNR is about 13.5 dB or less, the coding rates for both levels 0 and 1 exceed the respective 8-ASK capacity values. Assuming that 4-ASK retransmits, the capacity curves of level 0 and 1 shift from 8-ASK to 4-ASK, respectively. Then, the coding rate becomes lower than the capacity at the level 0 until the SNR is about 6 dB and the level 1 is about 6.9 dB. That is, if retransmission is performed in decoder 2 in mode 2, a gain of about 7.5 dB at level 0 and about 6.6 dB at level 1 is obtained.

  In this way, from the viewpoint of capacity, the retransmission mode may be selected so that the coding rate is lower than the capacity in a certain SNR. However, it must be noted that if any one level is wrong, there will be an error as a whole, so the coding rate of all levels must be set below the capacity. That is.

  While the present invention has been described with respect to several specific embodiments, the embodiments described herein are merely illustrative in view of the many possible embodiments to which the principles of the present invention can be applied. It is not intended to limit the scope of the invention. The configuration and details of the embodiment exemplified here can be changed without departing from the spirit of the present invention. Further, the illustrative components and procedures may be changed, supplemented, or changed in order without departing from the spirit of the invention.

It is a figure which shows an example of multilevel encoding modulation. It is a figure which shows an example of multistage decoding. It is a figure which shows the structural example of the transmitter by one Embodiment of this invention. It is a figure which shows the structural example of the receiver by one Embodiment of this invention. It is a figure which shows the structural example of the scheme 1 which applied ARQ mode 1 in all the levels according to this invention. It is a figure which shows the structural example of the scheme 2 which applied ARQ mode 2 to the lowest level 0 according to this invention, and applied ARQ mode 1 to the other levels 1 and 2. FIG. It is a figure which shows the characteristic of the bit error rate (BER) by simulation. It is a figure which shows the characteristic of the decoding error rate (DER) by simulation. It is a diagram showing a profile of an average required number of retransmissions N _r simulated. It is a figure which shows the characteristic of the throughput by simulation. It is a figure for demonstrating the capacity in simulation.

Explanation of symbols

DESCRIPTION OF SYMBOLS 300 Transmitter 302 Partitioning part 304 Modulation part 306 ARQ control part 308 ARQ receiver 400 Receiver 402 Demodulation part 404 Error detection part 406 ARQ control part 408 ARQ transmitter

Claims (5)

A retransmission method using multi-level coded modulation,
Transmitting a multi-level encoded modulation signal;
Decoding the signal at each level to detect errors;
Requesting a retransmission for the detected level of error;
Retransmitting the error-detected level codeword with a modulation scheme having a larger minimum distance between signal points than the original transmission;
In the detection level of said error, looking contains a to decode a combination of the original signal and the retransmission signal, to the retransmission, in addition to the code words of the detected level of the error, the upper level code Resending in combination with a word . A transmitter that implements retransmission using multi-level coded modulation,
An encoder that encodes each level of bits;
A modulation unit that maps bit combinations of codewords from each encoder to modulation symbols;
A receiver for receiving a retransmission request with a detected level of error from the receiver;
A retransmission control unit that, in response to the retransmission request, retransmits the codeword at the level in which the error is detected from the modulation unit using a modulation scheme in which a minimum distance between signal points is larger than that of an original transmission signal; And the retransmission control unit causes the modulation unit to retransmit in combination with a codeword of a higher level in addition to the codeword of the error detected level . The transmitter according to claim 2 , wherein
The transmitter is an LDPC encoder. A receiver that implements retransmission using multi-level coded modulation,
A demodulator that receives and demodulates the multi-level encoded modulation signal;
A decoder for decoding the demodulated signal at each level and detecting an error;
A transmitter for sending a retransmission request for the level of detected errors;
In response to the retransmission request, in addition to the codeword of the level where the error is detected, retransmitted using a modulation scheme having a minimum distance between signal points larger than that of the original transmission , A receiver comprising: a retransmission control unit that demodulates a signal modulated in combination by the demodulating unit and decodes the signal in combination with an original signal by a corresponding decoder. The receiver according to claim 4 , wherein
The receiver is an LDPC decoder.

Images