CN103916216B - A kind of QC LDPC code modulating methods in optical communication system based on 8 qam modes - Google Patents

Abstract

The invention discloses a kind of QC LDPC code modulating methods in a kind of optical communication system based on 8 qam modes, this programme is on the basis of the coded modulation scheme for studying existing 8 qam mode, using the collection partitioning scheme of UP BP BP, the Euclidean distance between signal is increased.Using not reciprocity protection coded system, first bit is encoded using the relatively low QC LDPC code types of redundancy, second and third bit is using the higher pattern coding of code check.In decoding, by multilevel coding/multilevel decoding MLC/MSD) scheme and multilevel coding/parallel decoding PDL schemes be combined, and obtains a kind of new QC LDPC coded modulation schemes.The computation complexity that the present invention proposes a plan for traditional scheme 2/3.Net coding gain (NCG) of the present invention is higher, and the error-correcting performance of proposed scheme is more excellent.

Description

A kind of QC-LDPC coded modulations in optical communication system based on 8-QAM modulation systems Method

Technical field

The invention belongs to based on 8-QAM modulation systems in field of signal processing in communicating, more particularly to optical communication system A kind of two benches QC-LDPC coded modulation schemes.

Background technology

Growing with optical communication technique, optical communication system is also therewith towards the longer distance of more Large Copacity more high speed side To growing, the 40Gb/s that single-channel optical communication system capacity develops into 2003 from the 10Gb/s of the nineties, use not To ten years.2010, ITU-T made the relevant criterion of 100Gb/s optical transmission systems, and relevant device has entered Enter industrialization test phase.While global main flow producer promotes 100Gb/s system globals to dispose, also actively carrying out 400Gb/s, 1Tb/s even beforehand research of 10Tb/s.Meanwhile, M system amplitude-shift keyings (Multiple Amplitude Shift Keying, MASK), M system phase shift keyings (Multiple Phase Shift Keying, MPSK), M Quadrature Amplitudes are adjusted System (Multilevel Quadrature Amplitude modulation, MQAM) can bring log₂The symbol utilization rate of M, Wherein M-ary QAM are a kind of power and the effective transmission means of bandwidth, in practice with most.In super 100Gb/s side In case, 8-QAM becomes a kind of popular scheme.At present in the method for various increase power system capacities, M-QAM is also a kind of joint The modulation system of the amplitude and phase place of control sinusoidal carrier.8-QAM therein becomes the popular tune of optical communication system of future generation Scheme processed.There is Experimental report that the optical transport of single channel speed 114Gb/s is realized using the system of 8-QAM modulation systems.

The present invention exactly in the case of this technical background, the requirement growing in order to adapt to optical communication system, this The bright coded modulation scheme based on 8-QAM modulation systems, furthers investigate to its modulation demodulation system and constellation mapping, and pin The characteristics of to its two ring of planisphere, more efficiently mapping mode is proposed, and then is proposed a kind of suitable for optical communication system and comprehensive Close the more superior two benches QC-LDPC coded modulation schemes of performance.

The content of the invention

The deficiency of the prior art for more than, it is an object of the invention to provide a kind of net coding gain, frequency efficiency, The aspect such as iterationses and complexity performance can preferably meet the code modulating method of the requirement of high speed optical communication system. Technical scheme is as follows：A kind of a kind of QC-LDPC coded modulation sides in optical communication system based on 8-QAM modulation systems Method, it is characterised in that comprise the following steps：

101st, the signal stream in optical communication system is divided into into information bit 1, information bit 2 and information bit 3, then will letter Breath bit 1 is encoded into first bit c for the encoder 1 of L1 through redundancy₀, and information bit 2 and information bit 3 are distinguished into Jing Cross encoder 2 and encoder 2 that redundancy is L2 and be encoded into the second bit c₁With the 3rd bit c₂；, wherein L1>L2；

102nd, the first bit c for obtaining will be encoded in step 101₀, the second bit c₁With the 3rd bit c₂Input 8 respectively is entered In quadrature amplitude modulator 8-QAM processed, respectively to the first bit c₀, the second bit c₁With the 3rd bit c₂Using UP-BP-BP stars Seat figure split-run is modulated, and is then transferred to decoder；

103rd, according to the first bit c after the modulation received in step 102₀, the second bit c₁With the 3rd bit c₂, it is first First by modulation after the first bit c₀It is soft demodulated, judges the first bit c₀Posterior probability_pIf p=0 judges the first ratio Special c₀For internal ring, the second bit c after modulation is calculated₁With the 3rd bit c after modulation₂Bit log-likelihood ratio LLR, Then it is delivered separately to the first bit c₀, the second bit c₁With the 3rd bit c₂Corresponding decoder Decode1, Decode2 and Decode3,

Wherein

Wherein r represents the signal that receives, and r coordinates are (x, y), and c represents transmitted bit the second bit c₁With the 3rd bit c₂If p=1 judges the first bit c₀For outer shroud, then repeat step 103 is decoded again, completes coded modulation.

Further, enter row decoding by the way of multistage decoding MSD peace row decoding PDL combines in step 103.

Further, in the bit log-likelihood ratio LLR Gaussian white noise channels in step 103, LLR table is shown as：

Wherein S₀And S₁The set of the point of the bit position bit 0 and 1 given in representing preferable constellation point respectively；s_xWith s_yRepresent two quadrature components in preferable coordinate points.δ²It is the noise in baseband signal.

Advantages of the present invention and have the beneficial effect that：

In terms of the computation complexity, the computation complexity of proposed scheme is the 2/3 of traditional scheme to the present invention.In error correction Energy aspect, the scheme for not only being proposed are compared with code check identical traditional scheme therewith, the net coding gain of proposed scheme (NCG) it is higher, and the error-correcting performance of proposed scheme is even better than the lower traditional scheme of code check.In sum, the present invention Whole scheme after the coded modulation scheme for being proposed is combined with QC-LDPC codes is equal at aspects such as spectrum efficiency, NCG and complexities The requirement of optical communication system can preferably be met.

Description of the drawings

Fig. 1 is traditional 8-QAM mapping modes

Fig. 2 is the collection segmentation of traditional 8-QAM

Fig. 3 is the parallel independent interpretation device of traditional 8-QAM schemes

Fig. 4 is the UP-BP-BP mixing partitioning schemes that the present invention proposes a plan

Fig. 5 is traditional 8-QAM mapping relations (left side) and mapping relations (right side) proposed by the invention

Fig. 6 is the encoder block diagram of scheme proposed by the invention

Fig. 7 is the decoder block diagram of scheme proposed by the invention

Fig. 8 is the amount of calculation contrast schematic diagram of traditional scheme and scheme proposed by the invention

Fig. 9 experiments one：The traditional scheme performance comparison of scheme proposed by the invention and close code check

Figure 10 present invention one each scheme performance comparison figures of experiment

Figure 11 experiments two：The traditional scheme performance comparison of scheme proposed by the invention and close code check

Figure 12 present invention two each scheme performance comparison figures of experiment

Figure 13 is scheme proposed by the invention and RS (255,239) performance comparison figure

Table 1 is QC-LDPC code pattern parameters in present invention experiment one

Table 2 is each scheme Performance comparision in present invention experiment one

Table 3 is QC-LDPC code pattern parameters in present invention experiment two

Table 4 is each scheme Performance comparision in present invention experiment two.

Specific embodiment

The invention will be further elaborated to provide the embodiment of an indefiniteness below in conjunction with the accompanying drawings.

As shown in figure 1, a kind of QC-LDPC code modulating methods in planting optical communication system based on 8-QAM modulation systems, its It is characterised by comprising the following steps：

101st, the signal stream in optical communication system is divided into into information bit 1, information bit 2 and information bit 3, then will letter Breath bit 1 is encoded into first bit c for the encoder 1 of L1 through redundancy₀, and information bit 2 and information bit 3 are distinguished into Jing Cross encoder 2 and encoder 2 that redundancy is L2 and be encoded into the second bit c₁With the 3rd bit c₂；, wherein L1>L2；

102nd, the first bit c for obtaining will be encoded in step 101₀, the second bit c₁With the 3rd bit c₂Input 8 respectively is entered In quadrature amplitude modulator 8-QAM processed, respectively to the first bit c₀, the second bit c₁With the 3rd bit c₂Using UP-BP-BP stars Seat figure split-run is modulated, and is then transferred to decoder；

103rd, according to the first bit c after the modulation received in step 102₀, the second bit c₁With the 3rd bit c₂, it is first First by modulation after the first bit c₀It is soft demodulated, judges the first bit c₀Posterior probability p, if p=0, judge first ratio Special c₀For internal ring, the second bit c after modulation is calculated₁With the 3rd bit c after modulation₂Bit log-likelihood ratio LLR, Then it is delivered separately to the first bit c₀, the second bit c₁With the 3rd bit c₂Corresponding decoder Decode1, Decode2 and Decode3,

Wherein

Wherein, represent the signal that receives, r coordinates are (x, y), c represents transmitted bit the second bit c₁With the 3rd bit c₂If p=1 judges the first bit c₀For outer shroud, then repeat step 103 is decoded again, completes coded modulation.

The present invention is a kind of two benches QC-LDPC coded modulation schemes in optical communication system based on 8-QAM modulation systems, Including：

1., in the method for various increase power system capacities, M-QAM is a kind of amplitude and phase place for jointly controlling sinusoidal carrier Modulation system.8-QAM therein becomes the popular modulation scheme of optic communication of future generation.In the possible signal constellation (in digital modulation) of many kinds In figure, a kind of generally acknowledged optimal 8-QAM signal constellation (in digital modulation)s figure and mapping mode are shown in Fig. 1, because it is for given signaling point Between minimum range required by power it is minimum.In Fig. 1 8-QAM signals be made up of the signaling point of two amplitudes and signaling point it Between minimum range be all 2A (coordinate of each signal is by A normalization).A symbol in 8-QAM is set to into c₀c₁c₂Number.8- The constellation point partition mechanism of QAM as shown in Fig. 2 as the division of constellation point each time does not all change minimum euclidean distance, because The BP collection partitioning schemes that this generally acknowledged optimal 8-QAM planisphere is adopted.In optical communications, when 8-QAM is applied to coded modulation During scheme, in general, using Bit Interleaved Coded Modulation (Bit-interweave Coding Modulation, BICM) side Case.In order to scheme comparison proposed by the invention, unification BICM schemes are referred to as into traditional scheme, or parallel independent interpretation Scheme.In traditional scheme, three bits are encoded using same pattern.In receiving terminal, using PDL decoding schemes, three Bit demodulation is parallel and independent with the process of decoding.As shown in Figure 3.Demod c₀Represent bit c₀Soft demodulating process, LLR(c₀) represent c₀Bit log-likelihood ratio (Log-likelihood Ratio, LLR).Bit c₁c₂In the same manner.In tradition side In case, after LLR is calculated respectively by three demodulators, corresponding decoder is passed to.The LLR of a certain bit c is defined as：

Wherein r represents the signal for receiving, and coordinate is (x, y).C be transmitted bit (in transmission symbol in K bit one It is individual), and assume that all bits are all equiprobability transmission.

In Gaussian white noise channel, LLR can be represented with equation below (2)：

Wherein S₀And S₁The set of the point of the bit position bit 0 and 1 given in representing preferable constellation point respectively；s_xWith s_yRepresent two quadrature components in preferable coordinate points.δ²It is the noise in baseband signal.

In the present invention, in order to make full use of 8-QAM annular planispheres the characteristics of, as shown in fig. 1,8-QAM planispheres It is formed by stacking by two QPSK constellation points for there are 45 ° of differences, and respectively on the different circle of two radiuses.Thus, the present invention is carried Go out in a kind of new mapping relations, these three bits are no longer the relations of equality, and 8-QAM planispheres are divided into two rings, by symbol In first bit decision its belong to internal ring or outer shroud, as shown in Figure 4.From the point of view of the theory of set-partition, divide for the first time Cut, the Euclidean distance of internal ring each point does not change, but the Euclidean distance between outer shroud is significantly increased, therefore can be by UP segmentations are regarded in one layer of segmentation as.By taking internal ring as an example, second segmentation and third time segmentation all do not make Euclidean distance become big, can be with Regard BP segmentations as.Therefore, the program is UP-BP-BP partitioning schemes.If mapping mode proposed by the invention and coding adjusted Scheme processed is combined, and coding and decoding mode all can change therewith.Generally speaking, mapping mode proposed by the invention and traditional approach Difference is exactly to take full advantage of the special constellation graph structures of 8-QAM, has distinguished inner ring and outer rings using first bit so that letter The minimum euclidean distance that number planisphere is shown in becomes big, as shown in Figure 5.

2., in scheme proposed by the present invention, three bits in a 8-QAM symbol are no longer the relations of equality, c₀'s Judged result directly affects c₁And c₂, c in other words₁And c₂To c₀Degree of dependence it is very big.In consideration of it, what if three bits were adopted Identical coded system, then overall performance will allow of no optimist.It is readily conceivable that due to c₀Importance, in encoding context plus The strong protection to which, the pattern for selecting performance more excellent, the pattern of such as High redundancy.In order to balance the redundancy of whole system, c₁And c₂In same grade, the pattern that another kind of redundancy can be adopted relatively low.Using not reciprocity protection encoder block diagram such as Shown in Fig. 6, dash area is c₀Encoder.

3. the concrete decoding flow process of the present invention program is as shown in Figure 7.The posteriority of first bit in symbol is calculated first Probability, judges which is 0 or 1, if 0, then calculates c according to the mapping relations of internal ring₁And c₂LLR.If 1, then for outer Ring.As seen from Figure 7, this kind of decoded mode is effective combination of two kinds of decoded modes in BCM.For c₀And c₁c₂Between sentence It is disconnected to belong to MSD i.e. multistage decoding, to c₁c₂Between judgement belong to the i.e. parallel decoded modes of PDL.Two ways is combined i.e. most Big the characteristics of make use of planisphere of degree, effective compromise of two kinds of decoded mode pluses and minuses of having got back.Generally speaking, tradition But scheme is in coding, although be all consistent, and scheme proposed by the invention using No. three encoders its code selection In coding, 3 bits are protected with different patterns.During decoding, traditional scheme by the way of the decoding of three bits parallels, And the decoding result of first bit in scheme proposed by the invention, is obtained first, judge the constellation point in internal ring or outer Ring, is further continued for decoding.This decoded mode, is effective combination of MSD decodings and PDL decodings in BCM.

4. the computation complexity analysis of scheme proposed by the invention

The complexity that calculate to compare two schemes of the present invention from LLR.In traditional scheme, the LLR of each bit Can be represented with equation below：

S₀And S₁The set of the point of the bit position bit 0 and 1 given in representing preferable constellation point respectively.Due to being 8 to enter System modulation, then have three bits in each symbol, each is than peculiar 0 and 1 two kind of possibility.Therefore in traditional scheme, 8-QAM stars There are 8 points in seat figure, (3) set S in formula₀Set S₁Four points are respectively taken, therefore when often calculating the LLR value of a bit, needs 8 It is secondaryComputing, the computing of such a combination in text is called a computing collection.In such a computing collection In, rough calculating needs 3 sub-additions, 3 multiplication, and 1 power operation.For a symbol of traditional algorithm, have three Individual bit, needs 24 computing collection altogether.In scheme proposed by the invention, as shown in Figure 8.For bit c₀Operation it is same Sample needs to carry out 8 computing collection.As known c₀Result after, two bits hereafter are then calculated according to the LLR value of QPSK, And for c₁And c₂, set S₀Set S₁Each there was only two points, that is to say, that S₀∪S₁No longer it is 8 points, but 4 points.Cause This each bit only needs to 4 computing collection.For in proportion, calculate complicated for the 2/3 of traditional scheme.Analysis above only limits one The amount of calculation of individual symbol.If one frame of analysis, so that code length is 5000 as an example.Propose a plan and will reduce 5000 × 8 computing collection, meaning Taste, and has lacked the computing of 40,000 power, 120,000 sub-addition computings, 120,000 multiplyings.The reduction of complexity still very may be used See.As shown in figure 8, the addition number of times needed for soft demodulation is intuitively given in figure.

5. scheme proposed by the invention combines the performance evaluation of finite field QC-LDPC codes

The LDPC code used in Error Control is to reach the most potential coding techniques of shannon limit at present.In general Structurized coded system can produce relatively low error floor, and this is most important for optic communication.Also, adopt cyclic code And quasi-cyclic code can become easy during coding by using shift register.Therefore will be proposed by the invention Scheme and two groups of QC-LDPC codes are combined, and can be referred to as QC-LDPC coded modulation schemes with reference to after.Two classes QC- are adopted in experiment LDPC code.Their performance is reduced with the increase of code check.In traditional parallel decoding scheme, three bits are using same A kind of pattern coding.Due to c in scheme proposed by the invention₁And c₂Value and c₀Judged result have much relations, so this Invention adopts not reciprocity protection (Unequal protection, UEP) coded system, and by c₀Bit on position strengthens coding Protection.In order to balance the code check of whole system, c₁And c₂Using the higher pattern of code check.In order to fully verify that the present invention is carried Go out the effectiveness of scheme.The present invention has carried out two experiments, in order to comparability and keep unified, three in same experiment Individual pattern adopts same make, and code length is consistent.For code 1, code check is R1, and the code check of code 2 is R2.Code 3 is R3.It is right In coded modulation scheme proposed by the invention, code 1 is used to protect c₀, code 3 be used for protect c₁And c₂.Relation between them It is, R₁<R₂<R₃.This relation is while 1 best performance of description code, secondly, worst is code 3 to code 2.And R₁+2R₃=3R₂, this Individual relational expression is it is meant that the coded modulation scheme code check generally for proposing is consistent with the traditional scheme using code 2.It is former according to this Reason, has carried out following two experiments.

Experiment one：

In experiment one, add the QC-LDPC codes constructed by group using based on finite field.The code length of 3 patterns is consistent, other Parameter is as shown in table 1.Scheme conjunctive use code 1 proposed by the invention and code 3, its code check are respectively 0.89 and 0.924 code Type, the code check after its synthesis are 0.912, with the code check 0.918 of code 2 closely.Therefore will analyze first and propose a plan and single The performance of the traditional scheme of code 2 is adopted solely.

In experiment one, the system frequency efficiency of scheme proposed by the invention is 2.74bit/symbol, and the bit error rate is 10^-7 When compare with uncoded 8-QAM systems, bring the NCG of 4.85dB.As can be known from Fig. 9, although in low signal-to-noise ratio region, sheet The performance of invention proposed scheme is weaker, but when EbN0 is more than 7.75dB, the performance of scheme proposed by the invention starts to be better than Traditional scheme, and bring obvious coding gain.The bit error rate is 10^-7When, scheme proposed by the invention is close compared with code check Traditional scheme brings the NCG of 0.2dB.Moreover, can be by scheme proposed by the invention and using other code check patterns Traditional scheme is compared, will three kinds of codes be all separately employed in traditional scheme.Specific performance performance is as shown in Figure 10.In experiment In one, the code check of three kinds of codes is sufficiently close to, and code 1 and 3 code check of code difference only have 3.68%.Therefore, in simulation result, three kinds of codes exist Less, performance curve is even overlapped the performance difference of traditional scheme.But still as can be seen that propose a plan in high s/n ratio region It is always maintained at clear superiority.10 are given in table 2^-6The performance parameter of each scheme.It is 10 in the bit error rate^-6When, propose a plan The NCG of 0.12dB or so is brought to other three kinds of schemes.It is noted that the performance of scheme proposed by the invention is better than Code check lower traditional scheme.

Experiment two：

In experiment two, using the QC-LDPC codes that constructed based on finite field subgroup, and according to the subgroup in GF (73) domain Construct one group of QC-LDPC code.Equally, analysis has the performance of two schemes of close code check first, i.e., using code 1 and code 3 Propose a plan (code check after comprehensively is 0.881), and using the traditional scheme of code 2 (0.89).As shown in figure 11.

The system frequency efficiency of scheme proposed by the invention is 2.697bit/ symbols, is 10 in the bit error rate^-7When and it is uncoded 8-QAM systems compare, bring 5.45dBNCG.In low signal-to-noise ratio region, the performance of scheme proposed by the invention slightly has not Foot, but when EbN0 is more than 6.8dB, the performance of scheme proposed by the invention starts torsion occur, and shows thereafter substantially Advantage.It is 10 in the bit error rate^-7When, bringing the net of 0.45dB the scheme proposed by the invention traditional scheme close compared with code check more Coding gain.Additionally, these three patterns are all respectively used to tradition side by the performance in order to fully demonstrate scheme proposed by the invention In case.As can be known from Fig. 12, two benches coded modulation scheme proposed by the invention brings obvious coding gain.In table 4 The bit error rate is given 10^-6When, the code check of each scheme and its NCG situations brought.It is again seen that one from Figure 12 and in table 4 Experiment two and one identical phenomenon of experiment, that is, the scheme that scheme performance proposed by the invention is not only close to better than code check, Some situations are even better than the lower traditional scheme of code check.Additionally, the code check difference of code 1 and code 3 is slightly larger in experiment two, it is 5.89%.Therefore, more net coding gain is brought in experiment two.

6. combination property analysis of scheme proposed by the invention

Generally speaking, above-mentioned two experiment QC-LDPC codes used, code length are different, and make is different, and performance is not Together, combined with coded modulation scheme proposed by the invention, can be than the traditional scheme with close code check in bit error rate Performance is excellent, even better than the lower traditional scheme of code check.The broad applicability for proposing a plan can be illustrated.And it was found that, when Bigger from the code check difference of code word, scheme proposed by the invention can bring coding gain more.

Next discuss whether scheme proposed by the invention meets the requirement of high speed optical communication system.Look first at the present invention The transfer rate that proposed scheme can reach.It is assumed that Symbol processing speed is R_s, according to palarization multiplexing (Polarization Division Multiplexing, PDM) the single channel speed that can reach of system be 2 × R_s× Rate × logM, wherein Rate is code check, and logM is the bit number in each symbol.The character rate R that current level of hardware can be realized_sCan reach 40Giga-Symbol/s(GS/s).If using coded modulation scheme proposed by the invention, code check Rate for 0.918, logM is 3, single channel speed can reach the speed of 220G/S.Additionally, the error floor of 3 LDPC codes in experiment one is verified, it is low In 10^-18.Therefore performance curve in experiment one can be obtained the estimation of the performance under high s/n ratio according to data of literatures Value.Then the LDPC code modulation systems and BPSK modulation in experiment one is compared using the scheme of RS (255,239).As schemed Shown in 13.Can see, scheme proposed by the invention is 10^-10Nearby performance is begun to better than RS codes, 10^-15Bring The NCG of 0.65dB.It is emphasized that the spectrum efficiency of scheme proposed by the invention is 2.74bit/symbol, and BPSK is adjusted The RS code schemes spectrum efficiency of system is 0.937bit/symbol.That is, in the case where frequency efficiency is three times higher, remain to band Carry out coding gain.In addition, by simulating, verifying, the adopted QC-LDPC codes iterationses of present invention experiment are 5 times, with other The pattern that universal iteration is more than ten times is compared, and saves more time and amount of calculation.

In sum, coded modulation scheme proposed by the invention net coding gain, frequency efficiency, iterationses and The aspect such as complexity performance can preferably meet the requirement of high speed optical communication system.

	Code length N	Information bit K	Code check R (dB)
				Code 1	5256	4679	0.890
Code 2	5256	4823	0.918
				Code 3	5256	4859	0.924

QC-LDPC code pattern parameters in 1 present invention experiment one of table

Each scheme Performance comparision in 2 present invention experiment one of table

	Code length N	Information bit K	Code check R
				Code 1	2592	2236	0.863
Code 2	2592	2307	0.890
				Code 3	2592	2378	0.917

QC-LDPC code pattern parameters in 3 present invention experiment two of table

Coded modulation scheme	Code check	In 10-6 net coding gains
			Parallel decoding+code 1	86.3%	4.65
Parallel decoding+code 2	89.0%	4.55
			Parallel decoding+code 3	91.7%	4.25
Propose a plan	88.1%	5.10

Each scheme Performance comparision in 4 present invention experiment two of table

The above embodiment is interpreted as being merely to illustrate the present invention rather than limits the scope of the invention. After the content of the record for having read the present invention, technical staff can be made various changes or modifications to the present invention, these equivalent changes Change and modification equally falls into the inventive method claim limited range.

Claims (3)

1. a kind of a kind of QC-LDPC code modulating methods in optical communication system based on 8-QAM modulation systems, it is characterised in that bag Include following steps：

101st, the signal stream in optical communication system is divided into into information bit 1, information bit 2 and information bit 3, then by information ratio Special 1 is encoded into first bit c for the encoder 1 of L1 through redundancy₀, and by information bit 2 and information bit 3 respectively through superfluous Remaining is encoded into the second bit c for the encoder 2 and encoder 3 of L2₁With the 3rd bit c₂, wherein L1>L2；

102nd, the first bit c for obtaining will be encoded in step 101₀, the second bit c₁With the 3rd bit c₂8 systems are input into respectively just In handing over amplitude modulator 8-QAM, respectively to the first bit c₀, the second bit c₁With the 3rd bit c₂Using UP-BP-BP planispheres Split-run is modulated, and the UP-BP-BP planispheres split-run is：8-QAM planispheres are divided into two rings, by symbol in One bit decision its belong to internal ring or outer shroud, from the point of view of the theory of set-partition, split for the first time, internal ring each point it is European Distance does not change, but the Euclidean distance between outer shroud is significantly increased, therefore can regard ground floor segmentation as UP point Cut；Second segmentation and third time segmentation of internal ring does not all make Euclidean distance become big, regards BP segmentations as, is then transferred to decoding Device；

103rd, according to the first bit c after the modulation received in step 102₀, the second bit c₁With the 3rd bit c₂, first will The first bit c after modulation₀It is soft demodulated, judges the first bit c₀Posterior probability p, if p=0, judge the first bit c₀ For internal ring, the second bit c after modulation is calculated₁With the 3rd bit c after modulation₂Bit log-likelihood ratio LLR, Ran Houfen Supplementary biography passs the first bit c₀, the second bit c₁With the 3rd bit c₂Corresponding decoder Decode1, Decode2 and Decode3,

Wherein

Wherein r represents the signal that receives, and r coordinates are (x, y), and c represents transmitted bit the second bit c₁With the 3rd bit c₂If, p =1, then judge the first bit c₀For outer shroud, then repeat step 103 is decoded again, completes coded modulation.

2. a kind of QC-LDPC coded modulation sides in optical communication system according to claim 1 based on 8-QAM modulation systems Method, it is characterised in that：Enter row decoding by the way of multistage decoding MSD peace row decoding PDL combines in step 103.

3. a kind of QC-LDPC coded modulation sides in optical communication system according to claim 1 based on 8-QAM modulation systems Method, it is characterised in that：In bit log-likelihood ratio LLR Gaussian white noise channels in step 103, LLR table is shown as：

$L\left(c\right)=log\left(\frac{\underset{s\&Element;S_0}{\mathrm{\&Sigma;}}{e}^{-\frac{1}{{\mathrm{\&delta;}}^2}({(x-s_x)}^2+{(y-s_y)}^2)}}{\underset{s\&Element;S_T}{\&Sigma;}{e}^{-\frac{1}{{\mathrm{\&delta;}}^2}({(x-s_x)}^2+{(y-s_y)}^2)}}\right)$

Wherein S₀And S₁The set of the point of the bit position bit 0 and 1 given in representing preferable constellation point respectively；s_xAnd s_yRepresent Two quadrature components in preferable coordinate points, δ²It is the noise in baseband signal.