CN107682122B - A kind of iterative demodulation and decoding method of wireless optical communication multi-level coding modulation system - Google Patents

Abstract

The invention discloses an iterative demodulation and decoding method for multi-level coding modulation of wireless optical communication.

and decoded output

Steps such as iterative update and iterative stop are performed. In a multi-level coded modulation wireless optical communication system, when the component codes are improperly configured to cause error propagation, this method has better error performance than the multi-stage demodulation and decoding method. When there is no error propagation in the component code configuration, this method The error performance of the method is comparable to that of the multi-stage demodulation decoding method.

Description

A kind of iterative demodulation and decoding method of wireless optical communication multi-level coding modulation system

technical field

The invention relates to a demodulation and decoding method of a wireless optical communication multi-level coding modulation system, in particular to an iterative demodulation and decoding method and realization of the multi-level coding modulation.

Background technique

The channel conditions of wireless communication systems are generally poor, and the use of channel error correction coding is one of the common methods to improve communication reliability. Generally speaking, a good coded modulation design is to construct long codewords and implement random coding, and to approximate or achieve maximum likelihood demodulation and decoding at the receiving end. However, the maximum likelihood decoding of long codewords is often very complicated and difficult to achieve. Multi-stage coding and modulation technology can construct multiple short code words into one long code word, and realize random coding through simple interleaving in modulation. The performance of large-likelihood demodulation and decoding is greatly reduced, and the complexity is greatly reduced.

The schematic diagram of the multi-level coding modulation structure is shown in Figure 1. It consists of three parts: the serial-to-parallel conversion of the input source, multi-level coding and signal modulation. equal. For an M-order modulation system, the source data is first transformed into M-channel parallel data, which are respectively sent to M-level encoders for encoding. The codeword lengths of the component codes output by each encoder are the same. One bit is taken and combined to form a modulation packet, and the modulation packet is modulated in the modulator into a transmission signal for transmission.

Suppose the i-th (1≤i≤M) encoder is an encoder C _i with information bit length k _i and encoding length n , the output codewords of C ₁ , C ₂ , ..., C _M are respectively expressed as v ₁ =(a _1,1 ,a _1,1 ,....,a _1,n ), v ₂ =(a _2,1 ,a _2,2 ,....,a _2,n ), ..., v _M =(a _M,1 ,a _M,2 ,....,a _M,n ), then the above codeword can be regarded as a sequence with n symbols after modulation, where f( ) represents the mapping relationship between the combined sequence and the transmitted signal, and "*" represents the simple interleaving relationship of the signal formed by the modulated packet.

(称之为接收估计)，之后进入第一级的译码环节，将

送入对应译码器D₁中进行译码，得到第一级分量码的最终值

(称之为译码估计)；然后以R和

作为已知条件，对R进行解调得到第二级分量码v₂的接收估计

并送入译码器D₂得到第二级分量码的译码估计

再以R、

和

作为已知条件，对R解调获得第三级的接收估计

送入译码器D₃得到其译码估计

依次类推，直到得出最后一级的

最后，根据各级分量码的编译码法则，将信息比特从

中分离出来，并按发送的逆序作并串转换，得到最终的信息序列。这种方法的优势是：相比极大似然方法解调译码复杂度大大降低，同时还可以获得较优误码性能。At the receiving end, after the transmitted signal f(·) is received and detected, the original transmitted information bits must be recovered, which is called demodulation and decoding of multi-level coding and modulation. A typical demodulation and decoding method is multi-stage demodulation and decoding. The method starts from the first-level component code, one component code at a time, including receiving estimation (ie demodulation) and decoding two links, the previous stage. The obtained decoding information will be passed to the lower stage for the reception estimation of the lower stage, ending at the last stage. The schematic block diagram of its principle is shown in Figure 2. The specific method is: set the photoelectric receiving signal as a vector R, and take R as a known condition, first send it into the first-stage receiving estimation (demodulation) link, and demodulate the first stage. Estimation of the _first -order component code v1

(called reception estimation), and then enter the first-level decoding link, which will

It is sent to the corresponding decoder D ₁ for decoding to obtain the final value of the first-level component code

(called the decoding estimate); then with R and

As a known condition, demodulate R to obtain the received estimate of the _second -order component code v2

And send it to the decoder D2 to obtain the decoding estimation of the _second -level component code

Then use R,

and

As a known condition, demodulate R to obtain a third-level receive estimate

Send to decoder _D3 to get its decoding estimate

And so on, until the last level of

Finally, according to the coding and decoding rules of component codes at all levels, the information bits are converted from

It is separated from it, and the parallel-to-serial conversion is performed according to the reverse order of transmission to obtain the final information sequence. The advantage of this method is that compared with the maximum likelihood method, the demodulation and decoding complexity is greatly reduced, and at the same time, better bit error performance can be obtained.

如

在大概率条件下与v₁,v₂,...,v_i-1相等，则可以对后级的解调估计形成一定保护作用，从而提高

的准确性，获得较好的误码性能。但是，如估值

与实际发送的v₁,v₂,...,v_i-1存在较大差异，也易导致

不准确，造成错误率增加。由于多阶段解调译码是从第一级分量码开始，逐次进行，因此，当前级的估计出现较多错误时，会对后续所有分量码的译码产生影响并形成错误累积，这被称为多阶段解调译码的错误传播现象。错误传播会导致系统通信的误码性能严重下降，目前的做法是分析各级分量码所对应子信道容量，在此基础上分别对应配置不同纠错能力的信道编码。这种方式需要准确估计各子信道的信道状态信息，而在实际通信过程中这显然往往难以做到。It can be seen that in the multi-stage demodulation and decoding method, the estimation of the _i -th level component codeword vi needs to rely on the previous estimation

like

If it is equal to v ₁ , v ₂ ,...,v _i-1 under high probability conditions, it can form a certain protective effect on the demodulation estimation of the later stage, thereby improving the

accuracy and better bit error performance. However, if the valuation

There is a big difference with the actual sent v ₁ , v ₂ ,...,v _i-1 , and it is also easy to cause

Inaccurate, resulting in increased error rate. Since the multi-stage demodulation and decoding starts from the first-stage component code and proceeds successively, when there are many errors in the estimation of the current stage, it will affect the decoding of all subsequent component codes and form error accumulation, which is called Error propagation phenomenon for multi-stage demodulation decoding. Error propagation will seriously degrade the bit error performance of system communication. The current practice is to analyze the sub-channel capacity corresponding to the component codes at all levels, and on this basis, configure channel codes with different error correction capabilities. This method requires accurate estimation of the channel state information of each sub-channel, which is obviously often difficult to achieve in the actual communication process.

SUMMARY OF THE INVENTION

The purpose of the invention is to invent a new demodulation and decoding method based on the basic method of multi-stage demodulation and decoding, using the posterior probability Bayesian expansion principle and iterative technology to suppress the error propagation phenomenon in the multi-stage demodulation and decoding. , thereby improving error performance.

In order to achieve the above object, the present invention proposes a demodulation and decoding method based on iterative multi-level coding and modulation. The method starts from the basic idea of reducing the error of the pre-stage code, uses the diversity of the Bayesian expansion of the posterior probability and the characteristic that the pre-stage forms a protective effect on the post-stage in the multi-stage demodulation and decoding. The estimation process is designed as a full feedback structure of the decoding result, and through iterations, the receiving estimation at each level and the output of the decoding link are updated in parallel, so that the unidirectional protection effect of the previous stage on the latter stage in the original multi-stage demodulation and decoding is transformed. It is the mutual protection between sub-channel codes at all levels, thereby improving the accuracy of estimation. The principle of the single iterative demodulation and decoding structure of this method is shown in Figure 3.

The present invention is specifically achieved through the following technical measures:

被送入到除本级以外的其它各级接收估计环节中，形成译码结果到接收估计的全反馈结构，使得每级接收估计环节的输出

均由其它各级译码输出和接收信号共同进行估计。①Construct a full feedback structure from decoding to receiving estimation. The M-level multi-level coding and modulation system corresponds to M-level demodulation and decoding sub-channels. The sub-channels at each level include two links: reception estimation and decoding. The decoder output of the sub-channels at each level

It is sent to other receiving estimation links at all levels except this stage, and forms a full feedback structure from the decoding result to the receiving estimation, so that the output of each stage receives the estimation link.

All are estimated jointly by other levels of decoding output and received signal.

的初始值可由多阶段解调译码方法直接得到。②Get the initial value of the iteration. For the M-order multi-level coded modulation system, the decoding output of each level is

The initial value of can be obtained directly by the multi-stage demodulation and decoding method.

和译码输出

进行更新。设M阶多级编码调制的各级分量编码器输出码字为v₁＝(a_1,1,a_1,2,....,a_1,n)，v₂＝(a_2,1,a_2,2,....,a_2,n)，…，v_M＝(a_M,1,a_M,2,....,a_M,n)，接收信号表示为矢量R，迭代开始，对其中第i级

的估计值进行更新时，以其它所有

和R为已知量，分别计算

中第j(j＝1,.....n)个比特a_ij的似然度度量值：(3) The iteration starts, and the receivers at all levels are estimated through the full feedback structure.

and decoded output

to update. Assume that the output codewords of the component encoders of all levels of M-order multi-level coding and modulation are v ₁ =(a _1,1 ,a _1,2 ,....,a _1,n ), v ₂ =(a _2,1 ,a _2,2 ,....,a _2,n ),...,v _M =(a _M,1 ,a _M,2 ,....,a _M,n ), the received signal is represented as a vector R , the iteration starts, for which the ith level

is updated with all other

and R are known quantities, calculated separately

The likelihood metric of the jth (j=1,.....n) bit a _ij in:

更新后的置信度度量，对(Λ(a_i1),Λ(a_i2),...,Λ(a_in))进行二进制判决，即得到更新的接收估计

对置信度度量进行极大似然译码或对更新的

进行代数译码，得到更新的译码估计

同法并行对所有

更新，完成本次迭代。The obtained (Λ(a _i1 ),Λ(a _i2 ),...,Λ(a _in )) is

After updating the confidence measure, make a binary decision on (Λ(a _i1 ),Λ(a _i2 ),...,Λ(a _in )), that is, get the updated receiving estimate

Maximum likelihood decoding of confidence measures or updated

perform algebraic decoding to obtain updated decoding estimates

Parallel to all

Update to complete this iteration.

的更新，根据各级分量码码字的具体组成规则分离出信息比特，并将得到的各级信息比特按发送的逆序进行并串转换，从而恢复出最终信息序列，需要特别说明的是，迭代停止判据可以是多种形式和方法，如最常用的最大迭代次数等，但这不影响本发明的技术独创性主张。④ When the iteration stop criterion reaches the set value, stop

According to the specific composition rules of component codes at all levels, the information bits are separated, and the obtained information bits at all levels are converted to parallel-serial in the reverse order of transmission, so as to restore the final information sequence. It should be noted that iterative The stopping criterion can be in various forms and methods, such as the most commonly used maximum number of iterations, etc., but this does not affect the technical originality claim of the present invention.

Compared with the prior art, the present invention has the following beneficial effects:

① When the multi-stage coded modulation system adopts the multi-stage demodulation and decoding method and the error propagation phenomenon occurs, the present invention can obtain better error performance. For the multi-level coding PPM modulation system of wireless optical communication, under the condition of weak turbulent atmospheric channel with scintillation index of 0.1 and bit error rate of 10 ^-4 , three iterations can obtain a gain of more than 0.5dB compared with no iteration (Fig. 4), and The outputs of all levels in the iteration are updated synchronously, which saves the iteration time overhead.

(2) When the component codes of each level of multi-level coding and modulation are properly configured, that is, when multi-stage demodulation and decoding is used, there is almost no error propagation phenomenon, and the error performance of three iterations using the present invention is the same as that when no iteration is performed. At this time, the present invention does not Change system error performance. (Figure 5).

Description of drawings

FIG. 1 is a schematic diagram of the structure of multi-level coding and modulation.

FIG. 2 is a schematic block diagram of multi-stage demodulation and decoding.

Figure 3 is a block diagram of the demodulation and decoding principle of a single iteration.

Fig. 4 shows the third-order PPM multi-level coding and modulation system of wireless optical communication in the first embodiment (the three-level component codes are in the order of BCH(127,113) code, BCH(127,99) code and BCH(127,85) code configuration, there is error propagation) in the weak turbulent atmospheric channel, the error performance comparison curve obtained by the multi-stage demodulation and decoding method and the three iterations of this method.

Fig. 5 shows the third-order PPM multi-level coding and modulation system of wireless optical communication in the second embodiment (the three-level component codes are in the order of BCH(127,85) code, BCH(127,99) code and BCH(127,113) code configuration, no error propagation) using the multi-stage demodulation and decoding method and the error performance comparison curve obtained by three iterations of this method.

Detailed ways

Example 1

In this embodiment, the wireless optical communication system adopts the third-order PPM modulation, corresponding to the need for three-stage channel coding. 85) code, the PPM modulation mapping method is shown in Table 1, where (a ₁ , a ₂ , a ₃ ) represents a modulation group, and m _i (i=0, 1,..., 7) represents that the modulated optical pulse is The position on the time slot of the PPM symbol, the PPM modulation represents the information by the difference in the position of the time slot in the symbol where the optical pulse is located.

Table 1 PPM mapping method

First, three groups of source sequences are generated at the transmitting end. After these three groups of source data are respectively sent to the three-level channel encoder for encoding, they are modulated into optical PPM signals for transmission. After channel transmission, the receiving end will obtain a signal composed of 127 PPM signals. The received vector R of , where the received signal of the t-th PPM sending packet (a _1,t ,a _2,t ,a _3,t ) is r=(r ₀ ,r ₁ ,...,r ₇ ), r _i (i=0,...,7) represents the photoelectric conversion current value of the PPM receiving time slot, and the corresponding position information is denoted as m ₀ , m ₁ ,..., m ₇ .

迭代开始，更新

假设在有光脉冲时隙上接收得到转换的电流概率密度为f₁，

的初始估计值为“00”。根据式(1)和表1，可得：According to this method, the initial estimated value is first obtained by the multi-stage demodulation and decoding method

Iteration starts, update

Assuming that the converted current probability density is f ₁ when received on the optical pulse time slot,

The initial estimate of is "00". According to formula (1) and Table 1, we can get:

r₁＜r₀时

为其他值情况同理。最后通过译码器更新得到

Considering that the current probability density function f ₁ is an increasing function, the magnitudes of r ₀ and r ₁ can be directly compared, when r ₁ >r ₀

When r ₁ < r ₀

The same is true for other values. Finally, it is obtained by updating the decoder

The communication channel in this embodiment is set as a cascade of weak turbulent atmosphere and Gaussian channel, the scintillation index of turbulence is 0.1, the Gaussian channel noise average is zero, the variance is 2×10 ^-25 , the channel attenuation factor is 1, and the working wavelength of the light wave is 1.55 um, the detector quantum efficiency is 0.5, the multiplication gain is 100, and the source rate is 10 MBit/s. When the average power of the transmitter is -39.3dBm, the specific implementation processing results of this method are as follows.

Generate three groups of source data S1, S2, S3 to be transmitted:

After the three sets of data are encoded by the sub-channel encoders, they are modulated into PPM signals containing eight time slots. After being transmitted through the atmospheric turbulence channel, the received data is obtained at the receiving end as shown in Table 2 (Note: Since the noise introduced by the channel is stochastic process, so the results obtained from each experiment are different). The serial number represents the serial number of the received PPM signal, r ₀ to r ₇ are the received photocurrent values detected in each time slot of the PPM signal, and the unit is 1e ^-11 A.

Table 2 Received data at the receiving end

Using multi-stage demodulation and decoding on the received data, the initial estimated value is:

S1(0):

S2(0):

S3(0):

The result after the first iteration is:

The result after the second iteration is:

The result after the third iteration is:

Comparing the source data, we can see that the number of errors after different iterations is shown in Table 3. It can be seen that the number of bit errors is 3 when no iteration is performed, and the number of bit errors becomes 0 after the first iteration.

Table 3 The number of errors under different iterations

Repeat the above method, calculate the bit error rate through a large amount of data under different transmit average powers, and get the system bit error rate performance curve as shown in Figure 4. Among them, the abscissa is the average power, the ordinate is the bit error rate, ITE is the number of iterations, ITE=0 is the case when it is not iterated, and the multi-stage demodulation decoding method is used when it is not iterated. As can be seen from the figure, the bit error performance curve of the system has been improved after iteration. Among them, the performance of one iteration (ITE=1) compared with no iteration (multi-stage demodulation and decoding method, ITE=0) has obtained a relatively obvious performance improvement. Compared with the bit error rate of 10 ^-4 , one iteration obtained With a gain of about 0.5dB, the bit error performance curves of 2 and 3 iterations are slightly better than that of 1 iteration, but the two curves almost overlap. No performance improvement can be obtained. At the same time, since the component code outputs of all levels can be updated in parallel in each iteration, the time overhead of the iteration is also small.

Embodiment 2

In this embodiment, the wireless optical communication system still adopts third-order PPM modulation, and the corresponding third-order component codes are configured in the order of BCH (127, 85) code, BCH (127, 99) code and BCH (127, 113) code, In this configuration, multi-stage demodulation decoding has minimal error propagation. At this time, the bit error performance curves of iterative (when ITE=1, 2, and 3) and no iteration (when ITE=0, a multi-stage demodulation and decoding method is used) are shown in FIG. 5 . It can be seen from the figure that the four curves almost overlap, that is, the present invention cannot obtain gain improvement at this time, but also does not cause system performance degradation.

Claims (1)

1. a kind of iterative demodulation and decoding method of wireless optical communication multistage coding and modulation system, it is characterized in that: set multistage coding and modulation system to be M order, and the code word that each level component encoder outputs is respectively v ₁ =(a _1,1 ,a _1,2 ,....,a _1,n ), v ₂ =(a _2,1 ,a _2,2 ,....,a _2,n ),...,v _M = (a _M,1 ,a _M,2 ,....,a _M,n ), the received signal is represented as a vector R, consisting of the following steps: 1) Construct a full feedback structure from decoding to receiving estimation The M-level multi-level coding and modulation system corresponds to M-level demodulation and decoding sub-channels. The sub-channels at each level include two links: reception estimation and decoding. The decoder output of the sub-channels at each level

It is sent to the receiving and estimation links at all levels except the current stage to form a full feedback structure from the decoding result to the receiving estimation, and the output of each stage receives the estimation link.

It is estimated jointly by other levels of decoding output and received signal; 2) Obtain the initial estimated value of each level of decoding output by the multi-stage demodulation and decoding method

3) At the beginning of the iteration, using the full feedback structure decoded to the received estimation, the i-th (i=1,2,...,M)-level received estimation

and decoding output estimates

When updating, with all other

and R are known quantities, based on the formula

calculated separately

The likelihood metric of the jth (j=1,.....n) bit a _ij in the obtained (Λ(a _i1 ),Λ(a _i2 ),...,Λ(a _in ) ) is the updated

Confidence measure, make binary decision on (Λ(a _i1 ),Λ(a _i2 ),...,Λ(a _in )), that is, get the updated reception estimate

Maximum likelihood decoding of confidence measures or updated

perform algebraic decoding to obtain updated decoding estimates

Parallel to all

Update to complete this iteration; 4) After the iteration stop criterion reaches the set value, stop

The information bits are separated according to the specific composition rules of the codewords of the component codes at all levels, and the obtained information bits at all levels are subjected to parallel-serial conversion in the reverse order of transmission to restore the final information sequence.

Images