CN116015554A - A Fusion Method for Soft Information Extraction of Heterogeneous Signals Based on HF Multi-Channel Diversity Framework - Google Patents

Abstract

The invention belongs to the technical field of signal transmission, and in particular relates to a fusion method for extracting soft information of heterogeneous signals based on a short-wave multi-channel diversity framework. demodulate signals of different modulations separately, and extract the corresponding symbol soft information; uniformly map the symbol soft information into bit soft information; use the noise model of bit soft information to realize the normalized description of bit soft information; Diversity gain is obtained in the form of addition, and decoding is performed through multi-judgment. The present invention proposes a method for extracting symbol soft information for heterogeneous signals, which extracts symbol soft information for signals with different modulation modes, and analyzes different The difference of the symbol soft information of the modulation mode is mapped to the bit soft information by using the corresponding mapping method, so as to perform the fusion of the bit soft information.

Description

A fusion method for soft information extraction of heterogeneous signals based on shortwave multi-channel diversity framework

Technical Field

The invention belongs to the technical field of signal transmission, and in particular relates to a fusion method for extracting soft information of heterogeneous signals based on a shortwave multi-channel diversity framework.

Background Art

As a communication method beyond line of sight based on the principle of reflection of high-frequency electromagnetic waves by the ionosphere, shortwave communication has the advantages of flexibility and independence from fixed base station equipment, making it an irreplaceable communication method in large-scale regional emergency and military fields. Traditional point-to-point communication methods are difficult to compensate for the impact of deep fading of shortwave signals on communication. Therefore, diversity reception technology is usually applied to shortwave communication to form a shortwave wide-area diversity reception model, so that the receiving end obtains several independent branch signals, thereby obtaining diversity gain to resist fading and improve the anti-interference performance and communication reliability of shortwave communication in harsh environments.

At present, with the rapid development of shortwave adaptive modulation technology, each distributed station can use the optimal modulation method according to its own link conditions. In this way, the situation of heterogeneous links will make the waveform merging method ineffective. Traditional diversity fusion methods are all oriented to the same modulation waveform, so that the waveforms with the same frequency and phase are added at the receiving end to enhance the signal quality. When different stations use different modulation methods to send signals, such as amplitude domain modulation, frequency domain modulation, phase domain modulation, and composite modulation, if the waveform addition method is used, the band errors of different symbols will be added together, resulting in decoding errors.

Summary of the invention

In order to solve the problems existing in the above-mentioned prior art, the present invention provides a fusion method for extracting soft information of heterogeneous signals based on a shortwave multi-channel diversity framework. Instead of using the form of waveform addition, the soft information of each signal after demodulation and before decoding is combined to achieve the fusion of heterogeneous soft information of the shortwave wide-area diversity model, obtain diversity gain, and improve the reliability of shortwave communication.

The present invention provides a fusion method for extracting soft information of heterogeneous signals based on a shortwave multi-channel diversity framework, which mainly includes the following steps:

Step S1, each sending station modulates and sends the signal source using a suitable modulation method and transmission rate;

Step S2, demodulating the differently modulated signals separately and extracting the corresponding symbol soft information;

Step S3, uniformly mapping symbol soft information into bit soft information;

Step S4, realizing normalized mapping of bit soft information;

Step S5, obtaining a diversity gain by adding, and obtaining a decoding result by performing a majority decision.

Furthermore, the extraction of corresponding symbol soft information in step S2 includes:

The composite domain modulation QAM soft information extraction includes: the multi-domain joint modulation method is QAM modulation, and its information is mapped on two signal domains of amplitude and phase, and the distance between the received symbol constellation point and the reference constellation point is used to describe the soft information.

Furthermore, the step S3 includes:

PSK and QAM first perform correlation mapping and then bit mapping; ASK and MFSK perform bit mapping directly.

Further, the correlation mapping includes:

Swap the target result set, assuming the symbol set is:

P＝{P _i |i＝0,1,2,...,M-1}

The soft information corresponding to _Pi can be described as _di , where M represents the amount of symbol soft information. In QAM, _di represents the distance, and in MPSK, _di represents the absolute value of the angle difference. Here, the value of _di is negatively correlated. By negating the value, j represents the reference constellation point, and _Sj is the soft information of this reference constellation point, and a positively correlated description of the soft information _Sj can be obtained:

Further, the bit mapping includes:

The coefficient matrix is formed by mapping the bit information 0 to -1, and then the symbol soft information vector is multiplied by the coefficient matrix to map the symbol soft information to the bit soft information.

Furthermore, the bit soft information normalization mapping in step S4 includes:

By performing statistical analysis on the bit soft information and its noise, a statistical model is obtained to model the bit soft information at a certain moment, and the bit soft information is uniformly mapped into a probabilistic form of the bit soft information, limiting the value of the soft information to the range of [0,1].

Further, the uniform mapping of the bit soft information into a probability form of the bit soft information includes:

Assuming that the noise of the bit soft information obeys Gaussian distribution, the soft information after bit mapping is defined as the occurrence of event B, _A0 represents the event that the bit is 0, _A1 represents the event that the bit is 1, and _A0 and _A1 are two incompatible events.

The probabilities P(B|A ₀ ) and P(B|A ₁ ) are described as:

Wherein S is the soft information after bit mapping, μ ₀ and σ ₀ are normal distribution parameters of the metric value with bit 0, and μ ₁ and σ ₁ are normal distribution parameters of the metric value with bit 1.

According to the posterior probability formula:

Where P(A ₀ ) represents the probability that the current bit value in the received information is 0, and P(A ₁ ) represents the probability that the current bit value in the received information is 1. In order to ensure the maximum information entropy, the value sent needs to satisfy P(A ₀ )=P(A ₁ ), and finally the conversion from the metric value of the bit soft information to the probability value is obtained:

P(A ₀ |B) and P(A ₁ |B) are the final normalized bit soft information, denoted as m(0) and m(1).

Further, the step S5 comprises:

The normalized bit soft information is added and fused, and the larger value of M(0) and M(1) is taken as the decoding result:

M(0)＝m1(0)+m2(0)+,...,ms(0)

M(1)＝m1(1)+m2(1)+,...,ms(1)

mi(0) is the normalized bit soft information value of a bit of the i-th branch being 0, and M(0) is the fused and added value of a corresponding bit of all branches being 0. Similarly, mi(1) is the normalized bit soft information value of a bit of the i-th branch being 1, and M(1) is the fused and added value of a corresponding bit of all branches being 1.

The present invention proposes a method for extracting symbol soft information of heterogeneous signals, and respectively extracts symbol soft information of signals modulated in amplitude domain, frequency domain, phase domain, and composite domain; analyzes the difference of symbol soft information of different modulation modes, adopts corresponding mapping methods to map it to bit soft information, thereby fusing the bit soft information, and realizes the fusion of heterogeneous soft information of shortwave wide area diversity model by performing the above two points, obtains diversity gain, and improves the reliability of shortwave communication.

BRIEF DESCRIPTION OF THE DRAWINGS

Fig. 1 is a flow chart of the algorithm of the present invention;

2 is a schematic diagram of a method for extracting soft information from an amplitude domain modulated waveform 2ASK according to the present invention;

FIG3 is a schematic diagram of a method for extracting soft information from a frequency domain modulated waveform MFSK (M=4) according to the present invention;

FIG4 is a schematic diagram of a method for extracting soft information from a phase domain modulated waveform MPSK (M=8) according to the present invention;

5 is a schematic diagram of a method for extracting soft information from a composite domain modulation waveform 16QAM according to the present invention;

FIG6 is a schematic diagram of a 16QAM bit mapping composite domain modulation waveform according to the present invention;

FIG. 7 is a schematic diagram of normalized mapping of soft information of 16QAM bit of composite domain modulation waveform of the present invention.

DETAILED DESCRIPTION

The technical solutions in the embodiments of the present invention will be described in detail below in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments.

The present invention proposes a fusion method for extracting soft information of heterogeneous signals based on a shortwave multi-channel diversity framework, as shown in FIG1 , which mainly includes the following steps:

Step S1 : Each sending station modulates and sends a signal source using a suitable modulation method and transmission rate.

Furthermore, a suitable modulation mode is selected through adaptive modulation technology, and a threshold decision method based on a signal-to-noise ratio or a threshold selection method based on a bit error rate can be adopted.

Step S2, demodulate the differently modulated signals separately and extract the corresponding symbol soft information.

Furthermore, the method for extracting symbol soft information is as follows:

A. Amplitude Domain Modulation ASK Soft Information Extraction

The process of the traditional envelope detection method of ASK is as follows: after receiving the signal, it passes through the rectifier circuit, converts the exchange signal into a DC signal, and then passes through a low-pass filter to filter out the envelope of the baseband signal. Finally, it is judged and output to complete the demodulation of the ASK signal.

The extraction of soft information is that when making a decision, the bit is not directly output as 0 or 1, but the specific value of the decision is directly output as symbol soft information. As shown in Figure 2, the sampling decision result after low-pass filtering can be directly output.

B. Frequency Domain Modulation MFSK Soft Information Extraction

Assuming that the received signal is ideally synchronized, a short-time Fourier transform analysis is performed on the received signal. The step length of the window function of the short-time Fourier transform is the same as the symbol time. The number of Fourier transform points is made to satisfy the spectrum resolution and the same carrier frequency interval of the FSK signal, and the time domain signal is converted into the time-frequency domain.

The spectrum of each symbol in the time-frequency domain information is analyzed. Since the frequencies corresponding to different symbols at the transmitting end are known, the frequency components corresponding to the signal time-frequency spectrum in the frequency domain distribution are taken. That is, the energy vector composed of this energy information is used as the soft information of this received symbol.

When M = 4, the extraction method is shown in Figure 3, where _TS represents the length of a single symbol and the baud rate is 1/ _TS . As can be seen from the figure, for a specific transmission symbol _Si , the corresponding soft information is _fj (j = 0, 1, ..., M-1), which is a soft information vector of length M, where each value represents the corresponding position

→Energy information. This scheme outputs the energy vector f as a whole as soft information.

C. Phase Domain Modulation MPSK Soft Information Extraction

The receiving end receives the MPSK signal s(t)=acos( _wct )-bsin( _wct ), multiplies it with cos( _wct ) and -sin( _wct ) respectively and integrates it, as shown in the following formula, and obtains the amplitude values a and b of the I and Q channels through sampling and judgment. Map it on the constellation diagram, a represents the value of the horizontal axis, and b represents the value of the vertical axis.

In the above formula, _wc is the carrier frequency T which is an integer multiple of T0=2Π/ _wc .

The phase of the received symbol is calculated from a and b, arctanb/a, denoted as θ.

Calculate the absolute value of the angle difference between θ and the phase of the four reference constellation points closest to the received symbol, Δθi, (i = 0 to 3). If Δθ exceeds 180°, then Δθ = 360°-θ, because the range of the angle difference is [0, Π).

That is, the four phase soft information values Δθ ₀ , Δθ ₁ , Δθ ₂ , Δθ ₃ of the MPSK symbol are obtained.

The extraction method when M=8 is shown in FIG4 . Although the received demodulated symbols are described by constellation points, the amplitude of the symbols does not change, so only the phase domain carries the modulation information.

D. Complex Domain Modulation QAM Soft Information Extraction

The multi-domain joint modulation method is QAM modulation, and its information is mapped on two signal domains, amplitude and phase. The constellation diagram of its soft information is shown in Figure 5.

Similar to PSK, after obtaining a and b, the soft information cannot be described by phase alone. Therefore, the distance between the received symbol constellation point and the reference constellation point is used to describe the soft information.

In the figure, the distances between the constellation position of the actual received symbol (p) and the nearest four 16QAM standard constellation points are d ₀ , d ₁ , d ₂ , and d ₃ , respectively. Note that this is different from the description method of energy soft information. The smaller the distance, the higher the credibility.

Step S3: uniformly map the symbol soft information into bit soft information.

Furthermore, PSK and QAM are first subjected to correlation mapping and then to bit mapping; ASK and MFSK are directly subjected to bit mapping.

Furthermore, the mapping of bit soft information is as follows:

A. Dependency Mapping

Correlation mapping mainly solves the problem that the size of soft information has different meanings. To swap the target result set, assume that the symbol set is:

P＝{P _i |i＝0,1,2,...,M-1}

The soft information corresponding to _Pi can be described as d _i , where M represents the number of symbol soft information, d _i in QAM represents the distance, and d _i in MPSK represents the absolute value of the angle difference. Here, the value of d _i is negatively correlated, and by taking the negation method, j represents the reference constellation point, and S _j is the soft information of this reference constellation point, and the positive correlation description of the soft information S _j can be obtained:

Taking the composite modulation 16QAM in Figure 5 as an example, there are

S0, S1, S2, S3 represent the soft information of the four symbols 1001, 1011, 1101, 1111 in Figure 5. d0, d1, d2, d3 represent the distance values of the receiving constellation point P from the nearest four reference constellation points 1001, 1011, 1101, 1111 in the figure.

It is worth noting that in 16QAM, M=16, but in the above formula, only a subset of the 4 nearest points is used for calculation. The reason for this is that in 16QAM, the distance between the constellation points at a longer distance is larger and the accuracy is extremely low. In order to reduce the computational consumption in the fusion process, it can usually be removed from the identification framework P to be fused to achieve the purpose of efficient calculation.

B. Bit Mapping

Bit mapping mainly solves the problem of different degrees of freedom of soft information. Taking 16QAM in Figure 5 as an example, the process is shown in Figure 6. In Figure 5, the received symbol P is described by M (M=4) soft information S ₀ , S ₁ , S ₂ , S _3. As shown in the constellation diagram, the corresponding bit information is 1001, 1011, 1101, 1111. Therefore, the coefficient matrix in Figure 6 can be constructed by mapping 0 to -1, and then the symbol soft information vector is multiplied by the coefficient matrix to realize the mapping of M symbol soft information to N bit soft information.

At this time, the 4 bits of soft information represented by the symbol P are: B ₀ , B ₁ , B ₂ , B ₃ .

A 16QAM symbol represents 4 bits of information, which means that the values of the four bits of soft information calculated based on the received symbol P are: B ₀ , B ₁ , B ₂ , B ₃ .

Step S4, realizing normalized mapping of bit soft information.

Furthermore, the bit soft information normalization mapping includes:

Normalization mapping mainly performs statistical analysis on the bit soft information and its noise, obtains a statistical model that the bit soft information of a certain moment roughly obeys, uniformly maps the bit soft information into the probabilistic form of the bit soft information, and limits the value of the soft information to the range of [0,1].

Assume that the noise of the bit soft information follows a Gaussian distribution, as shown in Figure 7. In the figure, B is the soft information after bit mapping, which is defined as the occurrence of event B. _A0 represents the event that the bit is 0, and _A1 represents the event that the bit is 1. _A0 and _A1 are two incompatible events.

The probabilities P(B|A ₀ ) and P(B|A ₁ ) in the figure are described as:

Wherein S is the soft information after bit mapping, μ ₀ and σ ₀ are normal distribution parameters of the metric value with bit 0, and μ ₁ and σ ₁ are normal distribution parameters of the metric value with bit 1.

According to the posterior probability formula:

Where P(A ₀ ) represents the probability that the current bit value in the received information is 0, and P(A ₁ ) represents the probability that the current bit value in the received information is 1. In order to ensure the maximum information entropy, the sent value needs to satisfy P(A ₀ )=P(A ₁ ), and finally the conversion from the metric value of the bit soft information to the probability value is obtained:

P(A ₀ |B) and P(A ₁ |B) are the final normalized bit soft information, denoted as m(0) and m(1).

Step S5, obtaining a diversity gain by adding, and obtaining a decoding result by performing a majority decision.

Further, including:

The normalized bit soft information is added and fused, and the larger value of M(0) and M(1) is taken as the decoding result:

M(0)＝m1(0)+m2(0)+,...,ms(0)

M(1)＝m1(1)+m2(1)+,...,ms(1)

mi(0) is the normalized bit soft information value of a bit of the i-th branch being 0. M(0) is the fused and added value of a corresponding bit of all branches being 0. Similarly, mi(1) is the normalized bit soft information value of a bit of the i-th branch being 1. M(1) is the fused and added value of a corresponding bit of all branches being 1.

The above embodiments should be understood to be only used to illustrate the present invention and not to limit the protection scope of the present invention. After reading the contents of the present invention, technicians can make various changes or modifications to the present invention, and these equivalent changes and modifications also fall within the scope defined by the claims of the present invention.

Claims (8)

1. The fusion method for heterogeneous signal soft information extraction based on the short-wave multichannel diversity framework is characterized by comprising the following steps of:

step S1, each transmitting station modulates and transmits the information source by using a proper modulation mode and transmission rate;

s2, independently demodulating signals with different modulations, and extracting corresponding symbol soft information;

step S3, the symbol soft information is mapped into bit soft information in a unified way;

s4, realizing the normalized mapping of the bit soft information;

and S5, obtaining diversity gain by utilizing an addition form, and obtaining a decoding result through multi-decision.

2. The fusion method for heterogeneous signal soft information extraction based on short-wave multichannel diversity framework of claim 1, wherein the extracting of the corresponding symbol soft information in step S2 comprises:

the extraction of the composite domain modulation QAM soft information comprises the following steps: the multi-domain joint modulation mode is QAM modulation, information of which is mapped on two signal domains of amplitude and phase, and soft information is described by adopting the distance between a received symbol constellation point and a reference constellation point.

3. The fusion method of heterogeneous signal soft information extraction based on short-wave multi-channel diversity framework according to claim 1, wherein the step S3 comprises:

PSK and QAM firstly carry out correlation mapping and then carry out bit mapping; ASK and MFSK directly perform bit mapping.

4. A fusion method of heterogeneous signal soft information extraction based on a short wave multi-channel diversity framework according to claim 3, wherein the correlation mapping comprises:

wherein S is _j For this purpose soft information, d, of the reference constellation point j _i The symbol P _i Corresponding soft information, d in QAM _i Represents distance d in MPSK _i Representing the absolute value of the angle difference, the symbol set P is:

P＝{P _i |i＝0,1,2,...,M-1}

where M represents the number of symbol soft information.

5. A fusion method of heterogeneous signal soft information extraction based on a short wave multi-channel diversity framework according to claim 3, wherein the bit map comprises:

mapping the symbol soft information onto the bit soft information is achieved by mapping 0's of the bit information into-1's constituting a coefficient matrix and then multiplying the coefficient matrix by a symbol soft information vector.

6. The fusion method for heterogeneous signal soft information extraction based on short-wave multichannel diversity framework of claim 1, wherein the bit soft information normalization mapping in step S4 comprises:

and carrying out data statistics analysis on the bit soft information and noise thereof to obtain a certain statistical model modeling obeyed by the bit soft information of a certain moment, uniformly mapping the bit soft information into a probability form of the bit soft information, and limiting the value of the soft information within the range of [0,1 ].

7. The fusion method for heterogeneous signal soft information extraction based on short wave multi-channel diversity framework according to claim 6, wherein uniformly mapping the bit soft information into a probability form of the bit soft information comprises:

assuming that the noise of the bit soft information is subject to Gaussian distribution, the bit mapped soft information is defined as occurrence of events B and A ₀ An event representing that bit is 0, A ₁ An event representing the bit being 1, A ₀ And A ₁ 2 mutually incompatible events;

probability P (B|A) ₀ ) And P (B|A) ₁ ) The description is as follows:

wherein S is soft information after bit mapping, mu ₀ 、σ ₀ Normal distribution parameter, μ, for a metric value of 0 bits ₁ 、σ ₁ A normal distribution parameter that is a metric value with bit 1;

according to the posterior probability formula:

wherein P (A) ₀ ) Representing the probability that the current bit value in the received information is 0, P (A ₁ ) Representing the probability that the current bit value in the received information is 1, the transmitted value needs to satisfy P (A ₀ )＝P(A ₁ ) Finally, the bit soft is obtainedConversion of metric values to probability values for information:

P(A ₀ i B) and P (A) ₁ I B) is the final normalized bit soft information, denoted as m (0) and m (1).

8. The fusion method of heterogeneous signal soft information extraction based on short-wave multi-channel diversity framework according to claim 1, wherein the step S5 comprises:

adding and fusing the normalized bit soft information, and taking larger values in M (0) and M (1) as decoding results:

M(0)＝m1(0)+m2(0)+,...,ms(0)

M(1)＝m1(1)+m2(1)+,...,ms(1)

and in the same way, mi (0) is a normalized bit soft information value with a certain bit of 0 of the ith branch, M (0) is a fusion added value with a certain corresponding bit of 0 of all branches, mi (1) is a normalized bit soft information value with a certain bit of 1 of the ith branch, and M (1) is a fusion added value with a certain corresponding bit of 1 of all branches.

Images