Channel equalization method for short-wave communication system

Abstract

The invention belongs to the technical field of short-wave communication, and particularly relates to a channel equalization method for a short-wave communication system, which comprises the following steps of: constructing the short-wave communication system which comprises a transmitting end and a receiving end; acquiring an information sequence, and inputting the information sequence to a transmitting end; the sending end encodes the input information sequence and sends the encoded data to the receiving end; the receiving end carries out diversity receiving processing on the received data and carries out decoding processing on the received signal; feeding back the decoded signal to an equalizer to complete channel equalization; the invention provides a frequency domain Turbo equalization algorithm combining expectation propagation and diversity reception, which utilizes EP approximate iteration to estimate prior and posterior distribution of transmission symbols, and effectively solves inter-symbol crosstalk by iteratively exchanging soft information between an equalizer and a decoder.

Description

Technical Field

[0001] The invention belongs to the technical field of shortwave communication, and in particular relates to a channel equalization method for a shortwave communication system. Background Art

[0002] Shortwave communication is a wireless communication method in harsh environments. The communication frequency band is between 3 and 30MHz, with wide coverage and strong penetration, and is particularly suitable for long-distance transmission across the atmosphere. However, since the height and density of the ionosphere are affected by factors such as day and night, season, and climate, shortwave communication has inter-symbol interference caused by time and frequency dispersion and fast fading caused by multipath, which seriously affects the receiving performance of the receiving end.

[0003] Channel equalization is an effective method to deal with channel distortion and eliminate inter-symbol interference. Traditional equalization includes time domain equalization, frequency domain equalization and blind equalization. Among them, Turbo equalization, which draws on the idea of ​​Turbo code, can effectively solve inter-symbol interference. The Turbo equalization process includes two parts: equalization and decoding, among which the soft input soft output (SISO) equalizer is the core part. The maximum a posteriori probability (MAP) equalization algorithm is the algorithm with the lowest bit error rate and the best performance, but the computational complexity of the algorithm increases exponentially with the increase of channel length. When high-order modulation is used, it involves a large number of logarithmic and multiplication operations, which is too complex and difficult to implement. The soft interference cancellation (SIC) algorithm is relatively simple and has a small amount of calculation, but its performance is greatly affected by prior information. When the channel conditions are relatively bad, the performance is poor. The performance and complexity of the linear minimum mean square error (LMMSE) algorithm are a compromise between MAP and SIC. The SISO equalizer based on LMMSE obtains the linear filter coefficient by minimizing the mean square error, filters the received signal to obtain soft information, and exchanges soft information with the decoder. For the difficult-to-obtain prior information, the LMMSE algorithm assumes that both the signal and the noise obey the Gaussian distribution. In the first iteration, the discrete true prior information of the transmission symbol is replaced by the Gaussian prior information, and the conditional distribution of the combined observation data is calculated to minimize the mean square error, obtain the Gaussian posterior probability distribution, and complete the estimation of the transmission symbol. The above existing methods have the problem of poor communication performance of the shortwave channel due to the severe fading and multipath of the shortwave channel. Summary of the invention

[0004] In order to solve the problems existing in the above prior art, the present invention proposes a channel equalization method for a shortwave communication system, which method comprises: constructing a shortwave communication system, which comprises a transmitting end and a receiving end; obtaining an information sequence, and inputting the information sequence into the transmitting end; the transmitting end encodes the input information sequence, and sends the encoded data to the receiving end; the receiving end performs diversity reception processing on the received data, and decodes the received signal; and the decoded signal is fed back to the equalizer to complete channel equalization.

[0005] Beneficial effects of the present invention:

[0006] The present invention proposes a frequency domain Turbo equalization algorithm that combines expected propagation with diversity reception. The algorithm uses EP approximate iteration to estimate the prior and posterior distribution of the transmission symbol, and effectively solves the inter-symbol interference by iteratively exchanging soft information between the equalizer and the decoder. At the same time, in order to eliminate the fading caused by multipath, the diversity reception technology is used to respectively equalize the signals of different branches and then perform maximum ratio combining (MRC) on the external information output by the multi-path equalization. The sum of the combined external information is sent to the decoder after deinterleaving. The information output by the decoder is then interleaved and mapped and sent back to the equalizer for the next equalization. Finally, simulations are carried out under different modulation modes and shortwave channels. The results show that when the bit error rate is 10 -4 ,When the modulation mode is 4QAM, the proposed algorithm has a performance gain of about 0.9dB compared with the traditional algorithm in ,shortwave channels with different interference conditions. BRIEF DESCRIPTION OF THE DRAWINGS

[0007] Figure 1 It is a structural diagram of the transmitting end of the present invention;

[0008] Figure 2 It is a structural diagram of a receiving end of the present invention;

[0009] Figure 3 It is a structural diagram of an equalizer for diversity reception of the present invention;

[0010] Figure 4 It is a diagram showing the iteration of the EP-DC-FDTE algorithm of the present invention under the HFMD channel;

[0011] Figure 5 It is a three-dimensional analysis structure diagram of the EP-DC-FDTE algorithm of the present invention under the HFMD channel;

[0012] Figure 6 It is a diagram showing the iteration of the EP-DC-FDTE algorithm of the present invention under different channels;

[0013] Figure 7 It is a comparison chart of iterative performance between the algorithm of the present invention and the traditional LMMSE algorithm;

[0014] Figure 8 Iteration of the EP-DC-FDTE algorithm of the present invention under the HFMD channel. DETAILED DESCRIPTION

[0015] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions 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. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0016] A channel equalization method for a shortwave communication system, the method comprising: constructing a shortwave communication system, the system comprising a transmitting end and a receiving end; obtaining an information sequence, and inputting the information sequence into the transmitting end; the transmitting end performs encoding processing on the input information sequence, and sends the encoded data to the receiving end; the receiving end performs diversity reception processing on the received data, and decodes the received signal; and the decoded signal is fed back to an equalizer to complete channel equalization.

[0017] In this embodiment, the communication sending end structure block diagram is as follows: Figure 1 As shown, an information sequence consisting of K information bits a=[a 1 ,...,a k ] T Encoded at bit rate R = K / V as b = [b 1 ,...,b v ] T , after interleaving, we get c = [c 1 ,...,c v ] T The interleaved data sequence is modulated by an M-order constellation to obtain a symbol sequence x = [x 1 ,...,x N ] T Where N = [V / log 2 M], data frame x = R (x) + jI (x) through the channel h = [h 1 ,...,h L ] transmission, each symbol is represented by in represent The symbol set of the order constellation. k The representative variance is The average energy per symbol and per bit of the transmitted signal are represented by E s and E b express.

[0018] The channel impulse response is h = [h1 ,...,h L ] T , the noise variance is Received signal y = [y 1 ,...,y N+L-1 ] T , as shown below:

[0019]

[0020]

[0021] The corresponding matrix form is:

[0022] y=Hx+w

[0023] The discrete time expression is:

[0024]

[0025] in When k<1 or k>N, x k =0.

[0026] The above model can give the posterior probability of the transmitted symbol vector x:

[0027]

[0028] The indicator function The values ​​are:

[0029]

[0030] If the prior information is unknown to the channel decoder, it is assumed that the transmitted symbols satisfy a uniform distribution and their prior probabilities satisfy:

[0031]

[0032] This assumption is used to provide prior information to the equalizer before the Turbo receiver performs iterations.

[0033] like Figure 2 As shown in Figure 1, the key to Turbo equalization at the receiving end is that the equalizer and decoder iteratively exchange information on the same set of received symbols. The decoder input is the new information LLR L calculated by the equalizer. E (b t |y) is the information used for iteration output by the equalizer during the iteration process, abbreviated as L E (b t ), the decoder calculates an estimate of the information bits after one or more iterations And the outer LLRs of the coded bits:

[0034]

[0035] Among them, p D (x k ) are remapped, such as Figure 1 As shown (Π and Π -1 represents the interleaving mapping and its inverse process), and returns it to the equalizer as the updated prior probability. This process is performed for a given maximum number of iterations T until convergence.

[0036] In this embodiment, in the classic diversity combining technology, when the diversity multiplicity M is the same, the maximum ratio combining improves the received signal performance the most because it fully considers the signal of each branch. The weight p k The signal envelope r of this branch k (t) is proportional to the noise power n k Inversely proportional, that is:

[0037]

[0038] It can be seen that the signal envelope of the maximum ratio combining output is:

[0039]

[0040] In Turbo equalization, the received M-diversity branch signals are Each equalizer is used to perform equalization separately, and then the multi-channel external information output by each equalizer is combined at the maximum ratio. The sum of the combined external information is sent to the decoder after deinterleaving. The external information decoded by the decoder is then interleaved and mapped, and sent back to each equalizer for another equalization, thereby realizing the continuous iteration of soft information between multiple equalizers and a single decoder. The principle block diagram is shown in the figure. Figure 3 shown.

[0041] In this embodiment, the Turbo equalizer design includes: the posterior probability approximate solution provided by the MSE equalizer will have a mean and variance The independent Gaussian function of is used to approximate the discrete prior probability p(x), and the approximate distribution is expressed as:

[0042]

[0043] Among them, μ MMSE and Σ MMSE is the mean and covariance matrix of the symbol sequence, whose values ​​are:

[0044]

[0045]

[0046] Initialization settings in the first iteration

[0047]

[0048] In the iterative process, it is assumed that the probability of the k-th prior information symbol is equal, the calculated external information of the symbol is passed to the channel decoder, and the equalizer feedback statistics and is the updated prior probability p D (x k ) is obtained, and the calculation method is as follows:

[0049]

[0050] If it is assumed that the transmitted symbols are equally distributed, the posterior probability distribution can be written as:

[0051]

[0052] Where δ(t) represents the Dirichlet function, when t is 0, δ(t) is 1, otherwise δ(t) is 0.

[0053] In this embodiment, the expectation propagation algorithm, as an estimation algorithm based on Bayesian reasoning, uses an easy-to-handle probability distribution to approximate a probability distribution that needs to be solved, and iteratively updates the parameter values ​​in the approximate distribution according to the criterion of minimizing the KL divergence, thereby obtaining an accurate approximate distribution that is closer to the true distribution. The accuracy of the approximation is measured by the relative entropy KL divergence, as follows:

[0054]

[0055] From the above formula, we can see that the approximate accuracy of p(x) and q(x) increases as the relative entropy decreases. Therefore, in order to make the approximate p(x) optimal, we can get the optimal p(x) by taking the derivative of the relative entropy and setting it to 0.

[0056] Assuming that the statistical distribution and observed variables of a hidden variable x are given, the complex posterior probability distribution formed by multiplying I (i = 1, ..., I) non-negative factors is:

[0057]

[0058] where t(x) represents a set of known approximate probability functions that obey exponential distribution. i (x) represents the objective function that does not belong to the exponential family. By using the EP algorithm, according to the moment matching condition, the objective function that does not belong to the exponential family is approximated, that is, using the exponential family To replace the objective function f i (x). Then we use the probability distribution function in the exponential family To obtain the estimator q(x) of the posterior probability distribution:

[0059]

[0060] The specific implementation process is as shown in Algorithm 1, where q (l) (x) represents the approximate distribution corresponding to the lth iteration:

[0061] Algorithm 1 Expectation Propagation Algorithm

[0062] Input: Initialize the approximate distribution factor According to the initialized approximate distribution factor, q(x) is calculated, where S is the number of iterations of EP.

[0063] for i=1,...,Iandl=1,...,S do

[0064] 1) Calculate the following probability distribution:

[0065]

[0066] The above probability distribution is called the approximate marginal probability distribution, which is obtained by approximating the probability distribution.

[0067] 2) Calculate the following probability distribution and corresponding moments:

[0068] 3) Update the approximate probability distribution of the next iteration through "moment" matching

[0069]

[0070] end for

[0071] Output: approximate probability distribution

[0072] In this embodiment, the EP-MDR-FDTE algorithm includes: according to the EP algorithm, a non-regular exponential distribution of the exponential family is used to replace the non-Gaussian term in the posterior probability distribution, and when the channel condition is known a priori, the structure of the real distribution is retained as much as possible, and the unknown factor f is i (x) Separated from the expression of the probability distribution, the following Gaussian exponential family is a reasonable approximation to the posterior probability distribution:

[0073]

[0074] The EP-based equalization algorithm continuously updates the parameter pair (γ k , Λ k ), k = 1, ..., N, so as to further update the approximate posterior probability. The approximate posterior probability q(x) also obeys the exponential distribution. According to the Gaussian product lemma, its mean and variance are derived as follows:

[0075]

[0076] The above formula has a similar structure to the LMMSE equalization method, but the EP algorithm can achieve better performance gain with lower computational complexity through iterative update of parameter pairs and transfer between parameter pairs.

[0077] The calculation method of a single branch in the EP-MDR-FDTE algorithm is as follows. It can be regarded as a turbo equalization. The EP approximation algorithm is used to approximate the outer distribution of M branches through the cavity function, iterate S times, perform equalization separately, merge the demapped results, and then deinterleave and input them to the decoder, continue to iterate, and repeat T times. The output of the channel decoder is used to initialize the EP iteration process, and then its output is fed forward to the channel decoder.

[0078] Algorithm 2EP-MDR-FDTE algorithm

[0079] Input:

[0080] in are the parameters after Fourier transformation.

[0081] fort=1,...,T do

[0082] fork=1,...,Ndo

[0083] 1) Start EP iteration and calculate q under initial conditions (l) The mean and variance (second moment) of (x).

[0084] forl=1,...,S do

[0085] fork=1,...,N do

[0086] 2) Calculate q (l) Marginal distribution of the kth symbol in (x).

[0087] 3)

[0088]

[0089] in

[0090]

[0091] 4) Calculate the approximate posterior probability distribution

[0092]

[0093] Estimate its mean and variance 5)

[0095]

[0096] and The second moments of are equal.

[0097] 6) Update parameter pair

[0098] end for

[0099] end for

[0100] end for

[0101] 7) After the EP iteration, use the parameters to Calculate the final distribution q(x). Calculate p E (x k |y)=q (S +1)\k (x k ) and perform inverse Fourier transform on the detected symbols, input them into the demapper, and calculate the external LLRsL E (b t |y) and then MRC combined and transmitted to the channel decoder. From each soft output of the channel decoder, each symbol p in the decoder is recalculated D (x k ) and calculate its mean and variance Feedback to each branch.

[0102] 8) Reinitialize the parameter pair.

[0103] end for

[0104] In this embodiment, equalization mainly solves the problem of inter-code interference in shortwave communication, while multipath fading cannot be properly handled. Diversity combining is an effective way to handle multipath fading in shortwave communication. In order to further improve the shortwave data transmission rate and reduce the bit error rate of shortwave communication, multipath diversity technology and equalization technology can be combined at the receiving end. While reducing multipath fading, the influence of inter-code interference is reduced, thereby effectively improving the performance of the receiving end.

[0105] In order to verify the effectiveness of the algorithm in this paper, it is compared with the traditional Turbo equalization algorithm. The diversity multiplicity is 2, and simulations are performed under different shortwave channels. Taking into account the influence of factors such as terrain and atmosphere, as well as the channel characteristics of radio waves in shortwave communication, including reflection, scattering and diffraction, multipath fading problems are caused. The channel adopts iturHFMQ, iturHFMM, and iturHFMD models, with a multipath delay of 2ms and a Doppler shift of 1Hz. The channel parameter settings are consistent with the description of shortwave channels in ITU-R, and the specific parameters are as follows.

[0106] Table 1 Channel parameters

[0107]

[0108] The signal modulation mode is QAM modulation, the coding mode is 1 / 2 LDPC coding, the generating polynomial G = [576, 1152], the information bits per frame are 576 bits, the decoding uses LDPC soft decision decoding, and the interleaver uses random interleaving. The number of iterations is 5.

[0109] Figure 4 The iteration of the improved algorithm in HFMD channel is given. As the signal-to-noise ratio increases, the bit error rate also gradually decreases. Where T represents the number of Turbo iterations. After five iterations, when the bit error rate is 10 -4 When , there is a performance gain of about 1dB. It can be seen that under the HFMD channel condition, the algorithm reaches convergence after five iterations.

[0110] In order to more intuitively see the relationship between the number of iterations, signal-to-noise ratio and bit error rate, this paper conducts a three-dimensional analysis of the EP-DC-FDTE algorithm. Figure 5 It is not difficult to see that the bit error rate decreases with the increase of iteration number and signal-to-noise ratio. When both increase at the same time, the improvement of bit error performance is most obvious.

[0111] Figure 6 The bit error rates of the improved algorithm after one and five iterations in HFMQ, HFMM, and HFMD channels are given. It can be seen that the bit error rates in HFMQ and HFMM channels are lower than those in HFMD channels. This is because HFMQ and HFMM are shortwave mid-latitude quiet conditions and shortwave mid-latitude medium conditions, respectively. Compared with shortwave mid-latitude disturbance conditions, the bad channel conditions have been improved. Moreover, the performance improvement in the three channels can reach a maximum of about 1dB.

[0112] Figure 7The simulation results of the improved algorithm in this paper and the algorithm that only uses the LMMSE algorithm without using the EP approximation and diversity merging are given. It can be seen that the performance of the improved algorithm is almost the same as that of the unimproved algorithm after only one iteration. After five iterations to reach convergence, the improved algorithm has a bit error rate of 10 -4 There is approximately a 0.9dB performance gain.

[0113] Figure 8 The iteration of the improved algorithm under 4QAM and 16QAM modulation in HFMD channel is given. It can be seen that within a certain range, as the signal-to-noise ratio increases, the error performance is gradually improved, which shows that the algorithm has a high adaptability when facing different order modulations.

[0114] The above embodiments further illustrate the purpose, technical solutions and advantages of the present invention in detail. It should be understood that the above embodiments are only preferred implementation modes of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made to the present invention within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A channel equalization method for a shortwave communication system, characterized in that: include: Construct a shortwave communication system, which includes a transmitting end and a receiving end; The information sequence is obtained and input into the transmitter; the transmitter encodes the input information sequence and sends the encoded data to the receiver; the receiver performs diversity reception on the received data and decodes the received signal; the decoded signal is fed back to the equalizer to complete channel equalization.

2. A channel equalization method for a shortwave communication system according to claim 1, characterized in that: The transmitting end includes an encoder, an interleaver and a mapping module; the transmitting end processes the information sequence including: the encoder encodes the information sequence; the encoded information is input into the interleaver for interleaving; the interleaved signal is mapped through the mapping module, and the mapped signal is modulated by an M-order constellation to obtain a symbol sequence.

3. A channel equalization method for a shortwave communication system according to claim 2, characterized in that: The encoder encodes the information sequence including: information sequence a = [a1, ..., a k ] T It consists of K information bits, and the information sequence is encoded into b = [b1, ..., b v ] T , where the code rate is the ratio of the information sequence length to the encoded sequence length, that is, K information bits are encoded into V information bits; a k is the kth information bit, T is the transposition, K is the number of information bits, V is the number of encoded information bits, b v is the vth information bit in the encoded sequence.

4. A channel equalization method for a shortwave communication system according to claim 2, characterized in that: The interleaving process of the encoded information includes: using a random interleaving method to interleave the encoded information sequence b = [b1, ..., b v ] T Interleave to generate a random permutation index P = [p1, ..., p n ], where p i ∈{1,...,n}, and all p i Uniquely, map the i-th element of the original data to p i The interleaved sequence c = [c1,...,c v ] T .

5. A channel equalization method for a shortwave communication system according to claim 2, characterized in that: Mapping the interleaved signal includes: the interleaved data sequence is modulated by an M-order constellation to obtain a symbol sequence x=[x1,...,x N ] T , where N = [V / log2M], the symbol sequence x = R(x) + jI(x) through the channel h = [h1, ..., h L ] transmission, each symbol is represented by in, represent The symbol set of the constellation is an order constellation, V is the number of coded information bits, N is the length of the modulated symbol sequence, M is the modulation order, R(x) is the real part of the complex signal, and I(x) is the imaginary part of the complex signal.

6. A channel equalization method for a shortwave communication system according to claim 1, characterized in that: The receiving end includes multiple equalizers, multiple demappers, a fusion device and a decoder; the diversity reception processing of the received data includes: using each equalizer to respectively equalize the M-diversity branch signals; inputting the equalized signals into the corresponding demappers for demapping, and inputting the demapped signals into the fusion device to obtain a fused signal; calculating the posterior probability of the fused signal, and inputting the signal into the decoder according to the posterior probability to obtain external information; interleaving and mapping the external information, and returning it to each equalizer for another equalization of each channel, thereby realizing the continuous iteration of soft information between multiple equalizers and a single decoder.

7. A channel equalization method for a shortwave communication system according to claim 6, characterized in that: The equalization of the M-fold diversity branch signals includes: calculating the prior distribution for each multipath branch, initializing the moment information of the prior distribution Calculate the posterior probability distribution from the prior distribution According to the moment matching condition, the EP algorithm is used to approximate the objective function that does not belong to the exponential family, that is, to use the To replace the objective function f i (x); using the probability distribution function in the exponential family To obtain the estimator q(x) of the posterior probability distribution: in, is the mean of the Kth symbol, x k is the Kth transmission symbol, is the variance of the Kth symbol, E s is the average energy of the transmitted symbol, Σ x is the covariance matrix, μ x is the mean vector, x is the received vector, y is the observed vector, ∝ means proportional to, It means it obeys complex Gaussian distribution, H is the channel matrix, is the noise variance, I is a diagonal matrix whose diagonal elements are all 1, δ is the impulse function, and t(x) is the part of the objective function that belongs to the exponential function.

8. A channel equalization method for a shortwave communication system according to claim 6, characterized in that: The demapper demaps the signal by mapping each symbol after equalization into a bit sequence and performing deinterleaving at the same time.

9. A channel equalization method for a shortwave communication system according to claim 6, characterized in that: The decoder decodes the signal and restores the encoded bit sequence.

Images