Direct adaptive bidirectional Turbo equalization method for FMT underwater acoustic communication

Abstract

The invention discloses a direct adaptive bidirectional Turbo equalization method for FMT underwater acoustic communication, and the method comprises the steps: carrying out the FMT modulation of an information bit stream at a transmitting end after the encoding, interleaving and mapping, and transmitting the information bit stream to an underwater acoustic channel through a transmitting array element; and at a receiving end, the signal demodulated by the FMT is processed by utilizing direct self-adaptive bidirectional Turbo equalization. The provided direct adaptive bidirectional Turbo equalization is mainly composed of bidirectional equalization, interleaving, de-interleaving and a decoder. The bidirectional equalization is composed of forward equalization and reverse equalization which are the same in structure, the forward equalization and the reverse equalization are both preprocessed by adopting a time reversal combiner, the adaptive decision feedback equalizer based on soft information is adopted for post-processing, and soft information output by the forward equalization and soft information output by the reverse equalization are weighted and merged to form soft information of the bidirectional equalization. Soft information exchange is continuously carried out between the bidirectional equalization and the decoder until the communication performance tends to be stable. The method improves the reliability of a communication system, and is suitable for high-performance information transmission in a complex underwater acoustic channel.

Description

Technical Field

[0001] The present invention belongs to the technical field of underwater acoustic communication, and particularly relates to a direct adaptive two-way Turbo equalization method for FMT underwater acoustic communication. Background Technique

[0002] Multi-carrier underwater acoustic communication is an important part of high-rate and high-performance underwater acoustic communication. In the past few decades, due to its simple implementation, high bandwidth efficiency, and the ability to avoid inter-symbol interference (ISI), orthogonal frequency division multiplexing (OFDM) has been the most widely studied and applied multi-carrier technology in underwater acoustic communication. However, due to the overlapping of sub-bands and narrow bandwidth, OFDM underwater acoustic communication is very sensitive to frequency offset, and a slight frequency offset will lead to a significant decline in communication performance. In recent years, filtered multitone (FMT) underwater acoustic communication has attracted the research interest of researchers due to its advantages such as simple implementation, high frequency band utilization, and insensitivity to frequency offset.

[0003] Although FMT underwater acoustic communication can maintain the characteristic of being insensitive to frequency offset by dividing non-overlapping sub-bands with relatively wide bandwidths, there is still ISI in the communication because the bandwidth of the sub-bands is greater than the coherence bandwidth of the underwater acoustic channel. Initially, traditional adaptive equalizers were used in FMT underwater acoustic communication to suppress ISI. Subsequently, a method combining multi-channel time reversal (TR) merging and adaptive equalization was proposed. Compared with traditional adaptive equalization, this method can effectively suppress ISI in FMT underwater acoustic communication and has a lower computational complexity. However, there is also noise in the underwater acoustic channel, which limits the performance of this method.

[0004] Therefore, Turbo equalization, which combines channel equalization and decoding, is applied to FMT underwater acoustic communication. A low-complexity Turbo linear equalization method based on the Minimum Mean Square Error (MMSE) is proposed. The performance of this method is superior to that of traditional linear equalization. However, its equalization coefficients are adjusted according to accurate channel estimation, and a large number of iterations are required to achieve convergence, which is not suitable for time-varying channels. Therefore, a direct adaptive Turbo decision feedback equalization method is proposed. This method uses an adaptive decision feedback equalizer (DFE) based on soft information to equalize the signal, and the equalization coefficients are adaptively adjusted using the Recursive Expected Least Squares (RELS) algorithm. Although this method can be adaptively adjusted based on channel characteristics, and its computational complexity is of the same order of magnitude as the traditional Recursive Least Squares (RLS) algorithm and the amount of computation is also low, it still has the following disadvantages. First, Turbo DFE only eliminates causal ISI and cannot suppress non-causal ISI, so its interference suppression performance is limited. Second, when the decision values fed back in the DFE are incorrect, error propagation may occur, which will lead to a serious decline in communication performance. Summary of the Invention

[0005] To solve the technical problems existing in the background art, the present invention aims to provide a direct adaptive two-way Turbo equalization method for FMT underwater acoustic communication, which can not only avoid error propagation but also obtain additional diversity gain through two-way adaptive Turbo equalization to improve the ISI suppression performance of FMT underwater acoustic communication.

[0006] To solve the technical problems, the technical solution of the present invention is as follows:

[0007] A direct adaptive two-way Turbo equalization method for FMT underwater acoustic communication, the method comprising:

[0008] S1: At the transmitting end of FMT underwater acoustic communication, the original binary data stream is input, and the error control coding ECC is preset. After the encoded binary bit stream is interleaved, mapped, and FMT modulated, it is sent into the underwater acoustic channel through the transmitting array element;

[0009] S2: At the receiving end of FMT underwater acoustic communication, the received signal is input, and the forward equalization and the backward equalization with the same structure are combined to construct a two-way equalization. Both the forward and backward equalizations use a multi-channel TR combiner for preprocessing, and a direct adaptive decision feedback equalization (DFE) based on soft information is used for postprocessing. The bit probabilities output by the combined two-way equalization are transmitted to the decoder, and the information exchange between the two-way equalization and the decoder is repeated until the performance is stable. Finally, the original binary bit stream recovered by the decoder is output.

[0010] Further, in the receiving end and decoder of FMT underwater acoustic communication, the iterative information exchange between the two-way equalization and the decoder specifically includes:

[0011] S2101: In the forward equalization, first, the sub-band signals after FMT demodulation are preprocessed by the TR combiner to achieve pre-compression of the ISI. Subsequently, a direct adaptive DFE based on soft information is used for postprocessing, and the symbol probabilities output by the processing are input to the demapper.

[0012] S202: In the backward equalization, first, the sub-band signals after FMT demodulation are stored using a buffer, and time reversal processing is performed in the way that the signal that enters the buffer first is output last, while the signal that enters the buffer last is output first. Subsequently, the time-reversed signals are preprocessed by the TR combiner, and a direct adaptive DFE based on soft information is used for postprocessing.

[0013] S203: The symbol probabilities output in the backward equalization are stored using a buffer, and time reversal processing is performed in the way that the one that enters the buffer first is output last, while the one that enters the buffer last is output first. Then, the time-reversed symbol probabilities are input to the demapper.

[0014] S204: The two-way symbol probabilities are demapped into two-way binary bit probabilities respectively and input to a combiner for combination. The combined output bit probabilities are de-interleaved. The de-interleaved output bit probabilities are used as the prior information for this decoding after subtracting the bit probabilities of the previous decoding output, while the bit probabilities of the current decoding output are input to the two-way equalization for the next equalization processing after interleaving. The information interaction between the two-way equalization and the decoding is sequential and iterative, and stops when the system performance reaches stability. Finally, the original binary bit stream recovered by the decoder is output.

[0015] Further, the specific steps of step S201 include:

[0016] In the forward equalization, the signal output by the TR combiner is:

[0017]

[0018] In the above formula, m is the sub - band serial number, q (m,m) (nT, lT) represents the m - th sub - band sequence s at the transmitting end m of (lT) and the output signal r of the m - th TR combiner in the forward structure at the receiving end m the combined channel response between (nT), s m (nT) represents s when l = n m the value of (lT), and T represents the symbol interval;

[0019] The direct adaptive DFE based on soft information is used for post - processing, and the equalized output estimate is:

[0020]

[0021] In the above formula, f m (n) represents the filter coefficient vector, represents the forward filter coefficient vector, represents the feedback filter coefficient vector, [·] Η represents the conjugate transpose operation, and the filter coefficient vector is adaptively adjusted by the recursive expectation - minimum - squares (RELS) method, r m (n) represents the input signal vector of the equalizer, represents the vector composed of the decision values of the feedback filter input;

[0022] For the signal make a decision to obtain the output probability of the symbol:

[0023]

[0024] In the above formula, p in [s m (n)=s] represents the input prior probability when the transmitted symbol s m (n)=s, represents m conditioned on s the probability distribution of, and this value depends on the filter coefficient vector f m (n), and all possible values of s form the symbol set

[0025] Furthermore, the step S202 specifically includes:

[0026] The TR combiner in the reverse equalization is used to process the time - reversed version of the FMT demodulation signal, and the output signal is:

[0027]

[0028] In the above formula and Respectively represent the combined channel response and noise after TR combination in the reverse equalization. Since the transmit filter matches the receive filter and the m-th TR filter matches the m-th sub-band channel, the combined channel response of the reverse equalization is equal to the combined channel response q (m,m) (lT,nT);

[0029]

[0030] In the reverse equalization, an adaptive DFE based on soft information is used to post-process the signal The adjustment process of the equalization coefficient is the same as that of the forward equalization described in step S201.

[0031] Furthermore, in step S204, the gain combination is specifically:

[0032] Γ out (d m (n)) = λ n Γ nor (d m (n)) + λ b Γ bw (d m (n))

[0033] In the above formula, λ n and λ b respectively represent the combination coefficients of the forward and reverse equalizations, and their value ranges are from 0 to 1. Γ nor (d m (n)) and Γ bw (d m (n)) represent the log-likelihood values of the binary information bits obtained after mapping the output probabilities of the forward and reverse equalizations.

[0034] Compared with the prior art, the advantages of the present invention are as follows:

[0035] Two-way equalization combination and time reversal processing:

[0036] Traditional equalization methods usually only rely on forward equalization, while the present invention effectively processes ISI caused by multipath effects by introducing reverse equalization and combining TR technology. The reverse equalization can eliminate non-causal interference by time-reversing the received signal and using a filter with the same structure as the forward equalization, thereby improving the signal quality and providing additional diversity gain.

[0037] Direct adaptive two-way Turbo equalization structure:

[0038] The present invention adopts a bidirectional equalization structure, combines the output information of forward and reverse equalizations, and performs iterative information exchange with the decoder, enabling continuous optimization of the prior information between the equalizer and the decoder, thereby enhancing the accuracy of signal recovery. This iterative processing combines the dynamic update of soft information, effectively improving the anti-interference performance and robustness of the system.

[0039] Adaptive DFE based on soft information:

[0040] In the equalization post-processing stage, an adaptive DFE based on soft information is used to optimize decision feedback. This method dynamically adjusts the filter coefficients based on RELS, effectively suppressing the ISI caused by multipath in complex underwater acoustic channels.

[0041] Equal-gain combining scheme:

[0042] When combining the outputs of forward and reverse equalizations, an equal-gain combining strategy is adopted, simplifying the combining process while ensuring good performance. This makes the system more efficient and easier to implement.

[0043] In summary, the innovation of the present invention lies in the combination of bidirectional equalization, TR combining processing, and adaptive DFE based on soft information, as well as the soft information exchange mechanism based on Turbo equalization, effectively improving the transmission performance of the FMT underwater acoustic communication system. Description of the Drawings

[0044] Figure 1 、The principle block diagram of the FMT underwater acoustic communication based on direct adaptive bidirectional Turbo equalization in the embodiment of the present invention;

[0045] Figure 2 、The principle block diagram of FMT modulation and demodulation in the embodiment of the present invention;

[0046] Figure 3 、The principle block diagram of direct adaptive bidirectional Turbo equalization in the embodiment of the present invention;

[0047] Figure 4 、The structure of bidirectional equalization in the embodiment of the present invention;

[0048] Figure 5 、The measured underwater acoustic channel response for simulation verification in the embodiment of the present invention;

[0049] Figure 6 、The principle block diagram of the FMT underwater acoustic communication based on the traditional bidirectional DFE (Bidirection DFE, BiDFE) without coding in the embodiment of the present invention;

[0050] Figure 7 、The principle block diagram of the traditional uncoded BiDFE in the embodiment of the present invention;

[0051] Figure 8 , The principle block diagram of the equalization process in the traditional unidirectional Turbo DFE in the embodiment of the present invention;

[0052] Figure 9 , The performance comparison diagram between the method proposed in the embodiment of the present invention and the uncoded traditional BiDFE method;

[0053] Figure 10 , The performance comparison diagram after equalization between the method proposed in the embodiment of the present invention and the traditional unidirectional Turbo DFE method;

[0054] Figure 11 , The performance comparison diagram after decoding between the method proposed in the embodiment of the present invention and the traditional unidirectional Turbo DFE method. Specific embodiments

[0055] The following describes the specific embodiments of the present invention in conjunction with the embodiments:

[0056] It should be noted that the structures, ratios, sizes, etc. shown in this specification are only used to cooperate with the content disclosed in the specification for those familiar with this technology to understand and read, and are not used to limit the limiting conditions for the implementation of the present invention. Any modification of the structure, change of the proportional relationship, or adjustment of the size, without affecting the effects that the present invention can produce and the purposes that can be achieved, should still fall within the scope covered by the technical content disclosed in the present invention.

[0057] At the same time, the terms such as "upper", "lower", "left", "right", "middle", and "one" cited in this specification are only for the convenience of clear narration, and are not used to limit the scope of implementation of the present invention. The change or adjustment of their relative relationships, without substantial change in the technical content, should also be regarded as the scope within which the present invention can be implemented.

[0058] Embodiment 1:

[0059] 1. Structure of the communication system

[0060] Figure 1 The principle block diagram of the FMT underwater acoustic communication based on direct adaptive bidirectional Turbo equalization is given. At the transmitter of the communication system, the information bit stream is first serially-parallel converted into M parallel sub-bit streams. Each sub-bit stream is respectively encoded, interleaved, and mapped. All symbol sub-sequences output by the mapping are transmitted into the underwater acoustic channel after FMT modulation. The transmitted signal can be expressed as:

[0061]

[0062] In the above formula, s i (nT) represents the symbol transmitted by the i-th sub-sequence at the n-th moment, T is the symbol interval, and T c = T / K representsFigure 2 (a) shows the time interval after K-fold upsampling in FMT modulation, g t (kT c ) represents the discrete-time domain response of the transmit filter in FMT modulation.

[0063] After transmission through the underwater acoustic channel, the signal received by the β-th array element can be expressed as:

[0064]

[0065] In the above formula, η β (kT c ) represents the noise received by the β-th array element, h β (kT c ) represents the discrete-time response between the transmit array element and the β-th receive array element.

[0066] At the receiving end, first, the received signal is demodulated by FMT. The principle block diagram of FMT demodulation is as shown in Figure 2 (b), and the demodulated signal can be expressed as:

[0067]

[0068] In the above formula, g r (nT c ) represents the discrete-time domain response of the receive filter in FMT demodulation. The receive filter is matched to the transmit filter, that is, their frequency-domain responses satisfy

[0069] The sub-bands of FMT are relatively wide and the passbands do not overlap, and it is insensitive to frequency offset. Therefore, when both the transmitter and the receiver are fixedly deployed, the impact of inter-carrier interference on the communication performance can be ignored. Ignoring the inter-carrier interference, Equation (3) can be expressed as:

[0070]

[0071] In the above formula represents the discrete-time response of the m-th sub-channel between the transmit array element and the β-th receive array element.

[0072] All the output signals after FMT demodulation are re-divided into M groups of signals according to the sub-band numbers and sent to the direct adaptive bidirectional Turbo equalizer for interference suppression. After the information bit sub-stream output by the Turbo equalizer undergoes parallel-to-serial conversion, the decision value of the original information bit stream can be obtained.

[0073] 2. Structure of direct adaptive bidirectional turbo equalization

[0074] Figure 3 shows the m-th group of signals The process of direct adaptive two-way Turbo equalization, where direct adaptive two-way Turbo equalization consists of two-way equalization, interleaving, de-interleaving, and decoding. As Figure 3 shown, first, two-way equalization is used to suppress ISI and the corresponding soft bit information Γ out (d m (nT b )) is output. Subsequently, the output soft information Γ out (d m (nT b )) is de-interleaved, and the de-interleaved output soft information after subtracting the soft information Λ out (c m (nT b )) output by the decoder last time is used as the prior information for this decoding. The soft information output by the decoder is interleaved and then transmitted to the two-way equalizer as the prior information for the next equalization. The information interaction between two-way equalization and decoding is carried out continuously and iteratively until the performance of the decoding output reaches stability.

[0075] Figure 4 The structure diagram of two-way equalization is given. It can be seen from the figure that two-way equalization includes two parallel equalization structures: one is the forward equalization, including the TR combiner for preprocessing and the soft information-based adaptive DFE for postprocessing; the other is the backward equalization, which has the same structure as the forward equalization. It should be noted that Figure 4 the time reversal in

[0076] 3. TR combiner

[0077] Combined with Figure 4 , the output signal of the forward equalization TR combiner can be expressed as:

[0078]

[0079] q (m,m) (lT,nT) and ξ m (nT) respectively represent the combined channel response and noise after TR combining processing, and their expressions are:

[0080]

[0081] where represents the discrete-time response of the TR filter matched to the m-th sub-band channel, and its corresponding frequency-domain response is

[0082] The TR combiner in the reverse equalization is used to process the time-reversed version of the FMT demodulation signal, and its output signal is:

[0083]

[0084] and respectively represent the combined channel response and noise after the TR combining process in the reverse equalization, and their expressions are:

[0085]

[0086] Since the transmit filter matches the receive filter and the m-th TR filter matches the m-th sub-band channel, the combined channel response q (m,m) (lT,nT) and are theoretically the same, and their corresponding frequency-domain forms satisfy

[0087] Substitute with q (m,m) (lT,nT), and Equation (7) can be further expressed as:

[0088]

[0089] 4. Adaptive DFE Based on Soft Information

[0090] For the sake of easy expression, the symbol interval T and the bit interval T b will be omitted in the following analysis. Figure 4 It can be seen that the adaptive DFE based on soft information consists of two parts: signal filtering represented by solid lines and adaptive filter coefficient adjustment represented by dashed lines. The output signal of the adaptive DFE based on soft information in the forward equalization can be expressed as:

[0091]

[0092] where, f m (n) represents the filter coefficient vector, which is composed of the forward filter coefficient vector and the feedback filter coefficient vector ; [·] H represents the conjugate transpose operation; u m (n) represents the signal vector, which is composed of the output signal vector r m (n) of the TR combiner and the decision vector output by the feedback filter. The signal vector r m (n) == [r m (n),..., r m (n-(N f -1))]T where N f represents the length of the forward filter, where N b represents the length of the feedback filter, [·] T represents the transpose operation. In the first iteration, the signal is equal to the hard decision value of the filtered output signal . In subsequent iterative processing, where p in (s m (n)=s) represents the input prior information when s m (n)=s, represents the symbol set consisting of all possible values of s.

[0093] The RELS algorithm is used to adjust the coefficients of the adaptive DFE based on soft information. The processing procedure of the algorithm is as follows:

[0094] (1) Calculate the Kalman gain,

[0095]

[0096] where J m (n - 1) represents the inverse covariance matrix, and λ represents the forgetting factor.

[0097] (2) Update the coefficient vector

[0098]

[0099] where represents the decision value used for filtering processing. When the equalizer processes the training mode, is equal to the known training sequence. When the equalizer processes the decision mode and ignores the information about s provided by the first n - 1 estimated values m (n), it can be calculated according to the following formula,

[0100]

[0101] where represents the probability distribution of m with s (n)=s as the condition, and this value depends on the filter coefficient vector f m (n).

[0102] (3) Update the inverse covariance matrix

[0103]

[0104] After decision, the signal Is mapped to the output probability of the symbol according to the following formula:

[0105]

[0106] In the reverse equalization, the adaptive DFE based on soft information utilizes the time reversal of the prior probability to process the signal And its structure is the same as that of the adaptive DFE based on soft information in the forward equalization. Therefore, the processing procedure for the reverse equalization will not be elaborated here.

[0107] 5. Diversity combining

[0108] Since the forward equalization and the reverse equalization can suppress the causal ISI and the non-causal ISI respectively, the diversity gain can be obtained by combining the outputs of the forward equalization and the reverse equalization, and the choice of the combining scheme must comprehensively consider the performance after combination and the combination complexity. The log-likelihood value of the combined output is:

[0109] Γ out (d m (n)) = λ n Γ nor (d m (n)) + λ b Γ bw (d m (n))(16)

[0110] where λ n and λ b represent the combining coefficients of the forward and the reverse respectively.

[0111] Since the forward equalization and the reverse equalization have the same structure, and the combined channel response is theoretically symmetric, the proposed method selects the equal-gain combining scheme, that is, both λ n and λ b are set to 0.5.

[0112] Embodiment 2:

[0113] Combined with Embodiment 1, this embodiment presents a specific computational example of underwater acoustic communication. The underwater acoustic communication method provided by the present invention has been verified by simulation using a group of measured underwater acoustic channels. A specific computational example is given below to illustrate the effectiveness of the present invention.

[0114] Figure 5The measured underwater acoustic channel response used in the simulation is given. This set of channels was obtained in an FMT underwater acoustic communication experiment. The experimental pool is 45 meters long, 6 meters wide, and 5 meters deep, surrounded by sound-absorbing materials. Only the bottom and the surface of the pool can reflect signals. The transmitting element is placed 1.5 m below the water surface, and two receiving elements are placed 0.7 m and 0.9 m below the water surface respectively. The transmitted signal consists of a linear frequency modulation signal from 8 - 16 kHz with Hamming window, a 100 ms guard time interval, and an information signal. The communication frequency band used for the information signal is 8 - 16 kHz, which is divided into 4 FMT sub-bands. The roll-off factors of the transmitting and receiving filters are set to 0.5, the number of transmitted symbols is 1600, and BPSK mapping is used, where the first 250 symbols are used for channel estimation.

[0115] To evaluate the performance of the proposed method, the performance of the proposed method is also compared with that of two other methods used in FMT underwater acoustic communication. The first one is the traditional uncoded BiDFE method, and the second one is the traditional one-way Turbo DFE method. The principle block diagram of FMT underwater acoustic communication based on the traditional BiDFE is as Figure 6 shown, where the structure of BiDFE is as Figure 7 shown. The principle block diagram of FMT underwater acoustic communication based on the traditional one-way Turbo DFE is similar to the principle block diagram of the proposed method as Figure 1 shown, except for the structure of Turbo equalization, so it will not be repeated here. Figure 8 The principle block diagram of the equalization process in the traditional one-way Turbo DFE is given.

[0116] Table 1 shows the simulation parameter settings of the proposed method and the two comparison methods. It should be emphasized that to ensure the correctness of the simulation analysis, the settings of the communication frequency band, the number of sub-bands, and the roll-off factor in Table 1 are the same as those in the experiment for obtaining the underwater acoustic channel response used in the simulation.

[0117] Table 1 - Simulation Parameter Settings of the Proposed Method and Two Comparison Methods

[0118]

[0119]

[0120] Taking the bit error rate (BER) as the performance index, the performance of the proposed method is compared with that of the traditional uncoded BiDFE method under different signal-to-noise ratio (SNR) conditions. The results are as Figure 9As shown. It should be noted that, in order to ensure that the two methods have the same bit rate, in the simulation analysis, the proposed method uses QPSK mapping for the binary bit stream after cyclic convolutional coding with a code rate of 1 / 2, while the uncoded traditional BiDFE method uses BPSK mapping for the binary bit stream. From Figure 9 it can be seen that: First, due to the adoption of ECC, after the first iteration, the performance of the proposed method is significantly better than that of the uncoded traditional BiDFE method. Second, except for the second sub-band which requires 3 iterations to stabilize the performance, the performance of other sub-bands tends to be stable after 2 iterations.

[0121] Figure 10 and Figure 11 give the performance comparison between the proposed method and the traditional one-way Turbo DFE. From Figure 10 and Figure 11 the following conclusions can be drawn. First, since the backward equalization suppresses the non-causal ISI, after the first iteration, the BER of the proposed method after equalization and decoding processing is significantly lower than that of the traditional one-way Turbo DFE method. Second, the performance of both methods tends to be stable after 3 iterations, but the BER of the proposed method is significantly lower. Third, after 3 iterations, when the BER of the equalized outputs of the two methods is the same, the SNR corresponding to the proposed method is lower. Taking the equalization result of the first sub-band shown in Figure 10 (a) as an example, when the BER is 1%, the SNR corresponding to the proposed method is about 14.2 dB, while the SNR of the traditional one-way Turbo DFE is close to 15.7 dB, and the proposed method saves about 1.5 dB of power. It should be noted that in Figure 10 (a), when the SNR is low and the BER is the same, the power saved by the proposed method compared to the traditional Turbo DFE is small. For example, when the SNR is 8 dB, the BER of the proposed method is about 4%, while in the case of the same bit error rate, the SNR of the traditional one-way Turbo DFE is about 9 dB, and at this time the proposed method only saves about 1 dB. This is because when the SNR is low, both noise and ISI are the main factors affecting the communication performance, while when the SNR is high, the influence of noise is small and ISI is the main factor affecting the communication performance. The proposed method suppresses the causal and non-causal ISI through two-way equalization and obtains the diversity gain. Its improvement in the communication performance is mainly reflected in the suppression of ISI, so the performance improvement is more significant in the case of high SNR.

[0122] The above has made a detailed description of the preferred embodiments of the present invention, but the present invention is not limited to the above embodiments. Within the knowledge scope of those of ordinary skill in the art, various changes can be made without departing from the purpose of the present invention.

[0123] Many other changes and modifications can be made without departing from the spirit and scope of the present invention. It should be understood that the present invention is not limited to the specific embodiments, and the scope of the present invention is defined by the appended claims.

Claims

1. A direct adaptive two-way Turbo equalization method for FMT underwater acoustic communication, characterized in that, The method includes: S1: At the transmitting end of FMT underwater acoustic communication, the original binary data stream is input, and the error control coding ECC is preset. After the encoded binary bit stream undergoes interleaving, mapping, and FMT modulation, it is sent into the underwater acoustic channel through the transmitting array element. S2: At the receiving end of FMT underwater acoustic communication, the received signal is input. The forward equalization and the backward equalization with the same structure are combined to construct a two-way equalization. Both the forward and backward equalizations use a multi-channel TR combiner for preprocessing and a soft information-based direct adaptive decision feedback equalization DFE for postprocessing. The bit probabilities output by the combined two-way equalization are transmitted to the decoder, and the information exchange between the two-way equalization and the decoder is repeated until the performance is stable. Finally, the original binary bit stream recovered by the decoder is output.

2. The direct adaptive two-way Turbo equalization method for FMT underwater acoustic communication according to claim 1, characterized in that In the receiving end and decoder of FMT underwater acoustic communication, the iterative information exchange between the two-way equalization and the decoder specifically includes: S2101: In the forward equalization, first, the sub-band signals after FMT demodulation are preprocessed by the TR combiner to achieve pre-compression of the ISI. Subsequently, a soft information-based direct adaptive DFE is used for postprocessing, and the symbol probabilities output by the processing are input to the demapper. S202: In the backward equalization, first, the sub-band signals after FMT demodulation are stored using a buffer. The time reversal process is performed in the manner that the signal that enters the buffer first is output last, and the signal that enters the buffer last is output first. Subsequently, the time-reversed signal is preprocessed by the TR combiner, and a soft information-based direct adaptive DFE is used for postprocessing. S203: The symbol probabilities output in the backward equalization are stored using a buffer. The time reversal process is performed in the manner that the signal that enters the buffer first is output last, and the signal that enters the buffer last is output first. The symbol probabilities after the time reversal process are input to the demapper. S204: The two-way symbol probabilities are demapped into two-way binary bit probabilities and input to the combiner for combination. The bit probabilities output by the combination are de-interleaved. The bit probabilities output by the de-interleaving, after subtracting the bit probabilities of the previous decoding output, are used as the prior information for the current decoding. The bit probabilities of the current decoding output, after being interleaved, are input to the two-way equalization for the next equalization process. The information interaction between the two-way equalization and the decoding is sequential and iterative, and stops when the system performance reaches stability. Finally, the original binary bit stream recovered by the decoder is output.

3. A direct adaptive two-way Turbo equalization method for FMT underwater acoustic communication according to claim 2, characterized in that The specific steps of S201 include: In the forward equalization, the signal output by the TR combiner is: In the above formula, m is the sub - band serial number, q (m,m) (nT, lT) represents the m - th sub - band sequence s at the transmitting end m (lT) and the combined channel response between the output signal r of the m - th TR combiner in the forward structure at the receiving end m (nT), s m (nT) represents the value of s m (lT) when l = n, and T represents the symbol interval; The soft information-based direct adaptive DFE is used for postprocessing, and the equalization output estimate value is: In the above formula, f m (n) represents the filter coefficient vector, represents the forward filter coefficient vector, represents the feedback filter coefficient vector, [·] Η represents the conjugate transpose operation, and the filter coefficient vector is adaptively adjusted using the recursive extended least squares (RELS) method. r m (n) represents the input signal vector of the equalizer, represents the vector formed by the decision values of the feedback filter input; Make a decision on the signal to obtain the output probability of the symbol: In the above formula, p in [s m (n) = s] represents that when the transmitted symbol s m (n) = s is the input prior probability, represents that with s m (n) = s as the condition of the probability distribution, and this value depends on the filter coefficient vector f m (n), all possible values of s form the symbol set 4. A direct adaptive two-way Turbo equalization method for FMT underwater acoustic communication according to claim 2, characterized in that The specific steps of S202 include: The TR combiner in the backward equalization is used to process the time reversal version of the FMT demodulated signal, and the output signal is: In the above formula and respectively represent the combined channel response and noise after TR combination in the reverse equalization. Since the transmit filter matches the receive filter, and the m-th TR filter matches the m-th sub-band channel, the combined channel response of the reverse equalization is equal to the combined channel response q (m,m) (lT,nT); In the reverse equalization, an adaptive DFE based on soft information is used to post-process the signal The adjustment process of the equalization coefficient is the same as that of the forward equalization described in step S201.

5. A direct adaptive two-way Turbo equalization method for FMT underwater acoustic communication according to claim 2, characterized in that In step S204, the gain combination is specifically: Γ out (d m (n)) = λ n Γ nor (d m (n)) + λ b Γ bw (d m (n)) In the above formula, λ n and λ b respectively represent the combining coefficients of the forward and reverse equalizations, and their value ranges from 0 to 1. Γ nor (d m (n)) and Γ bw (d m (n)) represent the log-likelihood values of the binary information bits obtained after mapping the output probabilities of the forward and reverse equalizations.

Images