CN116155663A - Design method of underwater acoustic communication decision feedback iterative equalization receiver based on symbol-by-symbol posterior information soft decision - Google Patents
Design method of underwater acoustic communication decision feedback iterative equalization receiver based on symbol-by-symbol posterior information soft decision Download PDFInfo
- Publication number
- CN116155663A CN116155663A CN202211202518.4A CN202211202518A CN116155663A CN 116155663 A CN116155663 A CN 116155663A CN 202211202518 A CN202211202518 A CN 202211202518A CN 116155663 A CN116155663 A CN 116155663A
- Authority
- CN
- China
- Prior art keywords
- symbol
- equalization
- equalizer
- training
- iteration
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 47
- 238000004891 communication Methods 0.000 title claims abstract description 25
- 238000013461 design Methods 0.000 title claims abstract description 9
- 238000012549 training Methods 0.000 claims abstract description 47
- 230000008569 process Effects 0.000 claims abstract description 19
- 230000003044 adaptive effect Effects 0.000 claims abstract description 17
- 239000011159 matrix material Substances 0.000 claims description 21
- 238000012545 processing Methods 0.000 claims description 14
- 238000001914 filtration Methods 0.000 claims description 10
- 230000005540 biological transmission Effects 0.000 claims description 9
- 238000013507 mapping Methods 0.000 claims description 9
- 125000004122 cyclic group Chemical group 0.000 claims description 6
- 108010076504 Protein Sorting Signals Proteins 0.000 claims description 3
- 238000012804 iterative process Methods 0.000 claims description 2
- 238000004364 calculation method Methods 0.000 description 6
- 238000003672 processing method Methods 0.000 description 5
- 238000010586 diagram Methods 0.000 description 4
- 230000000694 effects Effects 0.000 description 3
- 235000019800 disodium phosphate Nutrition 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 238000010606 normalization Methods 0.000 description 2
- 238000012795 verification Methods 0.000 description 2
- 102100026758 Serine/threonine-protein kinase 16 Human genes 0.000 description 1
- 101710184778 Serine/threonine-protein kinase 16 Proteins 0.000 description 1
- 230000002411 adverse Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 230000007123 defense Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000009977 dual effect Effects 0.000 description 1
- 238000005562 fading Methods 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 230000010287 polarization Effects 0.000 description 1
Images
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L25/00—Baseband systems
- H04L25/02—Details ; arrangements for supplying electrical power along data transmission lines
- H04L25/03—Shaping networks in transmitter or receiver, e.g. adaptive shaping networks
- H04L25/03006—Arrangements for removing intersymbol interference
- H04L25/03165—Arrangements for removing intersymbol interference using neural networks
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04B—TRANSMISSION
- H04B13/00—Transmission systems characterised by the medium used for transmission, not provided for in groups H04B3/00 - H04B11/00
- H04B13/02—Transmission systems in which the medium consists of the earth or a large mass of water thereon, e.g. earth telegraphy
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L25/00—Baseband systems
- H04L25/02—Details ; arrangements for supplying electrical power along data transmission lines
- H04L25/03—Shaping networks in transmitter or receiver, e.g. adaptive shaping networks
- H04L25/03006—Arrangements for removing intersymbol interference
- H04L25/03012—Arrangements for removing intersymbol interference operating in the time domain
- H04L25/03019—Arrangements for removing intersymbol interference operating in the time domain adaptive, i.e. capable of adjustment during data reception
- H04L25/03057—Arrangements for removing intersymbol interference operating in the time domain adaptive, i.e. capable of adjustment during data reception with a recursive structure
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L25/00—Baseband systems
- H04L25/02—Details ; arrangements for supplying electrical power along data transmission lines
- H04L25/03—Shaping networks in transmitter or receiver, e.g. adaptive shaping networks
- H04L25/03006—Arrangements for removing intersymbol interference
- H04L2025/03592—Adaptation methods
- H04L2025/03598—Algorithms
- H04L2025/03611—Iterative algorithms
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
- Y02D30/00—Reducing energy consumption in communication networks
- Y02D30/70—Reducing energy consumption in communication networks in wireless communication networks
Landscapes
- Engineering & Computer Science (AREA)
- Computer Networks & Wireless Communication (AREA)
- Signal Processing (AREA)
- Power Engineering (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Computation (AREA)
- Filters That Use Time-Delay Elements (AREA)
Abstract
The invention discloses a design method of an underwater acoustic communication decision feedback iterative equalization receiver based on symbol-by-symbol posterior information soft decision, which comprises the steps of constructing the iterative equalization receiver based on decision feedback, which consists of a feedforward filter, a feedback filter, an interleaver, a phase-locked loop, a decoder and the like; training a feedforward filter, a phase-locked loop, covariance and the like by using training data based on the selected adaptive algorithm when performing first iterative equalization; and then carrying out symbol judgment on the data sequence based on the equalizer obtained by training, calculating the mean value and the variance of the equalized data sequence at the previous symbol time to obtain equalizer external information of the current symbol, further obtaining the posterior average symbol of the current equalized symbol, and synchronously updating the mean value and the variance of the equalized data sequence. The invention can accurately acquire the variation of the mean value and the variance of the equalization symbol in the equalization process, thereby improving the accuracy of decoding and the reliability of a receiver.
Description
Technical field:
the invention belongs to the field of underwater acoustic communication, and particularly relates to a design method of an underwater acoustic communication decision feedback iterative equalization receiver based on symbol-by-symbol posterior information soft decision.
The background technology is as follows:
in underwater wireless remote information transmission, underwater acoustic communication is one of the effective means at present, and is widely applied to the fields of hydrologic observation, warning monitoring, regional defense and the like. In underwater acoustic communications, different modulation methods are typically employed according to different data rate application requirements. In an application scenario of real-time high-speed transmission of underwater acoustic data, an underwater acoustic signal modulation mode based on single carrier phase modulation or orthogonal frequency division multiplexing modulation is generally adopted. The sending peak of the orthogonal frequency division multiplexing modulation is higher, and is easily influenced by the fading of the channel frequency domain, so that the orthogonal frequency division multiplexing modulation is mostly used for occasions with fixed receiving and sending and stable channels; single carrier phase modulation can be used in channel time-varying applications, but this results in a channel delay spread typically of tens to hundreds of symbols, which presents challenges for the equalization design of the receiver.
At present, a common receiving end equalization processing mode in single carrier underwater acoustic communication mainly comprises frequency domain linear equalization, block equalization based on Linear Minimum Mean Square Error (LMMSE) channel estimation, adaptive equalization based on decision feedback and the like, and the method has different algorithm performances, implementation complexity, system overhead and the like and needs to be selected according to actual application scenes. In applications such as real-time high-speed transmission of underwater acoustic data, trade-offs need to be made in terms of decoding effect, processing delay, computational complexity, overhead, and the like. Meanwhile, with the development of underwater acoustic signal processing, an iterative processing method is introduced in the equalization, and the confidence of the output soft information is improved by introducing the iterative interaction of the soft information between the equalizer and the decoder, so that the error code is further reduced in the iterative processing process.
In three typical underwater acoustic communication receiving end processing methods, such as frequency domain linear equalization, block equalization based on LMMSE channel estimation, decision feedback equalization and the like, a sub-optimal calculation method is adopted in the decision feedback equalization method, and symbol judgment is carried out by using self-adaptive algorithms, such as Recursive Least Square (RLS), least Mean Square (LMS), variable step length normalization mean square error (IPNLMS) and the like, and compared with the block equalization based on LMMSE channel estimation, the performance of steady-state minimum mean square error is lost. However, when an iterative equalization processing means is adopted, after a plurality of iterative equalization processes are performed, there is no obvious performance difference in terms of steady-state minimum mean square error between the iterative equalization based on decision feedback and the block iterative equalization based on LMMSE channel estimation. On the other hand, the receiver processing method based on decision feedback equalization adopts a self-adaptive updating means, so that complex computation such as matrix inversion, matrix division and the like, which are involved in channel inversion, can be avoided, and the method is convenient to realize on low-power consumption embedded platforms such as DSPs and the like. Therefore, in the processing of the receiving end in the underwater acoustic communication real-time data application scene, a better balance between the system overhead of the receiving end and the decoding accuracy can be obtained by adopting an iterative equalization processing method based on decision feedback.
In the iterative equalization processing method based on decision feedback, various factors such as adaptive equalization criteria and algorithms, feedforward feedback filter coefficient updating methods, adaptive filter coefficient selection, mean value and variance of equalization symbols and the like need to be carried out.
In summary, in the face of application scenarios of real-time data transmission in underwater acoustic communication, iterative equalization based on decision feedback is utilized to process on real-time computing platforms such as low-power consumption DSPs, and key parameter acquisition processes such as mean value and variance of equalization symbols playing an important role on processing results are optimized, so that comprehensive performance in aspects such as decoding effect, processing delay, calculation complexity and system overhead is optimal, and the method becomes an important problem to be solved urgently by technicians in the field of underwater acoustic communication networks.
The invention comprises the following steps:
the invention aims to solve the technical problems of providing a design method of an underwater acoustic communication decision feedback iteration equalization receiver based on symbol-by-symbol posterior information soft decision, so as to solve the technical problems of processing flow of a receiving end of iterative equalization based on decision feedback, acquiring key parameters and the like in real-time data processing in underwater acoustic communication.
The technical solution of the invention is to provide a design method of an underwater acoustic communication decision feedback iterative equalization receiver based on symbol-by-symbol posterior information soft decision, which comprises,
constructing an iterative equalization receiver based on decision feedback, which is composed of a feedforward filter, a feedback filter, an interleaver, a deinterleaver, a phase-locked loop, a mapper, a demapper, a decoder and the like;
training a feedforward filter, a phase-locked loop, covariance and the like by using training data based on the selected adaptive algorithm when performing first iterative equalization;
then, based on the equalizer obtained by training, symbol judgment is carried out on the data sequence, the average value and the variance of the equalized data sequence at the previous symbol moment are utilized to calculate out equalizer external information of the current symbol, the posterior average symbol of the current equalized symbol is further obtained, and the average value and the variance of the equalized data sequence are synchronously updated;
and when the subsequent iteration equalization is carried out, the decoder external information of the previous iteration is used as prior information to be input into an equalizer, a training data multiplexing method is adopted to train the adaptive filtering related parameters obtained by the previous equalization again, after the training is finished, symbol judgment is carried out on the data sequence, the average value and the variance of the equalized data sequence at the previous symbol moment are utilized to calculate the equalizer external information of the current symbol, meanwhile, the equalizer external information of the current symbol is obtained by adding the equalizer prior information of the current symbol obtained by the previous iteration, the posterior average symbol is obtained by utilizing the posterior information of the current symbol, and the average value and the variance of the equalized data sequence are synchronously updated.
Preferably, the method comprises the following steps,
and 3, constructing an iterative equalization receiver based on decision feedback and composed of a feedforward filter, a feedback filter, a phase-locked loop, an interleaver, a de-interleaver, a mapper, a de-mapper, a decoder and the like, and finishing definition of each parameter of the receiver, wherein the method mainly comprises the steps of feedforward filter length, feedback filter length, a filter coefficient updating algorithm, an adaptive algorithm, a covariance matrix, phase-locked loop coefficients and the like, and simultaneously defining the maximum iterative equalization times.
And 4, performing first iterative equalization, and executing the following operations: firstly, training a feedforward filter, a feedback filter, a phase-locked loop, a covariance and the like by using training data based on a selected self-adaptive algorithm, and obtaining the mean value and the variance of a training sequence after training; then, based on a feedforward and feedback filter, a covariance matrix and a phase-locked loop of self-adaptive filtering obtained by training, carrying out symbol judgment on the data sequence, calculating the mean value and variance of the equalized data sequence at the previous symbol moment to obtain equalizer external information of the current symbol, further obtaining a posterior average symbol, and synchronously updating the mean value and variance of the equalized data sequence; after the first equalization is completed, inputting the obtained equalizer external information into a decoder to obtain output bits and performing CRC (cyclic redundancy check), outputting a result and stopping the iteration process if the output bits pass the CRC, and continuing the next iteration if the output bits do not pass the CRC.
Preferably, when the next iteration is continued, the decoder external information of the previous iteration is used as prior information to be input into the equalizer, and the following operations are executed: training data multiplexing is adopted, and adaptive filtering related parameters obtained in the previous equalization are trained again to correct adverse effects of burst noise and the like; after training, carrying out symbol judgment on the data sequence by utilizing a feedforward and feedback filter and a covariance matrix obtained by multiplexing training data, calculating the mean value and the variance of the equalized data sequence at the previous symbol moment to obtain equalizer external information of the current symbol, adding the equalizer external information with equalizer prior information of the current symbol obtained by the previous iteration to obtain posterior information of the current equalized symbol, obtaining posterior average symbols by utilizing posterior information of the current symbol, and synchronously updating the mean value and the variance of the equalized data sequence; after the equalization is completed, inputting the obtained equalizer external information into a decoder to obtain output bits, performing CRC (cyclic redundancy check), stopping the iterative process and outputting a result when the check passes or the iterative number reaches the maximum iterative number, and continuing the next iteration if the check does not pass.
Preferably, in step 3, the following definition is performed,
a) Defining the data length as L data Wherein the front L train Each is a training symbol;
b) Defining the received signal sequence as y and equalizer output sequence x dfe The soft decision sequence is x det Hard decision sequence x det ,x det Front L of (1) train Are known training symbol data train Equalizer output sequence mean value x ave Variance x var Equalizer a posteriori average symbol x bar ;
c) Defining the feedforward vector of the RLS adaptive algorithm as a and the length as L a The feedback vector is b, the length is L b ;
d) Defining covariance matrix P as dimension (L a +L b )×(L a +L b ) A unit array;
e) Definition of phase-locked loop coefficientsAnd θ, and phase-locked loop update coefficient kf 1 ,kf 2 ;
f) Defining lambda as the step length of the self-adaptive algorithm;
g) Defining equalizer soft information LLR equExit Decoder extrinsic information LLR decoExit ;
h) Defining iteration times n=1, 2,3 … and N, wherein the iteration times respectively correspond to first iteration equalization, second iteration equalization, … … and Nth iteration equalization;
i) The maximum number of iterative equalisation times is defined as ITmax.
Compared with the existing underwater acoustic communication equalization receiver technology after the scheme is adopted, the invention has the following advantages:
(1) Compared with the channel equalization mode based on the LMMSE, the self-adaptive equalization mode based on the decision feedback adopts a sub-optimal algorithm based on self-adaptation, reduces the system overhead and the calculation complexity, but has a performance difference with the channel equalization mode based on the LMMSE on the steady-state average square error; by introducing iterative equalization, after a plurality of iterative equalization, the equalization mode based on decision feedback has no obvious performance disadvantage on steady-state average square error. Therefore, iterative equalization based on decision feedback can achieve a better balance between the receiving end system overhead and decoding accuracy.
(2) In the equalization process, the invention adopts the symbol-by-symbol posterior information calculation method, and compared with the symbol-by-data block posterior information calculation method, the invention can more accurately track the time-varying changes of important parameters such as symbol mean value, variance and the like in the data sequence, thereby being beneficial to improving the accuracy of equalization processing of a receiving end.
(3) In the iteration process, the invention adopts a training strategy of training data multiplexing, and can correct the feedforward and feedback filter coefficients, covariance matrix, phase-locked loop coefficients and the like obtained in the previous iteration process, thereby further reducing the steady-state average square error in the equalization process and improving the reliability of the receiver.
Description of the drawings:
fig. 1 is a schematic diagram of an iterative equalization receiver based on decision feedback;
fig. 2 is a schematic diagram of the output of the iterative equalization equalizer and the output result of the decoding soft decision.
The specific embodiment is as follows:
the invention is further described in terms of specific embodiments in conjunction with the following drawings:
the embodiment describes the working mode of the underwater sound communication decision feedback iterative equalization receiver design method based on the symbol-by-symbol posterior information soft decision through specific steps. And will be described by taking QPSK modulation and RLS adaptive algorithm as examples.
Step 1: defining constellation points theta and q of QPSK modulation as mapping dimensions, and adopting Gray mapping, wherein the QPSK modulation constellation mapping table is shown in the table 1:
constellation pointsCorresponding bit (0, 0), constellation point +.>Corresponding bits (0, 1), constellation pointsCorresponding bit (1, 1), constellation point +.>Corresponding bit (1, 0), wherein +.>Is an energy normalization factor. After the bit-symbol mapping of the transmission symbol is completed, dual polarization mapping is performed on the transmission bit, namely, x= -2c+1, and the (0, 1) bit stream is mapped into a (-1, 1) data stream, so that the calculation of subsequent soft information is facilitated. After finishing data modulation, adding pilot frequency to form a baseband data signal, then forming a passband transmitting signal after interpolation, filtering and carrier loading, and transmitting a transmitting end signal.
Step 2: at a receiving end, demodulating, filtering, downsampling and synchronizing signals of the passband receiving signals to obtain baseband data signals, and entering a subsequent processing step;
step 3: constructing an iterative equalization receiver based on decision feedback, which is composed of a feedforward filter, a feedback filter, a phase-locked loop, an interleaver, a de-interleaver, a mapper, a de-mapper, a decoder and the like, as shown in fig. 1, and performing the following definition:
i) Defining the data length as L data Wherein the front L train Each is a training symbol;
j) Defining the received signal sequence as y and equalizer output sequence x dfe The soft decision sequence is x det Hard decision sequence x det ,x det Front L of (1) train Are known training symbol data train Equalizer output sequence mean value x ave Variance x var Equalizer a posteriori average symbol x bar ;
k) Defining the feedforward vector of the RLS adaptive algorithm as a and the length as L a The feedback vector is b, the length is L b ;
L) defining covariance matrix P as dimension (L) a +L b )×(L a +L b ) A unit array;
m) definition of phase-locked loop coefficientsAnd θ, and phase-locked loop update coefficient kf 1 ,kf 2 ;
n) defining lambda as the step length of the adaptive algorithm;
o) definition of equalizer soft information LLR equExit Decoder extrinsic information LLR decoExit ;
p) defining iteration times n=1, 2,3 … and N, wherein the iteration times respectively correspond to the first iteration balance, the second iteration balance, … … and the nth iteration balance;
q) defining the maximum iterative equilibrium number as ITmax;
step 4: an iterative equalization process is started. When the first iteration is performed (n=1), without prior information, the following steps are performed:
step 4.1: initializing a feedforward filter a as an all-zero vector, a feedback filter b as an all-zero vector and an initialization covariance matrix P as a unit matrix; initialization ofθ is zero;
step 4.2: for received symbols k=1 to k=l data -L b Training is carried out, and the following steps are mainly executed:
step 4.2.1: intercepting a feedforward signal y from a received signal ff (k)=y(k+L b :k+L a +L b ) Feedback signal y fb (k)=x det (k:k+L b ) The aggregate signal synthesized to obtain equalizer feed-forward and feedback inputs is u (k) = [ y ] ff (k)·e -jθ(k) ;y fb (k)];
Step 4.2.2: calculating the feedforward part output p (k) =a' (n; k) ·y of the equalizer ff (k)·e -jθ(k) The feedback part outputs q (k) =b' (n; k) ·y fb (k) And get an estimate x of the current symbol dfe (k)=p(k)+q(k);
Step 4.2.3: according to the current training symbol x det (k)=data train (k) Obtaining the current error x err (k)=x train (k)-x dfe (k);
Step 4.2.4: calculating coefficient matrix K (n; K) =p (n; K) ·u (K)/(λ+u' (K) ·p (n; K) ·u (K)), updating feedforward filter coefficients and feedback filter coefficients a (n; k+1) =a (n; K) ·k·conj (x) err (k)),b(n;k+1)=b(n;k)·K·conj(x err (k) And a covariance matrix P (n+1); k) =p (n; k) λ -k·u' ·p (n; k) Lambda; step 4.2.5: updating phase-locked loop coefficientsCumulative phase
Step 4.2.6: calculating the mean and variance from the 1 st symbol to the current training symbol in the 1 st iteration
Step 4.2.7: when training is completed, i.e. k=l train Recording feedforward filter coefficient a after 1 st iteration training out(n=1) =a (n; k), feedback filter coefficientsb out(n=1) =b (n; k), covariance matrix P out(n=1) =p (n; k), symbol meanSymbol variance->
Step 4.3: for the received symbol k=l data -L b +1 to k=l data Making a decision, mainly executing the following steps:
step 4.3.1: cutting-in a feedforward signal y from a received signal ff (k)=y(k+L b :k+L a +L b ) Feedback signal y fb (k)=x det (k:k+L b ) The aggregate signal synthesized to obtain equalizer feed-forward and feedback inputs is u (k) = [ y ] ff (k)·e -jθ(k) ;y fb (k)];
Step 4.3.2: calculating the feedforward part output p (k) =a' (n; k) ·y of the equalizer ff (k)·e -jθ(k) ,q(k)=b′(n)·y fb (k) And get an estimate x of the current symbol dfe (k)=p(k)+q(k);
Step 4.3.3: using symbol mean x of k-1 symbol instants ave (k-1) and symbol variance x var (k-1) calculating out equalizer outer information of the current symbol time:
wherein,,represents constellation point theta corresponding to the q-th bit of 1,/and the like>Representing constellation point θ corresponding to the q-th bit of-1, because QPSK mapping is employed, two soft information LLRs are generated per symbol equExit(n=1) (k, 1) and LLR equExit(n=1) (k,2);
Step 4.3.4: calculating a posterior average symbol of the current symbol:
step 4.3.5: let x bar(n=1) (k)=x det (k) And obtain soft decision error x err (k)=x det (k)-x dfe (k) Updating feedforward and feedback filter coefficients, covariance matrix, phase-locked loop coefficients with reference to steps 4.2.4 through 4.2.6;
Step 4.3.7: when the first iterative equalization is completed, i.e. k=l data Outputting equalizer external information LLR of the current iteration equExit(n=1) ;
Step 4.4: the decoder uses the equalizer external information LLR of this iteration equExit(n=1) The prior information serving as the decoder is input into the decoder and decoded to obtain the external decoder information LLR of the current balance decoExit(n=1) Adding the decoder priori information to the decoder extrinsic information to obtain decoder posterior information LLR decoPost(n=1) And for decoder posterior information LLR decoPost(n=1) Making hard decision to obtain output bits; meanwhile, the decoder external information LLR obtained by the iteration decoExit(n=1) As the priori information of the equalizer in the next iteration, determining whether to input the equalizer again for the next iteration according to the verification result;
step 4.5: performing CRC (cyclic redundancy check) on hard decision output bits of the decoder, outputting a result and stopping the iteration process if the hard decision output bits pass the CRC, and continuing the next iteration if the hard decision output bits do not pass the CRC;
step 5: when the n=m (1 < m < n) th iteration is performed, the following steps are performed when a priori information is present:
step 5.1: based on the idea of multiplexing training data, the method is restartedTraining is performed. Let the initial value of the feedforward filter a be the output a of the n-1 th iteration out(n-1) The initial value of the feedback filter b is the output b of the n-1 th iteration out(n-1) The initial value of the covariance matrix P is the output P of the n-1 th iteration out(n-1) ;
Step 5.2: executing the training process according to the step 4.2, and obtaining the training output a of the mth iterative equalization when the training is finished out(n=m) =a(m;k),b out(n=m) =b (m; k) and P out(n=m) =p (m; k), symbol meanSymbol variance->Wherein a is out(n=m) ,b out(n=m) P out(n=m) Can be used as the input parameters of the filter coefficient and covariance matrix in the next iterative equalization training multiplexing according to the requirement>And +.>The method is mainly used for the judgment stage of the iterative equalization;
step 5.3: for the received symbol k=l data -L b +1 to k=l data Making a decision, mainly executing the following steps:
step 5.3.1: intercepting a feedforward signal y from a received signal ff (k)=y(k+L b :k+L a +L b ) Feedback signal y fb (k)=x det (k:k+L b ) The aggregate signal synthesized to obtain equalizer feed-forward and feedback inputs is u (k) = [ y ] ff (k)·e -jθ(k) ;y fb (k)];
Step 5.3.2: calculating the feedforward part output p (k) =a' (m; k) ·y of the equalizer ff (k)·e -jθ(k) ,q(k)=b′(m;k)·y fb (k) And get an estimate x of the current symbol dfe (k)=p(k)+q(k);
Step 5.3.3: using the mean value x of k-1 symbol instants ave (k-1) and symbol variance x var (k-1) calculating out equalizer outer information of the current symbol time:
wherein,,represents constellation point theta corresponding to the q-th bit of 1,/and the like>Representing the constellation point θ corresponding to the q-th bit of-1, because QPSK mapping is adopted, each symbol generates 2-bit soft information LLR equExit(n=m) (k, 1) and LLR equExit(n=m) (k,2);
Step 5.3.4: decoder external information LLR of kth symbol obtained by equalizing nth=m-1 times decoExit(n=m-1) (k, 1) and LLR decoExit(n=m-1) (k, 2) equalizer a priori information LLR as the present equalization kth symbol equPri(n=m-1) (k, 1) and LLR equPri(n=m-1) (k, 2) and adding the extrinsic information to the prior information to obtain equalizer posterior information LLR for the kth symbol at the nth=m iterations equPost(n=m-1) (k, 1) and LLR equPost(n=m-1) (k,2);
Step 5.3.5: calculating a posterior average symbol from the obtained equalizer posterior information:
step 5.3.6: and obtain soft decision error x err (k)=x det (k)-x dfe (k) Updating the feedforward and feedback filter coefficients, covariance matrix, phase-locked loop coefficients, and equalizer mean and variance with reference to steps 4.2.4 through 4.2.6;
step 5.3.7: calculating the mean value from the 1 st symbol to the current equalizing symbolVariance of
Step 5.3.8: when the mth iteration is completed, i.e. k=l data Outputting equalizer external information LLR of the current iteration equExit(n=m) ;
Step 5.4: the decoder uses the equalizer external information LLR of this iteration equExit(n=m) The prior information serving as the decoder is input into the decoder and decoded to obtain the external decoder information LLR of the current balance decoExit(n=m) Adding the decoder priori information to the decoder extrinsic information to obtain decoder posterior information LLR decoPost(n=m) And for decoder posterior information LLR decoPost(n=m) Making hard decision to obtain output bits; meanwhile, the external information LLR of the decoder obtained by the iteration decoExit(n=m) As the priori information of the equalizer in the next iteration, determining whether to input the equalizer again for the next iteration according to the verification result;
step 5.5: performing CRC (cyclic redundancy check) on hard decision output bits of the decoder, outputting a result and stopping the iteration process if the hard decision output bits pass the CRC, and continuing the next iteration if the hard decision output bits do not pass the CRC;
step 6: if the decoding is correct in the n=m time, or the iteration number n=itmax is the maximum iteration number, stopping the iteration process and outputting the decoding result. The result of the equalizer output and the decoded soft decision output of the iterative equalization is shown in fig. 2. In the equalization process, after 5 iterations, the decoder outputs no error code, and the decoding process is completed.
The it=0 decoding constellation diagram and the decoding soft decision soft information output in fig. 2 show that the error is 223 bits before decoding and 68 bits after decoding;
as can be seen from the it=1 decoding constellation and the decoding soft-decision soft information output of fig. 2, the decisions before decoding are 145 bits, and the decisions after decoding are 26 bits;
as can be seen from the it=2 decoding constellation and the decoding soft decision soft information output of fig. 2, the error is 80 bits before decoding and 6 bits after decoding;
as can be seen from the it=3 decoding constellation diagram and the decoding soft decision soft information output of fig. 2, the error is 16 bits before decoding, and the error is 0 bits after decoding;
as can be seen from the it=4 decoding constellation and the decoding soft-decision soft information output of fig. 2, the pre-decoding decision is 0 bit, and the post-decoding decision is 0 bit.
Therefore, the invention calculates the soft decision posterior average symbol by using the self-adaptive algorithm in a symbol-by-symbol mode, and can realize the accurate tracking of the mean and variance change of the equalization symbol in the iterative equalization process, thereby improving the decoding accuracy and the reliability of the receiver.
The foregoing is illustrative of the preferred embodiments of the present invention, and is not to be construed as limiting the claims. All equivalent structures or equivalent flow path changes made by the specification of the invention are included in the protection scope of the invention.
Claims (6)
1. A design method of an underwater acoustic communication decision feedback iterative equalization receiver based on symbol-by-symbol posterior information soft decision is characterized by comprising the following steps of: the method includes the steps of,
constructing an iterative equalization receiver based on decision feedback;
training the feedforward filter, the phase-locked loop and the covariance by using training data based on the selected adaptive algorithm when performing first iterative equalization;
performing symbol judgment on the data sequence based on the equalizer obtained by training, calculating the mean value and the variance of the equalized data sequence at the previous symbol time to obtain equalizer external information of the current symbol, further obtaining a posterior average symbol of the current equalized symbol, and synchronously updating the mean value and the variance of the equalized data sequence;
and when the subsequent iteration equalization is carried out, the decoder external information of the previous iteration is used as prior information to be input into an equalizer, a training data multiplexing method is adopted to train the adaptive filtering related parameters obtained by the previous equalization again, after the training is finished, symbol judgment is carried out on the data sequence, the average value and the variance of the equalized data sequence at the previous symbol moment are utilized to calculate the equalizer external information of the current symbol, meanwhile, the equalizer external information of the current symbol is obtained by adding the equalizer prior information of the current symbol obtained by the previous iteration, the posterior average symbol is obtained by utilizing the posterior information of the current symbol, and the average value and the variance of the equalized data sequence are synchronously updated.
2. The method for designing the underwater acoustic communication decision feedback iterative equalization receiver based on the soft decision of the symbol-by-symbol posterior information according to claim 1, wherein the method comprises the following steps: the iterative equalization receiver based on decision feedback is composed of a feedforward filter, a feedback filter, an interleaver, a de-interleaver, a phase-locked loop, a mapper, a de-mapper and a decoder.
3. The method for designing the underwater acoustic communication decision feedback iterative equalization receiver based on the soft decision of the symbol-by-symbol posterior information according to claim 1, wherein the method comprises the following steps: comprises the steps of,
step 1, defining the mapping relation of constellation points, bits and symbols aiming at single-carrier underwater acoustic communication phase modulation, adding pilot frequency to form a baseband data signal after finishing data modulation, forming a passband transmission signal after interpolation, filtering and carrier loading, and transmitting a signal at a transmitting end;
step 2, the receiving end demodulates, filters, downsamples and synchronizes the passband received signal to obtain a baseband data signal, and starts equalization processing;
and step 3, constructing an iterative equalization receiver based on decision feedback, finishing definition of each parameter of the receiver, and defining the maximum iterative equalization times.
And 4, performing first iterative equalization, and executing the following operations: firstly, training a feedforward filter, a feedback filter, a phase-locked loop and covariance by using training data based on a selected self-adaptive algorithm, and obtaining the mean value and variance of a training sequence after training; then, based on a feedforward and feedback filter, a covariance matrix and a phase-locked loop of self-adaptive filtering obtained by training, carrying out symbol judgment on a data sequence, calculating the mean value and variance of the equalized data sequence at the previous symbol moment to obtain equalizer external information of the current symbol, further obtaining a posterior average symbol, and synchronously updating the mean value and variance of the equalized data sequence; after the first equalization is completed, inputting the obtained equalizer external information into a decoder to obtain output bits and performing CRC (cyclic redundancy check), outputting a result and stopping the iteration process if the output bits pass the CRC, and continuing the next iteration if the output bits do not pass the CRC.
4. The method for designing an underwater acoustic communication decision feedback iterative equalization receiver based on soft decision of symbol-by-symbol posterior information according to claim 3, wherein: if the next iteration is needed, the decoder external information of the previous iteration is used as prior information to be input into the equalizer, and the following operations are executed: firstly, training data multiplexing is adopted, and the adaptive filtering related parameters obtained in the previous equalization are trained again; after training, carrying out symbol judgment on the data sequence by utilizing a feedforward and feedback filter and a covariance matrix obtained by multiplexing training data, calculating the mean value and the variance of the equalized data sequence at the previous symbol moment to obtain equalizer external information of the current symbol, adding the equalizer external information with equalizer prior information of the current symbol obtained by the previous iteration to obtain posterior information of the current equalized symbol, obtaining posterior average symbols by utilizing posterior information of the current symbol, and synchronously updating the mean value and the variance of the equalized data sequence; after the equalization is completed, inputting the obtained equalizer external information into a decoder to obtain output bits, performing CRC (cyclic redundancy check), stopping the iterative process and outputting a result when the check passes or the iterative number reaches the maximum iterative number, and continuing the next iteration if the check does not pass.
5. The method for designing an underwater acoustic communication decision feedback iterative equalization receiver based on soft decision of symbol-by-symbol posterior information according to claim 3, wherein: in step 3, the parameter definition of the receiver includes feedforward filter length, feedback filter length, filter coefficient updating algorithm, adaptive algorithm, covariance matrix, and phase-locked loop coefficient.
6. The method for designing an underwater acoustic communication decision feedback iterative equalization receiver based on soft decision of symbol-by-symbol posterior information according to claim 3, wherein: in step 3, the following definition is performed,
a) Defining the data length as L data Wherein the front L train Each is a training symbol;
b) Defining the received signal sequence as y and equalizer output sequence x dfe The soft decision sequence is x det Hard decision sequence x det ,x det Front L of (1) train Are known training symbol data train Equalizer output sequence mean value x ave Variance x var Equalizer a posteriori average symbol x bar ;
c) Defining the feedforward vector of the RLS adaptive algorithm as a and the length as L a The feedback vector is b, the length is L b ;
d) Defining covariance matrix P as dimension (L a +L b )×(L a +L b ) A unit array;
e) Definition of phase-locked loop coefficientsAnd θ, and phase-locked loop update coefficient kf 1 ,kf 2 ;
f) Defining lambda as the step length of the self-adaptive algorithm;
g) Defining equalizer soft information LLR equExit Decoder extrinsic information LLR decoExit ;
h) Defining iteration times n=1, 2,3 … and N, wherein the iteration times respectively correspond to first iteration equalization, second iteration equalization, … … and Nth iteration equalization;
i) The maximum number of iterative equalisation times is defined as ITmax.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202211202518.4A CN116155663A (en) | 2022-09-29 | 2022-09-29 | Design method of underwater acoustic communication decision feedback iterative equalization receiver based on symbol-by-symbol posterior information soft decision |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202211202518.4A CN116155663A (en) | 2022-09-29 | 2022-09-29 | Design method of underwater acoustic communication decision feedback iterative equalization receiver based on symbol-by-symbol posterior information soft decision |
Publications (1)
Publication Number | Publication Date |
---|---|
CN116155663A true CN116155663A (en) | 2023-05-23 |
Family
ID=86351301
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202211202518.4A Pending CN116155663A (en) | 2022-09-29 | 2022-09-29 | Design method of underwater acoustic communication decision feedback iterative equalization receiver based on symbol-by-symbol posterior information soft decision |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN116155663A (en) |
-
2022
- 2022-09-29 CN CN202211202518.4A patent/CN116155663A/en active Pending
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN113242189B (en) | Adaptive equalization soft information iteration receiving method combined with channel estimation | |
Sandell et al. | Iterative channel estimation using soft decision feedback | |
Otnes et al. | Iterative channel estimation for turbo equalization of time-varying frequency-selective channels | |
US6819630B1 (en) | Iterative decision feedback adaptive equalizer | |
CN109981501B (en) | Underwater sound direct self-adaptive MIMO communication method | |
Sozer et al. | Iterative equalization and decoding techniques for shallow water acoustic channels | |
CN105553903B (en) | A kind of adaptive turbo equalization methods and balanced device, underwater sound communication system | |
Xu et al. | Spatial and time-reversal diversity aided least-symbol-error-rate turbo receiver for underwater acoustic communications | |
Xi et al. | Soft direct-adaptation based bidirectional turbo equalization for MIMO underwater acoustic communications | |
CN112039809B (en) | Block iterative equalizer based on mixed soft information and bidirectional block iterative equalizer | |
Hu et al. | Optimal channel shortener design for reduced-state soft-output Viterbi equalizer in single-carrier systems | |
CN115225437B (en) | Combined intelligent equalization and decoding method for underwater acoustic cooperative communication | |
CN116155663A (en) | Design method of underwater acoustic communication decision feedback iterative equalization receiver based on symbol-by-symbol posterior information soft decision | |
CN111682924B (en) | Bidirectional frequency domain Turbo equalization method adopting expected propagation | |
WO2021243607A1 (en) | Equalisation method and apparatus | |
Ye et al. | Adaptive Step-size Iterative Equalization for Underwater Communication | |
Yang et al. | Hybrid Turbo equalization based on Kalman filter for underwater acoustic communications | |
Kim et al. | An efficient blind Pseudo turbo equalizer with CMA and SAGMCMA for single-carrier system | |
Oyerinde et al. | Improved soft iterative channel estimation for turbo equalization of time varying frequency selective channels | |
CN111211870B (en) | Iteration termination decision method in turbo receiving system | |
Wan et al. | Joint channel estimation and Markov Chain Monte Carlo detection for frequency-selective channels | |
Duan et al. | Experimental evaluation of turbo receivers in single-input single-output underwater acoustic channels | |
Movahedian et al. | Low complexity estimation of fast fading radio channels for higher order modulation | |
Singer et al. | Linear iterative turbo-equalization (LITE) for dual channels | |
Xiong et al. | Low-Complexity Turbo Equalization for MIMO Underwater Acoustic Communications |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination |