CN109428673A - For the method for decoded signal, equipment and storage equipment - Google Patents

Abstract

The embodiment of the present disclosure provides a kind of method for decoded signal.This method comprises: being decoded using predetermined decoding scheme to signal is received, and the estimation to signal is sent is obtained based on decoding result；The estimation to interchannel noise is obtained using the correlation of interchannel noise；The estimation obtained to interchannel noise is subtracted from the reception signal, to obtain the reception signal of modification；And be decoded using reception signal of the predetermined decoding scheme to the modification, to obtain decoded signal.The embodiment of the present disclosure additionally provides the equipment and storage equipment for decoded signal.

Description

Method, device and storage device for decoding a signal

technical field

The present disclosure relates to signal decoding, and in particular, to methods, devices, and storage devices for decoding signals.

Background technique

Significant progress has been made in the field of channel coding and decoding since Shannon founded information theory. With proper coding design and efficient belief propagation (BP) algorithm, low-density parity-check (LDPC, low-density parity-check) codes can achieve performance close to the Shannon limit. However, traditional codec designs are mainly designed for simple channels such as Gaussian white noise channels. However, this codec design does not take into account the complex channel situation caused by factors such as filtering, oversampling, and multi-user interference in the actual channel (this noise is called colored noise).

Therefore, there is a need for a decoding scheme that can handle such colored noise.

SUMMARY OF THE INVENTION

An aspect of an embodiment of the present disclosure provides a method for decoding a signal. The method includes: decoding a received signal using a predetermined decoding scheme, and obtaining an estimate of the transmitted signal based on the decoding result; obtaining an estimate of the channel noise by utilizing a correlation of channel noise; subtracting the obtained signal from the received signal an estimate of channel noise to obtain a modified received signal; and decoding the modified received signal using the predetermined decoding scheme to obtain a decoded signal.

Optionally, utilizing the correlation of the channel noise to obtain the estimate of the channel noise may comprise: obtaining the first estimate of the channel noise by subtracting the estimate of the transmitted signal from the received signal; and utilizing the channel noise The obtained first estimate of channel noise is processed to obtain a second estimate of channel noise as the estimate of channel noise.

Optionally, using the correlation of the channel noise to process the obtained first estimate of the channel noise to obtain the second estimate of the channel noise may include: using the first estimate of the channel noise as a convolutional neural network the input of the network, and the output of the convolutional neural network as the second estimate of the channel noise.

Optionally, the method may further include training the convolutional neural network using a predetermined strategy, wherein the predetermined strategy includes any of the following: a training strategy that minimizes headroom noise power, and a training strategy that minimizes headroom noise power while making the distribution of the margin noise close to the Gaussian distribution of the training strategy.

Optionally, the method may further comprise using the modified received signal as the received signal to iteratively perform decoding of the received signal, estimation of the channel noise based on the correlation of the channel noise, and obtaining of the modified received signal, until the system state meets the Booking request.

Another aspect of an embodiment of the present disclosure provides an apparatus for decoding a signal. The apparatus includes a decoder, a transmit signal estimator, a noise estimator, and a modified signal generator. The decoder is used to decode the received signal using a predetermined decoding scheme. A transmit signal estimator is used to obtain an estimate of the transmit signal based on the decoding result. The noise estimator is used to obtain an estimate of the channel noise by exploiting the correlation of the channel noise. A modified signal generator for subtracting the obtained estimate of channel noise from the received signal to obtain a modified received signal. The decoder also decodes the modified received signal using the predetermined decoding scheme to obtain a decoded signal.

Optionally, the noise estimator may include a first noise estimation module and a second noise estimation module. A first noise estimation module is operable to obtain a first estimate of channel noise by subtracting the estimate of the transmitted signal from the received signal. The second noise estimation module may be configured to process the obtained first estimate of channel noise using the correlation of channel noise to obtain a second estimate of channel noise as the estimate of channel noise.

Optionally, the second noise estimation module is implemented using a convolutional neural network. In this case, the second noise estimation module may be further configured to: take the first estimation of the channel noise as the input of the convolutional neural network, and take the output of the convolutional neural network as the pair of channels A second estimate of noise.

Optionally, the apparatus may further include a network trainer for training the convolutional neural network using a predetermined strategy, wherein the predetermined strategy includes any of the following: minimizing a margin noise power A training strategy, and a training strategy to make the distribution of the margin noise close to a Gaussian distribution while minimizing the margin noise power.

Optionally, in the device, the operations of the decoder, the transmit signal estimator, the noise estimator and the modified signal generator are iteratively performed with the modified received signal as the received signal, until the system state meets the predetermined requirements.

Another aspect of an embodiment of the present disclosure provides an apparatus for decoding a signal. The device includes a memory and a processor. Memory is used to store executable instructions. The processor is configured to execute executable instructions stored in the memory to perform the above method.

Another aspect of an embodiment of the present disclosure provides a memory device having a computer program carried thereon, which when executed by a processor causes the processor to perform the above-described method.

Description of drawings

For a more complete understanding of the present invention and its advantages, reference will now be made to the following description taken in conjunction with the accompanying drawings, in which:

FIG. 1 schematically shows a brief flow chart of a method for decoding a signal according to an embodiment of the present disclosure;

FIG. 2 schematically shows a brief block diagram of an apparatus for decoding a signal according to an embodiment of the present disclosure;

FIG. 3 shows a brief schematic diagram of a decoding scheme according to an embodiment of the present disclosure;

A schematic diagram of a CNN structure used according to an embodiment of the present disclosure is shown in FIG. 4;

Figures 5 to 8 show schematic diagrams of performance comparisons for two different channel correlation models, respectively; and

FIG. 9 schematically shows a brief block diagram of an electronic device according to an embodiment of the present disclosure.

Detailed ways

Hereinafter, embodiments of the present disclosure will be described with reference to the accompanying drawings. It should be understood, however, that these descriptions are exemplary only, and are not intended to limit the scope of the present disclosure. Also, in the following description, descriptions of well-known structures and techniques are omitted to avoid unnecessarily obscuring the concepts of the present disclosure.

The terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the present disclosure. As used herein, the words "a," "an," and "the" and the like shall also include the meanings of "plurality," "plurality," unless the context clearly dictates otherwise. Furthermore, the terms "comprising", "comprising" and the like used herein indicate the presence of stated features, steps, operations and/or components, but do not preclude the presence or addition of one or more other features, steps, operations or components .

All terms (including technical and scientific terms) used herein have the meaning as commonly understood by one of ordinary skill in the art, unless otherwise defined. It should be noted that terms used herein should be construed to have meanings consistent with the context of the present specification and should not be construed in an idealized or overly rigid manner.

Some block diagrams and/or flow diagrams are shown in the figures. It will be understood that some of the blocks in the block diagrams and/or flowcharts, or combinations thereof, can be implemented by computer program instructions. The computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, or other programmable data processing apparatus, such that the instructions, when executed by the processor, may be created to implement the functions illustrated in the block diagrams and/or flow diagrams /Operating the device.

Accordingly, the techniques of this disclosure may be implemented in hardware and/or software (including firmware, microcode, etc.). Additionally, the techniques of the present disclosure may take the form of a computer program product on a computer-readable medium having stored instructions for use by or in conjunction with an instruction execution system. In the context of this disclosure, a computer-readable medium can be any medium that can contain, store, communicate, propagate, or transmit instructions. For example, a computer-readable medium may include, but is not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, device, or propagation medium. Specific examples of computer-readable media include: magnetic storage devices, such as magnetic tapes or hard disks (HDDs); optical storage devices, such as compact disks (CD-ROMs); memories, such as random access memory (RAM) or flash memory; and/or wired /Wireless communication link.

FIG. 1 schematically shows a brief flowchart of a method for decoding a signal according to an embodiment of the present disclosure.

As shown in FIG. 1 , the method includes operation S110 , decoding the received signal using a predetermined decoding scheme, and obtaining an estimate of the transmitted signal based on the decoding result.

The predetermined decoding scheme referred to here can be any suitable decoding scheme, for example, it can be a belief propagation decoding (belief propagation, BP) scheme, of course, according to specific application scenarios, it can also be the same as the encoding scheme used by the sending side. corresponding to any decoding scheme.

In some examples, obtaining an estimate of the transmitted signal based on the decoding result may be to re-encode the decoding result using an encoding scheme corresponding to the decoding scheme used, using the re-encoded coded signal obtained as the estimate of the transmitted signal. Of course, any other technical solutions in the art that can be used for estimating the transmitted signal at the receiving side can also be applied here, and the embodiment of the present disclosure is not limited by the specific implementation of the estimation of the transmitted signal.

In operation S120, an estimate of the channel noise is obtained using the correlation of the channel noise.

By considering the correlation of noise in the channel caused by factors such as filtering, oversampling, multi-user interference, etc., the correlation can be used to estimate the noise in the actual channel more accurately in the embodiments of the present disclosure.

In some examples, utilizing the correlation of the channel noise to obtain the estimate of the channel noise may include: obtaining the first estimate of the channel noise by subtracting the estimate of the transmitted signal from the received signal; and utilizing the correlation of the channel noise to process the obtained first estimate of channel noise to obtain a second estimate of channel noise as the estimate of channel noise described in operation S120.

In some examples, a convolutional neural network (CNN) may be used to implement the correlation of channel noise. In this case, the first estimate of the channel noise may be used as the input to the convolutional neural network, and the output of the convolutional neural network may be used as the second estimate of the channel noise.

Before utilizing the convolutional neural network to obtain the second estimate of the channel noise, the convolutional neural network may be trained with a predetermined strategy so that the convolutional neural network can simulate the scenario of a real-world channel. In some examples, the predetermined strategy may include a training strategy that minimizes headroom noise power, such as a Baseline BP-CNN strategy. In other examples, the predetermined strategy may also include a training strategy that minimizes the power of the margin noise while making the distribution of the margin noise close to a Gaussian distribution, such as an Enhanced BP-CNN (Enhanced BP-CNN) strategy. Of course, the strategies that can be used to train the convolutional neural network in the embodiments of the present disclosure are not limited to the above two, but may include any other strategies that can enable the trained convolutional neural network to more accurately reflect the real channel.

After obtaining an estimate of the channel noise using the correlation of the channel noise in operation S120, in operation S130, the obtained estimate of the channel noise is subtracted from the received signal to obtain a modified received signal.

Since a good estimate of the channel noise is obtained using the correlation of the channel noise in operation S120, by subtracting the good estimate from the received signal, the obtained modified received signal may be better close to the transmission transmitted at the transmitting side Signal.

In some cases, performing operation S120 only once may not be able to obtain a good estimate of actual channel noise (eg, colored noise). In this case, the decoding of the received signal ( S110 ), the estimation of the channel noise based on the correlation of the channel noise ( S120 ), and the obtaining of the modified received signal ( S130 ) may be performed iteratively with the above-mentioned modified received signal as the received signal. , until the system state meets the predetermined requirements. The predetermined requirement mentioned here may be that the number of system iterations reaches a preset number or the decoded signal does not change, or it may be any other requirement commonly used by those skilled in the art for noise estimation.

Then, in operation S140, the modified received signal is decoded using the same predetermined decoding scheme as in operation S110 to obtain a decoded signal.

In the method shown in Fig. 1, by using the correlation of the channel noise to obtain a good estimate of the channel noise and subtracting the good estimate from the received signal, the estimate of the transmitted signal can be obtained more accurately, and thus the High quality decoding.

FIG. 2 schematically shows a brief block diagram of an apparatus for decoding a signal according to an embodiment of the present disclosure. The block diagram shown in FIG. 2 corresponds to the flowchart of the method shown in FIG. 1 . It should be noted that, for the sake of clarity and conciseness, the block diagram shown in FIG. 2 only shows functions/modules that are helpful for understanding the embodiments of the present disclosure. In a specific implementation, more or less functions/modules may also be included.

As shown in FIG. 2 , the apparatus includes a decoder 210 , a transmit signal estimator 220 , a noise estimator 230 and a modified signal generator 240 .

The decoder 210 is used to decode the received signal using a predetermined decoding scheme.

The predetermined decoding scheme referred to here can be any suitable decoding scheme, for example, it can be a belief propagation decoding (belief propagation, BP) scheme, of course, according to specific application scenarios, it can also be the same as the encoding scheme used by the sending side. corresponding to any decoding scheme.

The transmit signal estimator 220 is used to obtain an estimate of the transmit signal based on the decoding result.

In some examples, obtaining an estimate of the transmitted signal based on the decoding result may be to re-encode the decoding result using an encoding scheme corresponding to the decoding scheme used, using the re-encoded coded signal obtained as the estimate of the transmitted signal. In this case, the transmit signal estimator may be an encoder using a coding scheme corresponding to the used decoding scheme, eg the same encoder as the transmitting side. Of course, any other technical solutions in the art that can be used for estimating the transmitted signal at the receiving side can also be applied here, and the embodiment of the present disclosure is not limited by the specific implementation of the estimation of the transmitted signal.

The noise estimator 230 is used to obtain an estimate of the channel noise using the correlation of the channel noise.

By considering the correlation of noise in the channel caused by factors such as filtering, oversampling, multi-user interference, etc., the correlation can be used to estimate the noise in the actual channel more accurately in the embodiments of the present disclosure.

In some examples, noise estimator 230 may include a first noise estimation module 232 and a second noise estimation module 234 . The first noise estimation module 232, which may be an adder/subtractor, may be used to obtain a first estimate of channel noise by subtracting the above-mentioned estimate of the transmitted signal from the received signal. The second noise estimation module 234 may be configured to process the obtained first estimate of the channel noise by utilizing the correlation of the channel noise to obtain a second estimate of the channel noise, as the above-mentioned estimate of the channel noise.

In some examples, a convolutional neural network (CNN) may be used to implement the correlation of channel noise, that is, the second noise estimation module 234 is implemented as a convolutional neural network. In this case, the first estimate of the channel noise may be used as the input to the convolutional neural network, and the output of the convolutional neural network may be used as the second estimate of the channel noise.

In some examples, prior to utilizing the convolutional neural network to obtain a second estimate of channel noise, the convolutional neural network may be trained with a predetermined strategy to enable the convolutional neural network to simulate real-world channel scenarios. In this case, the apparatus shown in FIG. 2 may further include a network trainer 250 for training the convolutional neural network using a predetermined strategy. In some examples, the predetermined strategy may include a training strategy that minimizes headroom noise power, such as a Baseline BP-CNN strategy. In other examples, the predetermined strategy may also include a training strategy that minimizes the power of the margin noise while making the distribution of the margin noise close to a Gaussian distribution, such as an Enhanced BP-CNN (Enhanced BP-CNN) strategy. Of course, the strategies that can be used to train the convolutional neural network in the embodiments of the present disclosure are not limited to the above two, but may include any other strategies that can enable the trained convolutional neural network to more accurately reflect the real channel.

Since the noise estimator 230 obtains a good estimate of the channel noise using the correlation of the channel noise, by subtracting this good estimate from the received signal, the resulting modified received signal can be better approximated to the transmission transmitted at the transmitting side Signal.

After the noise estimator 230 uses the correlation of the channel noise to obtain an estimate of the channel noise, the modified signal generator 230 is used to subtract the obtained estimate of the channel noise from the received signal to obtain a modified received signal.

In some cases, one operation of noise estimator 230 may not be able to obtain a good estimate of actual channel noise (eg, colored noise). In this case, the operations of the decoder 210, the transmit signal estimator 220, the noise estimator 230 and the modified signal generator 240 may be iteratively performed with the above modified received signal as the received signal until the system state meets the predetermined requirements. The predetermined requirement mentioned here may be that the number of system iterations reaches a preset number or the decoded signal does not change, or it may be any other requirement commonly used by those skilled in the art for noise estimation.

When the modified received signal is obtained, the decoder 210 also decodes the modified received signal using the predetermined decoding scheme to obtain a decoded signal.

In the device shown in Fig. 2, by using the correlation of the channel noise to obtain a good estimate of the channel noise and subtracting the good estimate from the received signal, the estimate of the transmitted signal can be obtained more accurately, and thus the High quality decoding.

The technical solutions of the embodiments of the present disclosure have been described above through the method shown in FIG. 1 and the device shown in FIG. 2 . The technical solutions according to the embodiments of the present disclosure will be described in detail below through specific examples. It should be noted that the technical solutions of the embodiments of the present disclosure are not limited to the specific example, but may also include various modifications made to the example that fall within the protection scope of the present invention.

FIG. 3 shows a brief schematic diagram of a decoding scheme according to an embodiment of the present disclosure. It should be noted that FIG. 3 is only a specific example for illustrating the embodiment of the present disclosure, and should not be regarded as a limitation on the embodiment of the present disclosure. For example, FIG. 3 uses LDPC coding and Binary Phase Shift Keying (BPSK) modulation scheme on the transmitting side, and uses BP decoding scheme on the receiving side. However, those skilled in the art can understand that the technical solutions of the embodiments of the present disclosure can also be Applied to different codecs and modulation methods.

In the technical solution of FIG. 3 , it is assumed that the information bit block x at the transmitting end obtains the codeword u through the LDPC encoder, and the symbol s is obtained through BPSK modulation. After the symbol s passes through the colored noise channel, the receiving end will receive the noisy signal y=s+n, where n represents the colored channel noise. The receiving end decodes the transmitted information through an iterative structure mainly consisting of a BP decoder and a convolutional neural network (CNN). An iterative process mainly consists of two steps. In the first step, the noisy signal y first passes through a standard BP decoder, and the estimation of the transmitted symbol s can be obtained according to the decoding result, denoted as By subtracting the estimated value from the noisy signal y received by the receiver, an estimate of the channel noise can be obtained, that is, Because of the existence of possible BP decoding errors, There is an error between the true channel noise. So the noise estimate can be written as ξ represents the noise estimation error due to BP decoding errors. In the second step, the Input a deep convolutional neural network (CNN), CNN will use the correlation of the channel noise to estimate the noise again, and get a more accurate estimate of the noise, denoted as Will Subtracting from the received signal y gives in is defined as the margin noise. When the CNN is more accurate in estimating the channel noise, the power of the margin noise is lower, has a higher signal-to-noise ratio than y. Therefore, will Input to the BP decoder again, and the BP decoder will get a more accurate decoding result. The above process can be iterated to gradually compress the influence of the margin noise and improve the decoding performance.

In the example of Figure 3, the beginning and end of the above-described iterative process is shown in the form of a single-pole, double-throw switch. For example, when the noisy signal y is decoded by the standard BP decoder in the first step above, the switch can be turned on the loop formed by the BP, the CNN and the adder to start the iterative operation of estimating the noise, while the system state meets the After a predetermined requirement (eg, the number of system iterations reaches a preset number or the decoded signal does not change), the loop can be turned off and the noisy signal y and the BP decoder can be turned on to end the iterative operation. The control signal of the switch can be generated by any method currently used or to be used in the future, and the protection scope of the present invention is not limited by the specific generation method of the control signal. In addition, the start and end of the iterative loop in the embodiment of the present disclosure is not limited to the form of the SPDT switch shown in FIG. 3 , but any specific implementation form for loop operation in the art may be used.

A schematic diagram of a CNN structure used according to an embodiment of the present disclosure is shown in FIG. 4 . In the structure shown in Figure 4, the input is an N×1 vector, the noise estimate In the first layer, k ₁ feature maps are obtained through the convolution operation, and the mathematical form is Where c _{1, j} represents the j-th feature map of the first layer, h _{1, j} is a one-dimensional convolution kernel of length f ₁ , * represents the convolution operation, b _{1, j} represents the j-th feature map corresponding to Bias, ReLU (Rectified Linear Unitfunction) represents the activation function (ie max(x, 0)). In the i-th layer, the convolution operation is performed on all feature maps output by the upper layer, so it can be regarded as a two-dimensional convolution, and the mathematical form is c _{i, j} = ReLU(hi _{, j} *c _i-1 +b _{i, j} ). c _i,j is the j-th feature map of the i-th layer, h _i,j is the j-th convolution kernel of the i-th layer, the size is f _i ×ki _-1 , f _i and k _i-1 represent respectively The size of the convolution kernel of the i-th layer and the number of feature maps of the i-1th layer c _i-1 represent the two-dimensional matrix formed by all the feature maps of the i-1th layer arranged. Use L to represent the number of layers of the network. In the last layer, the Lth layer, the final output of the network is i.e. with noise estimation compared to more accurate noise estimates. The network structure shown in FIG. 4 can be abbreviated as {L; f ₁ , f ₂ , ..., f _L ; k ₁ , k ₂ , ..., k _L }.

As mentioned above, the CNN network can be trained before using it for noise estimation. The training of the network can consist of two major steps. First, training data needs to be generated for a specific channel. In this step, the source bits x can be randomly generated, and the channel noise can be collected in the actual channel or trained using a pre-known channel model. In practical applications, it is usually necessary to train for some common communication scenarios (models), and the trained network model can be saved at the receiving end to select a corresponding channel model for a specific communication scenario. In actual use, an appropriate model can be selected and used according to the estimation of the channel. In the case of channel noise data, the input data of CNN can be obtained according to the process shown in Figure 3 The CNN can be trained only once, using the same network model throughout the iterations, and of course, in some examples, it can be trained multiple times.

In order to train the network, a loss function needs to be defined. The following embodiments of the present disclosure provide two methods for defining loss functions, which correspond to different network training strategies. It should be noted that the following loss function definitions and/or network training strategies are only examples provided to illustrate the solutions of the embodiments of the present disclosure, and other loss function definitions and/or network training strategies used in the art can also be applied in the embodiments of the present disclosure.

Network training strategy 1: Baseline BP-CNN (Baseline BP-CNN), this strategy minimizes the power of the margin noise, and the loss function at this time can be defined as

where r is the residual noise vector and N is the length of this vector.

In this training strategy, the empirical distribution of the residual noise is counted after training, and this distribution is used to initialize the log-likelihood ratios (LLR, log-likelihood ratios) values of the variable nodes in the next BP decoding.

Network training strategy 2: Enhanced BP-CNN (Enhanced BP-CNN), which adjusts the distribution of the margin noise while compressing the margin noise to make it close to the Gaussian distribution. Since most encoders are currently designed for Gaussian channels, this strategy can be better adapted to encoders. At this point, the loss function is defined as

The definition of the first part is the same as that of network training strategy 1, which is used to measure the power of the margin noise. The second part is derived from the Jarque-Bera test, a commonly used normality test method, specifically,

r _i is the ith element of the residual noise vector r, to its mean. λ is a weighting factor used to adjust the weights of the two parts. At this time, it is not necessary to count the empirical distribution of the residual noise, and it is only necessary to count its variance and calculate the initial value of the LLR of the variable node of the next BP decoding according to the Gaussian distribution.

In order to further illustrate the effect of the technical solutions of the embodiments of the present disclosure, FIG. 5 to FIG. 8 respectively show schematic diagrams of performance comparison for two different channel correlation models. 5 to 6 show a schematic diagram of the performance comparison of the first channel correlation model under different degrees of correlation, and FIG. 7 shows a schematic diagram of the relationship between the performance of the first channel correlation model and the number of iterations , Figure 8 shows a schematic diagram of the performance comparison for the second channel correlation model.

The first channel correlation model:

The elements in the correlation matrix R of the model are defined as Ri _,j = η ^|ij| . The LDPC code rate is 3/4, the code length is 576, and the coding matrix comes from the WiMax standard. It should be noted that the above parameters are only examples, and the application scenarios of the technical solutions of the embodiments of the present disclosure do not depend on a specific coding rate and coding matrix. Under the above parameters, the adopted CNN network structure can be {4; 9, 3, 3, 15; 64, 32, 16, 1}. Test results for only one iteration of BP-CNN under the parameters η = 0.8 (strong correlation) and η = 0.5 (moderate correlation) representing correlation are presented in Figures 5 and 6, respectively. The decoding structure at this time can be abbreviated as BP(x)-CNN-BP(x), where the numbers in parentheses represent the number of iterations of the BP decoder. The BP(5)-CNN-BP(5) structural complexity given in Figures 5 and 6 is roughly equivalent to 12 standard BP iterations. It can be seen from the results in the figures that whether the baseline BP-CNN training strategy or By enhancing the BP-CNN training strategy, the technical solutions of the embodiments of the present disclosure can obtain better results than the standard BP algorithm. At the same time, increasing the number of iterations of standard BP will further improve the performance, but the improvement is very limited. This shows that the technical solutions of the embodiments of the present disclosure can achieve higher performance with lower complexity. Figure 7 shows the results of performing multiple BP-CNN iterations. It can be seen from the figure that multiple BP-CNN iterations can further improve the decoding performance.

The second channel correlation model:

The correlation of the channel correlation model can be described using the power spectral density, ie P(f)∝1/|f| ^α . In particular, when α=1, this noise is called pink noise. FIG. 8 shows the performance comparison between the technical solution of the embodiment of the present disclosure and the standard BP decoder under the condition that the remaining conditions are the same as the first channel correlation model. It can be seen from FIG. 8 that the technical solutions of the embodiments of the present disclosure can still achieve better decoding performance under this channel model.

Figure 9 schematically shows a block diagram of a device according to an embodiment of the present disclosure. The device shown in FIG. 9 is only an example, and should not impose any limitation on the function and scope of use of the embodiments of the present disclosure.

As shown in FIG. 9 , an apparatus 900 according to this embodiment includes a central processing unit (CPU) 901 that can be loaded into a random access memory (RAM) according to a program stored in a read only memory (ROM) 902 or from a storage section 908 The program in 903 executes various appropriate actions and processes. In the RAM 903, various programs and data necessary for the operation of the device 900 are also stored. The CPU 901 , the ROM 902 , and the RAM 903 are connected to each other through a bus 904 . An input/output (I/O) interface 905 is also connected to bus 904 .

The device 900 may also include one or more of the following components connected to the I/O interface 905: an input portion 906 including a keyboard or mouse, etc.; including a cathode ray tube (CRT) or liquid crystal display (LCD), etc., and a speaker etc.; a storage part 908 including a hard disk or the like; and a communication part 909 including a network interface card such as a LAN card or a modem. The communication section 909 performs communication processing via a network such as the Internet. A drive 910 is also connected to the I/O interface 905 as needed. A removable medium 911, such as a magnetic disk, an optical disk, a magneto-optical disk, or a semiconductor memory, etc., is mounted on the drive 910 as needed so that a computer program read therefrom is installed into the storage section 908 as needed.

In particular, according to embodiments of the present disclosure, the processes described above with reference to the flowcharts may be implemented as computer software programs. For example, embodiments of the present disclosure include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the method illustrated in the flowchart. In such an embodiment, the computer program may be downloaded and installed from the network via the communication portion 909, and/or installed from the removable medium 911. When the computer program is executed by the central processing unit (CPU) 901, the above-described functions defined in the apparatus of the embodiment of the present disclosure are executed.

It should be noted that the computer-readable medium shown in the present disclosure may be a computer-readable signal medium or a computer-readable storage medium, or any combination of the above two. The computer readable storage medium can be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared or semiconductor system, apparatus or device, or a combination of any of the above. More specific examples of computer readable storage media may include, but are not limited to, electrical connections with one or more wires, portable computer disks, hard disks, random access memory (RAM), read only memory (ROM), erasable Programmable read only memory (EPROM or flash memory), optical fiber, portable compact disk read only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the above. In this disclosure, a computer-readable storage medium may be any tangible medium that contains or stores a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. In the present disclosure, however, a computer-readable signal medium may include a data signal propagated in baseband or as part of a carrier wave, carrying computer-readable program code therein. Such propagated data signals may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the foregoing. A computer-readable signal medium can also be any computer-readable medium other than a computer-readable storage medium that can transmit, propagate, or transport the program for use by or in connection with the instruction execution system, apparatus, or device . Program code embodied on a computer readable medium may be transmitted using any suitable medium including, but not limited to, wireless, wireline, optical fiber cable, or RF, etc., or any suitable combination of the foregoing.

The flowchart and block diagrams in the Figures illustrate the architecture, functionality, and operation of possible implementations of systems, methods and computer program products according to various embodiments of the present disclosure. In this regard, each block in the flowchart or block diagrams may represent a module, segment, or portion of code, which contains one or more possible functions for implementing the specified logical function(s) Execute the instruction. It should also be noted that, in some alternative implementations, the functions noted in the blocks may occur out of the order noted in the figures. For example, two blocks shown in succession may, in fact, be executed substantially concurrently, or the blocks may sometimes be executed in the reverse order, depending upon the functionality involved. It is also noted that each block of the block diagrams or flowchart illustrations, and combinations of blocks in the block diagrams or flowchart illustrations, can be implemented in special purpose hardware-based systems that perform the specified functions or operations, or can be implemented using A combination of dedicated hardware and computer instructions is implemented.

Methods, apparatus, units, and/or modules according to various embodiments of the present disclosure may also use, for example, field programmable gate arrays (FPGAs), programmable logic arrays (PLAs), systems on chips, systems on substrates, systems on packages, Application specific integrated circuits (ASICs) may be implemented in hardware or firmware in any other reasonable manner for integrating or packaging circuits, or in a suitable combination of software, hardware and firmware implementations. The system may include a storage device to implement the storage described above. When implemented in these ways, the software, hardware and/or firmware used is programmed or designed to perform the respective above-described methods, steps and/or functions in accordance with the present disclosure. Those skilled in the art can appropriately implement one or more of these systems and modules, or some or more of them, using different above-mentioned implementation manners according to actual needs. These implementations all fall within the protection scope of the present invention.

As will be understood by those skilled in the art, for any and all purposes, such as in providing a written description, all ranges disclosed in this application also encompass any and all possible subranges and subranges thereof. combination. Any listed range can be readily identified as sufficiently descriptive and to enable the same range to be broken down at least into equivalent two-part, three-part, four-part, five-part, ten-part, etc. equivalents. As a non-limiting example, each range discussed in this application can be easily broken down into a lower third, a middle third, an upper third, and so on. As will also be understood by those skilled in the art, all language such as "until," "at least," "greater than," "less than," etc. includes the stated quantity and means that it can be subsequently broken down into as discussed above range of subranges. Finally, as will be understood by those skilled in the art, ranges include each individual ingredient. So, for example, a group with 1-3 cells refers to a group with 1, 2 or 3 cells. Similarly, a group with 1-5 cells refers to a group with 1, 2, 3, 4, or 5 cells, and so on.

While the invention has been shown and described with reference to specific exemplary embodiments of the present disclosure, those skilled in the art will appreciate that, without departing from the spirit and scope of the invention as defined by the appended claims and their equivalents, Various changes in form and detail have been made in the present disclosure. Therefore, the scope of the present invention should not be limited to the above-described embodiments, but should be determined not only by the appended claims, but also by their equivalents.

Claims (12)

1. a kind of method for decoded signal, comprising:

It is decoded using predetermined decoding scheme to signal is received, and the estimation to signal is sent is obtained based on decoding result；

The estimation to interchannel noise is obtained using the correlation of interchannel noise；

The estimation obtained to interchannel noise is subtracted from the reception signal, to obtain the reception signal of modification；And

It is decoded using reception signal of the predetermined decoding scheme to the modification, to obtain decoded signal.

2. according to the method described in claim 1, wherein, the estimation to interchannel noise is obtained using the correlation of interchannel noise Include:

The first estimation to interchannel noise is obtained by subtracting the estimation of described pair of transmission signal from the reception signal；With And

The first estimation obtained to interchannel noise is handled using the correlation of interchannel noise to obtain to channel Second estimation of noise, using as the estimation to interchannel noise.

3. according to the method described in claim 2, wherein, using the correlation of interchannel noise come to obtained to interchannel noise The first estimation handle obtain estimate the second of interchannel noise include:

By the input to the first estimation of interchannel noise as convolutional neural networks, with the output of the convolutional neural networks As second estimation to interchannel noise.

4. according to the method described in claim 3, further include:

The convolutional neural networks are trained using predetermined policy, wherein the predetermined policy includes any one of following: being minimized The Training strategy of surplus noise power, and make the distribution of surplus noise close to Gauss while minimizing surplus noise power The Training strategy of distribution.

5. according to the method described in claim 1, further include:

The reception signal of the modification is carried out into iteration as reception signal and executes the decoding of reception signal, based on the phase of interchannel noise Acquisition of the closing property to the reception signal of estimation and the modification of interchannel noise, until system mode meets pre-provisioning request.

6. a kind of equipment for decoded signal, comprising:

Decoder, for being decoded using predetermined decoding scheme to signal is received；

Signal estimator is sent, for obtaining the estimation to signal is sent based on decoding result；

Noise estimator obtains the estimation to interchannel noise for the correlation using interchannel noise；And

Signal generator is modified, for subtracting the estimation obtained to interchannel noise from the reception signal, to be repaired The reception signal changed；

Wherein, the decoder is also decoded using reception signal of the predetermined decoding scheme to the modification, to obtain Decoded signal.

7. equipment according to claim 6, wherein the noise estimator includes:

First noise estimation module, for being obtained pair by subtracting the estimation of described pair of transmission signal from the reception signal First estimation of interchannel noise；And

Second noise estimation module, for the correlation using interchannel noise come to the first estimation obtained to interchannel noise It is handled to obtain to the second of interchannel noise the estimation, using as the estimation to interchannel noise.

8. equipment according to claim 7, wherein the second noise estimation module is realized using convolutional neural networks , the second noise estimation module is also used to:

By the input to the first estimation of interchannel noise as the convolutional neural networks, with the convolutional neural networks It exports as second estimation to interchannel noise.

9. equipment according to claim 8, further includes:

Network training device, for training the convolutional neural networks using predetermined policy, wherein the predetermined policy include with It is any one of lower: to minimize the Training strategy of surplus noise power, and so that surplus is made an uproar while minimizing surplus noise power Training strategy of the distribution of sound close to Gaussian Profile.

10. equipment according to claim 6, wherein using the reception signal of the modification as signal is received, iteratively hold The row decoder, the operation for sending signal estimator, the noise estimator and the modification signal generator, until System mode meets pre-provisioning request.

11. a kind of equipment for decoded signal, comprising:

Memory, for storing executable instruction；And

Processor, for executing the executable instruction stored in memory, to execute according to claim 1 described in any one of -5 Method.

12. one kind carries the memory devices of computer program, when executing the computer program by processor, institute thereon Stating computer program makes the processor execute method according to any one of claims 1-5.