Polar code decoding method and related equipment based on intelligent path selection

Abstract

The present disclosure provides a polar code decoding method and related equipment based on intelligent path selection, including: obtaining a polar code and determining parameters of the polar code; inputting the parameters into a pre-trained neural network model to perform path selection on the polar code to obtain a path selection result; determining the path selection result to obtain a determination result, and inputting the determination result into a decoder; and decoding the polar code by the decoder. By using a preset neural network model to perform path selection on the polar code, the present disclosure reduces the latency generated during polar code decoding and further improves the performance of the polar code decoder.

Description

Technical Field

[0001] The present invention relates to the field of communication technologies, and in particular to a polar code decoding method based on intelligent path selection and related equipment. Background Art

[0002] In communication systems, polar codes, proposed by Arikan, represent a breakthrough in channel coding. They are also a coding scheme that has been proven to achieve the Shannon channel capacity, sparking widespread interest among researchers. Furthermore, due to their low-complexity encoding and decoding scheme, polar codes have become one of the most important codes in the field.

[0003] In the prior art, when decoding polar codes, if the code length is infinite, the serial cancellation (SC) decoding algorithm can achieve the channel capacity of binary-input discrete memoryless channels (BMCs). When the code length is finite, decoding is performed using a serial cancellation list (SCL) aided by a cyclic redundancy check (CRC).

[0004] In traditional SCL / CA-SCL decoding algorithms, path metrics are typically used as a criterion for path selection for polar codes. By ranking the path metrics, L paths with the smallest metrics are selected as candidate paths.

[0005] However, in the actual decoding process, metric sorting of path metrics seriously affects the decoder throughput. Moreover, as the list size increases, the decoding delay also increases exponentially, which in turn limits the development of high-performance polar code decoders. Summary of the Invention

[0006] In view of this, the purpose of the present disclosure is to provide a polar code decoding method based on intelligent path selection and related devices.

[0007] As one aspect of the present disclosure, a polar code decoding method based on intelligent path selection is provided, characterized by comprising:

[0008] Obtaining a polarization code and determining parameters of the polarization code;

[0009] Inputting the parameters into a pre-trained neural network model to perform path selection on the polar code to obtain a path selection result;

[0010] Determining the path selection result to obtain a determination result, and inputting the determination result into a decoder;

[0011] The polar code is decoded by the decoder.

[0012] Optionally, the acquiring a polarization code and setting parameters of the polarization code includes:

[0013] determining a code length of the polar code, and calculating a codeword of the polar code using the code length;

[0014] Modulating the codeword to obtain a transmission vector of the codeword, and inputting the transmission vector into a transmission channel to obtain a reception vector;

[0015] A log-likelihood ratio of each decoded bit in each path of the polar code is calculated according to the received vector, and a path metric value of the corresponding path is calculated using the log-likelihood ratio of each decoded bit in each path.

[0016] Optionally, inputting the parameters into a preset neural network model to perform path selection on the polar code includes:

[0017] performing feature extraction on the path metric value to determine the length of the path corresponding to the path metric value;

[0018] classifying the path in response to a length of the path;

[0019] Based on the classification result, path selection is performed on the polar code.

[0020] Optionally, inputting the parameters into a preset neural network model to perform path selection on the polar code further includes:

[0021] Calculating the loss of the neural network model based on the classification result;

[0022] The neural network model is optimized using the loss.

[0023] Optionally, determining the result of the path selection and inputting the determination result to a decoder includes:

[0024] Determining the result of the path selection by a preset threshold value;

[0025] In response to determining that the result of the path selection is less than the preset threshold value, inputting the result of the path selection to the decoder;

[0026] In response to determining that the result of the path selection is greater than the preset threshold value, the result of the path selection is discarded.

[0027] Optionally, determining the result of the path selection and inputting the determination result to the decoder further includes: in response to determining that the determination result is less than the path requirement of the decoder, completing the determination result.

[0028] Optionally, the neural network model includes an input layer, an output layer and a hidden layer, wherein the number of neurons in the input layer and the output layer is the same as the scale of the decoder.

[0029] As a second aspect of the present disclosure, the present disclosure further provides a polar code decoder based on intelligent path selection, including:

[0030] Parameter setting module, which sets the parameters of the polarization code;

[0031] a path selection module, inputting the parameters into a preset neural network model to perform path selection on the polar code;

[0032] A path determination module determines the result of the path selection;

[0033] The path completion module completes the result of the determination in response to determining that the result of the determination is less than the path requirement of the decoder.

[0034] As a third aspect of the present disclosure, the present disclosure further provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. The computer program is characterized in that when the processor executes the program, it implements the polar code decoding method based on intelligent path selection provided in the present disclosure.

[0035] As a fourth aspect of the present disclosure, the present disclosure further provides a non-transitory computer-readable storage medium, wherein the non-transitory computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a computer to execute any of the methods described above.

[0036] As described above, the present disclosure provides a polar code decoding method and related equipment based on intelligent path selection. In this disclosure, polar code parameters are set to obtain path metrics for the polar code. These path metrics are then input into a preset neural network model, thereby implementing intelligent path selection for the polar code. This reduces the latency and, consequently, the time required for polar code decoding. Furthermore, the preset neural network model in this disclosure can be continuously updated and optimized, making polar code path selection even faster in practical applications using this neural network model, significantly reducing path selection time. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] In order to more clearly illustrate the technical solutions in the present disclosure or related technologies, the following briefly introduces the drawings required for use in the embodiments or related technical descriptions. Obviously, the drawings described below are only embodiments of the present disclosure. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0038] Figure 1A A schematic diagram of a polar code decoding method based on intelligent path selection provided by an embodiment of the present disclosure.

[0039] Figure 1B A schematic diagram of a method for determining polar code parameters provided by an embodiment of the present disclosure.

[0040] Figure 1C A schematic diagram of a polar code path selection method provided by an embodiment of the present disclosure.

[0041] Figure 1D A schematic diagram of a method for determining a path selection result provided by an embodiment of the present disclosure.

[0042] Figure 2 A schematic diagram of a polar code decoder based on intelligent path selection provided by an embodiment of the present disclosure.

[0043] Figure 3 A schematic diagram of the electronic device structure of a polar code decoding method based on intelligent path selection provided in an embodiment of the present disclosure. DETAILED DESCRIPTION

[0044] In order to make the objectives, technical solutions and advantages of the present disclosure more clearly understood, the present disclosure is further described in detail below in conjunction with specific embodiments and with reference to the accompanying drawings.

[0045] It should be noted that, unless otherwise defined, the technical terms or scientific terms used in the embodiments of the present disclosure should have the usual meanings understood by people with ordinary skills in the field to which the present disclosure belongs. The "first", "second" and similar words used in the embodiments of the present disclosure do not indicate any order, quantity or importance, but are only used to distinguish different components. "Include" or "comprise" and similar words mean that the elements or objects appearing before the word include the elements or objects listed after the word and their equivalents, without excluding other elements or objects. "Connect" or "connected" and similar words are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. "Up", "down", "left", "right" and the like are only used to indicate relative position relationships. When the absolute position of the described object changes, the relative position relationship may also change accordingly.

[0046] In the prior art, polar code decoding is primarily achieved using the SCL / CA-SCL decoding algorithm. When decoding using these algorithms, path metrics are sorted to determine multiple paths with smaller metrics, which serve as candidate paths for input to the decoder.

[0047] The inventors of the present disclosure have discovered that selecting candidate paths using the aforementioned metric sorting method will limit the throughput efficiency of the decoder and, when the decoding list is large, will also significantly increase the latency of the decoding process.

[0048] To address the aforementioned issues, the present disclosure provides a polar code decoding method and related equipment based on intelligent path selection. This method first sets polar code parameters and uses these parameters to obtain a path metric for the polar code. This path metric is then input into a preset neural network model, which then performs intelligent path selection. Finally, the selected path is input into a decoder for decoding.

[0049] After introducing the basic principles of the present disclosure, various non-limiting embodiments of the present disclosure are described in detail below.

[0050] Figure 1A A schematic diagram of a polar code decoding method based on intelligent path selection provided by an embodiment of the present disclosure.

[0051] Figure 1A The polar code decoding method based on intelligent path selection further includes the following steps:

[0052] Step S10: Obtain a polarization code and determine parameters of the polarization code.

[0053] In some embodiments, when decoding a received polar code, the relevant parameters of the polar code can be initialized. These parameters are then calculated to determine the path metric values ​​of each path in the polar code. Based on the path metric values, multiple paths are ultimately input to the decoder, which then decodes the input paths.

[0054] Figure 1B A schematic diagram of a method for determining polar code parameters provided by an embodiment of the present disclosure.

[0055] In some embodiments, as Figure 1B FIG. 1 is a further explanation of determining the polar code parameters in step S10, which specifically includes the following steps:

[0056] S101: Determine a code length of the polar code, and calculate a codeword of the polar code based on the code length.

[0057] In some embodiments, when we receive a polar code that needs to be decoded, we can first set its code length and calculate the codeword capacity of the polar code based on the set code length. Specifically, we can set the length of the polar code, which is a power of 2, to N, and the length of the polar code information bit can be expressed as K. Then, the code rate of this polar code can be expressed as: R = K / N. After an encoding, the polar code codeword can be expressed as: c = uF, where c = {c0, c1, ..., c N-1}, F is the polarization kernel matrix, represents the n-th Kronecker product, u={u0,u1,...,u N-1} is the original vector.

[0058] S102: Modulate the codeword to obtain a transmission vector of the codeword, and input the transmission vector into a transmission channel to obtain a reception vector.

[0059] In some embodiments, after obtaining the polar code codeword, we can also modulate it using binary phase-shift keying (BPSK). This modulation process uses a reference sine wave and a phase-reversed wave, shifting one side to 0 and the other to 1. This allows the modulated polar code to simultaneously transmit and receive two-value (1-bit) information, converting the polar code analog signal into an accurate data value.

[0060] In some embodiments, when the aforementioned codeword c is modulated by binary phase shift keying (BPSK), a transmission vector can also be obtained. The transmission vector can be expressed as: x=1 N -2c, where x={x0,x1,...,x N-1}, 1 N Represents a vector of all 1s of length N.

[0061] In some embodiments, after the transmit vector is obtained through the above process, the transmit vector may be subjected to noise processing. For example, the transmit vector may be transmitted through a channel of white Gaussian noise (AWGN). After the transmit vector is transmitted through a channel of white Gaussian noise (AWGN), a corresponding receive vector may be received at the receiving end corresponding to the transmitting end. The receive vector may be expressed as: y = x + z, where y = {y0, y1, ..., y N-1}, z={z0,z1,...,z N-1} represents an AWGN noise vector. It is understandable that other noise adding methods may be used when adding noise to the transmission vector. In this embodiment, only Gaussian white noise (AWGN) is used as an example for illustration.

[0062] S103: Calculate a log-likelihood ratio of each decoded bit in each path of the polar code according to the received vector, and calculate a path metric of the corresponding path using the log-likelihood ratio of each decoded bit in each path.

[0063] In some embodiments, after the transmit vector is noised, a noisy receive vector can be obtained. Next, the receive vector can be calculated to obtain the log-likelihood ratio of each decoded bit in each path of the polar code.

[0064] In some embodiments, we can first calculate the log-likelihood ratio of the received vector, which can be expressed as: Where L={L0,L1,...,L N-1}, σ 2 is the variance of the AWGN noise vector, and y represents the received vector. After calculating the log-likelihood ratio, we can further use the log-likelihood ratio to obtain the path metric for each path in the polar code.

[0065] In some embodiments, before calculating the path metric, we can also determine the decoding algorithm used to decode the polar code. In this embodiment, we select the SCL / CA-SCL decoding algorithm. It is understood that other decoding algorithms can also be used to decode the polar code, such as serial cancellation decoding (SC). This disclosure only uses the SCL / CA-SCL decoding algorithm as an example for description.

[0066] In some embodiments, after determining that the decoding algorithm is SCL / CA-SCL, we can also determine that the list size of the decoding algorithm is L. At this time, based on the selected decoding algorithm and the log-likelihood ratio obtained above, we can also calculate the path metric value of each path in the polar code. The path metric value PM can be expressed as:

[0067] In some embodiments, the path metric value PM can also be expanded to obtain 2L PM metric values, and the L PM value labels with smaller metric values ​​are set as d i =0, the label with the larger metric value is set to d i =1, and form vector d in order.

[0068] As described above, in this disclosure, by determining the relevant parameters of the acquired polar code, the path metrics of each path within the polar code are calculated. Next, these path metrics are input into a preset neural network model for feature extraction, thereby identifying multiple candidate paths.

[0069] Step S20: Inputting the parameters into a pre-trained neural network model to perform path selection on the polar code to obtain a path selection result.

[0070] In some embodiments, after determining the path metric values ​​of each path in the polar code, we can input the multiple path metric values ​​as inputs into a preset neural network model for path selection.

[0071] In some embodiments, the parameters of a preset neural network model can be determined first. Generally speaking, the parameter list of the neural network model can be determined as L, which is the same size as the decoding algorithm list. This design allows the preset neural network model to be fully matched with the relevant parameters obtained under the aforementioned decoding algorithm. It is understood that the neural network model designed in this disclosure can be a universal neural network. This design also allows the neural network model used in this disclosure to theoretically adapt to all polar codes with code length N and code rate R, and can also change with the size L of the decoding algorithm list.

[0072] In some embodiments, the neural network model provided herein may further include an input layer, an output layer, and a hidden layer, wherein the number of neurons in the input layer and the output layer is the same as the size of the decoder. This design ensures that the path selection results output from the neural network model match those of the decoder, eliminating any numerical errors.

[0073] In some embodiments, the input and output layers of the neural network model preset in this embodiment should both consist of one layer, but the hidden layer can be composed of multiple layers, and this embodiment does not impose specific restrictions on this. Accordingly, the number of neurons in the input layer of the neural network model of this disclosure can be 2L, the same as the decoder list size, and the number of neurons in the output layer is also 2L, the same as the number of information bits in the polar code. The number of hidden layers and the number of neurons in each hidden layer can be determined based on specific circumstances.

[0074] In some embodiments, the neural network model based on intelligent path selection in the present disclosure can also convert the input path metric value into a discrete value in the range of (0, 1). It is understood that the neural network model based on intelligent path selection in the present disclosure can include all neural network models with the same input and output, including but not limited to fully connected neural networks.

[0075] Figure 1C A schematic diagram of a polar code path selection method provided by an embodiment of the present disclosure.

[0076] In some embodiments, as Figure 1C FIG. 2 is a further explanation of step S20, which specifically includes the following steps:

[0077] S201: Extract features of the path metric value and determine the length of the path corresponding to the path metric value.

[0078] In some embodiments, after multiple path measurement values ​​are input into the input layer of a preset neural network model, the neural network model will perform feature extraction on the multiple path measurement values ​​to determine the length of the path corresponding to the multiple path measurement values, that is, the reliability of the path. It can be understood that the smaller the value of the path reliability, the higher the reliability of the path.

[0079] In some embodiments, the preset neural network model can obtain the numerical value of the path metric and perform corresponding calculations for feature extraction via the hidden layer. The specific calculation process can be expressed as: i =f(W i-1,i [o i-1 ] T +b i ),i=1,2,...,I-1where [o i-1 ] T is the transpose of the output vector of the i-1th layer, W i-1,i is the weight matrix between the i-1th layer and the i-th layer, and its dimension is (J i ,J i-1 ), matrix W i-1,i There is J i Line J i-1 Column, the matrix is ​​the trainable weight, all initialized to 1 during training, b i is the bias vector of the i-th layer, with a length of J i , that is, b i There is J i elements, all of which are initialized to 1 during training.

[0080] In some embodiments, the preset neural network model can determine the lengths of multiple paths corresponding to the multiple path metrics based on the above-mentioned calculation and conversion process. The determined path lengths can then be used to further determine the path selection result, i.e., the candidate path.

[0081] S202: Classify the path in response to the length of the path.

[0082] In some embodiments, we determine the lengths of multiple paths corresponding to the multiple path metrics using a preset neural network model. Next, we can classify the multiple paths based on the determined lengths of the multiple paths.

[0083] In some embodiments, we can determine a median value for the path length based on the path length. Specifically, we can calculate the mean of all path lengths to obtain the mean of the path lengths, and classify all path lengths based on the mean, classifying paths with lengths less than the mean into the first category and paths with lengths greater than the mean into the second category.

[0084] S203: Perform path selection on the polar code based on the classification result.

[0085] In some embodiments, after the preset neural network model classifies the path lengths, it can also perform path selection for the polar code based on the classification results, ultimately determining multiple candidate paths that can be input to the decoder.

[0086] In some embodiments, the preset neural network model can select multiple paths with shorter path lengths as candidate paths. Specifically, the preset neural network model can discard the second classification results, retain the first classification results, and use the multiple paths from the first classification results as the final candidate paths. It will be understood that the determination of the candidate paths also has a corresponding relationship with the value of the aforementioned label d.

[0087] In some embodiments, the aforementioned preset neural network model can also perform self-optimization based on the classification results. Specifically, the preset neural network model can calculate the loss of the neural network model based on the classification results. In the present disclosure, the loss of the neural network model can use cross entropy as the loss function, which can be expressed as: Among them, d i is the label vector, is the estimate of the information vector The elements in , and

[0088] In some embodiments, the neural network model can provide self-feedback on the calculated loss and optimize itself based on the feedback results. This continuous optimization allows the preset neural network model to more accurately extract corresponding features when extracting features from the aforementioned path metrics, thereby reducing the decoding process latency while maintaining the same bit error rate, resulting in a shorter decoding process.

[0089] As described above, the present disclosure uses a preset neural network model to classify the aforementioned path metrics and ultimately determine multiple candidate paths. This eliminates the time required to sort path metrics in the prior art, significantly reducing decoding latency and increasing decoder throughput, thereby improving decoder performance.

[0090] Next, the present disclosure performs threshold determination on the multiple candidate paths finally determined to determine multiple polar code paths that can be input into a decoder, and then decodes the polar codes through the decoder.

[0091] Step S30: Determine the path selection result to obtain a determination result, and input the determination result into a decoder.

[0092] In some embodiments, after we obtain multiple candidate paths through a preset neural network model, since the numerical values ​​of the obtained candidate paths may not meet the requirements of binary discrete values ​​in the decoder, we may need to use a preset threshold value to perform threshold value judgment on the multiple candidate paths obtained.

[0093] In some embodiments, the threshold value can be determined first. In actual use, each output node of the neural network model can be connected to a threshold, which can give rise to a variety of decision methods. Specifically, the intelligent path selection threshold in the present disclosure can include a threshold value T and 2L comparators, responsible for selecting the path with higher reliability (network output greater than T). Intelligent path selection can also include multiple implementations, such as: 1. Using a fixed threshold T uniformly, simulating with different threshold values, and selecting the one with the best performance as the fixed threshold value; 2. Using a variable threshold T uniformly, dynamically adjusting the threshold value at different decoding bits so that the output greater than T is as close to L as possible. Specific adjustment methods vary, for example, a binary method can be used for dynamic adjustment, represented by: presetting a threshold T = 0.5, determining the number of outputs greater than T, and lowering the threshold if it is greater than L, and raising it if it is less than L; 3. Using different fixed thresholds for different list positions, using different threshold values, and this threshold value can also vary based on the decoding position.

[0094] Figure 1DA schematic diagram of a method for determining a path selection result provided by an embodiment of the present disclosure.

[0095] In some embodiments, as Figure 1D FIG. 3 is a further description of step S30, which specifically includes the following steps:

[0096] S301: Determine the result of the path selection by using a preset threshold value.

[0097] In some embodiments, a corresponding threshold value can be set by the aforementioned method, and when multiple candidate paths selected by a preset neural network model pass through the gate line module, they can be judged by the preset threshold value.

[0098] S302: In response to determining that the result of the path selection is less than the preset threshold value, input the result of the path selection to the decoder.

[0099] In some embodiments, when the path length of multiple candidate paths is less than a preset threshold, the candidate path can be input as the final path to the decoder for decoding. Specifically, multiple candidate paths less than the preset threshold can be assigned a value of 0 and input into the decoder for decoding.

[0100] S303: In response to determining that the result of the path selection is greater than the preset threshold, discard the result of the path selection.

[0101] In some embodiments, when the path lengths of multiple candidate paths are less than a preset threshold, candidate paths greater than the threshold may be discarded. Specifically, multiple candidate paths greater than the threshold may be assigned a value of 1, thereby discarding them. This assignment during the threshold determination process can further automate the determination process.

[0102] S40: Decoding the polar code by the decoder.

[0103] In some embodiments, after the result of the path selection is determined by the threshold value, a final determination result can be obtained. Next, the final determination result can be input into a decoder for decoding.

[0104] In some embodiments, the final determination result may be smaller than the path requirement of the decoder. In this case, the final determination result may be supplemented so that the number of paths that ultimately enter the decoder can meet the decoder throughput requirement.

[0105] In some embodiments, various methods can be used to complete the final determination result. For example, a direct discard and complete matching strategy can be employed. Specifically, if the number of paths that ultimately pass the determination is greater than the number of paths required by the decoder, the excess is discarded in sequence. If the number is less than the number of paths required by the decoder, the paths are completed in sequence until the required number of paths is reached. It is understood that the method for completing the paths that ultimately pass the determination can also employ a random discard and complete matching strategy. This embodiment only uses the direct discard and complete strategy as an example.

[0106] In summary, in the present disclosure, parameters of the acquired polar code are first determined, ultimately calculating a path metric value. Next, multiple path metric values ​​are input into a preset neural network model for feature extraction, thereby determining multiple candidate paths. These multiple candidate paths are then subjected to threshold determination to obtain a final determined path. Finally, the determined path is input into a decoder to complete decoding of the acquired polar code.

[0107] The above method eliminates the time required to sort path metric values ​​in the prior art, significantly reduces the latency of polar code decoding, and increases decoder throughput, significantly contributing to decoder improvement and innovation.

[0108] Based on the same technical concept, and corresponding to any of the aforementioned embodiments and methods, the present disclosure further provides a polar code decoder based on intelligent path selection. The polar code decoder based on intelligent path selection provided by the present disclosure can implement the polar code decoding method based on intelligent path selection described in any of the aforementioned embodiments.

[0109] Figure 2 A schematic diagram of a polar code decoder based on intelligent path selection provided by an embodiment of the present disclosure.

[0110] In some embodiments, as Figure 2 As shown, the polar code decoder based on intelligent path selection includes:

[0111] A parameter setting module 100 is used to set the parameters of the polarization code;

[0112] The path selection module 200 inputs the parameters into a preset neural network model to perform path selection for the polar code;

[0113] The path determination module 300 determines the result of the path selection;

[0114] The path completion module 400 completes the result of the determination in response to determining that the result of the determination is less than the path requirement of the decoder.

[0115] Based on the same technical concept, and corresponding to any of the aforementioned embodiments and methods, the present disclosure further provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, it implements the polar code decoding method based on intelligent path selection described in any of the aforementioned embodiments.

[0116] Figure 3 10 is a schematic diagram showing a more specific hardware structure of an electronic device provided in this embodiment. The device may include: a processor 1010, a memory 1020, an input / output interface 1030, a communication interface 1040, and a bus 1050. The processor 1010, the memory 1020, the input / output interface 1030, and the communication interface 1040 are communicatively connected to each other within the device via the bus 1050.

[0117] The processor 1010 can be implemented using a general-purpose CPU (Central Processing Unit), a microprocessor, an application-specific integrated circuit (ASIC), or one or more integrated circuits, and is used to execute relevant programs to implement the technical solutions provided in the embodiments of this specification.

[0118] The memory 1020 can be implemented in the form of ROM (Read Only Memory), RAM (Random Access Memory), static storage devices, dynamic storage devices, etc. The memory 1020 can store an operating system and other application programs. When the technical solutions provided in the embodiments of this specification are implemented through software or firmware, the relevant program code is stored in the memory 1020 and is called and executed by the processor 1010.

[0119] The input / output interface 1030 is used to connect input / output modules to implement information input and output. The input / output modules can be configured as components within the device (not shown in the figure) or can be externally connected to the device to provide corresponding functions. Input devices may include a keyboard, mouse, touch screen, microphone, various sensors, etc., and output devices may include a display, speaker, vibrator, indicator light, etc.

[0120] The communication interface 1040 is used to connect to a communication module (not shown) to enable communication between the device and other devices. The communication module can communicate via a wired method (such as USB, network cable, etc.) or a wireless method (such as mobile network, WiFi, Bluetooth, etc.).

[0121] The bus 1050 comprises a path for transmitting information between the various components of the device (eg, the processor 1010 , the memory 1020 , the input / output interface 1030 , and the communication interface 1040 ).

[0122] It should be noted that although the above device only shows the processor 1010, the memory 1020, the input / output interface 1030, the communication interface 1040, and the bus 1050, in a specific implementation, the device may also include other components necessary for normal operation. In addition, it will be understood by those skilled in the art that the above device may only include the components necessary to implement the embodiments of this specification, and does not necessarily include all the components shown in the figure.

[0123] The electronic device in the above embodiment is used to implement the corresponding polar code decoding method based on intelligent path selection in any of the above embodiments, and has the beneficial effects of the corresponding method embodiment, which will not be described in detail here.

[0124] Based on the same technical concept, and corresponding to any of the aforementioned embodiments and methods, the present disclosure further provides a non-transitory computer-readable storage medium storing computer instructions for causing a computer to execute the polar code decoding method based on intelligent path selection as described in any of the aforementioned embodiments.

[0125] The computer-readable media of this embodiment include permanent and non-permanent, removable and non-removable media that can be used to store information by any method or technology. The information can be computer-readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, read-only compact disc read-only memory (CD-ROM), digital versatile disc (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device.

[0126] The computer instructions stored in the storage medium of the above embodiment are used to enable the computer to execute the polar code decoding method based on intelligent path selection as described in any of the above embodiments, and have the beneficial effects of the corresponding method embodiments, which are not further described here.

[0127] Those skilled in the art should understand that the discussion of any of the above embodiments is merely illustrative and is not intended to imply that the scope of the present disclosure (including the claims) is limited to these examples. Within the scope of the present disclosure, the technical features in the above embodiments or different embodiments may be combined, the steps may be implemented in any order, and there are many other variations of the different aspects of the embodiments of the present disclosure as described above, which are not provided in detail for the sake of simplicity.

[0128] In addition, to simplify the description and discussion, and so as not to obscure the embodiments of the present disclosure, known power / ground connections to integrated circuit (IC) chips and other components may or may not be shown in the provided figures. In addition, devices may be shown in the form of block diagrams to avoid obscuring the embodiments of the present disclosure, and this also takes into account the fact that the details of the implementation of these block diagram devices are highly dependent on the platform on which the embodiments of the present disclosure are to be implemented (i.e., these details should be fully within the purview of those skilled in the art). Where specific details (e.g., circuits) are set forth to describe exemplary embodiments of the present disclosure, it will be apparent to those skilled in the art that the embodiments of the present disclosure may be implemented without these specific details or with variations in these specific details. Therefore, these descriptions should be considered illustrative rather than restrictive.

[0129] Although the present disclosure has been described in conjunction with specific embodiments thereof, many alternatives, modifications, and variations of these embodiments will be apparent to those skilled in the art based on the foregoing description. For example, other memory architectures (e.g., dynamic RAM (DRAM)) may use the embodiments discussed.

[0130] The embodiments of the present disclosure are intended to cover all such substitutions, modifications, and variations that fall within the broad scope of the appended claims. Therefore, any omissions, modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the embodiments of the present disclosure should be included in the scope of protection of the present disclosure.

Claims

1. A polar code decoding method based on intelligent path selection, characterized in that: include: Obtaining a polarization code and determining parameters of the polarization code; determining a code length of the polar code, and calculating a codeword of the polar code using the code length; The codeword is modulated to obtain a transmit vector of the codeword, and the transmit vector is input into a transmission channel to obtain a receive vector. A log-likelihood ratio of each decoded bit in each path of the polar code is calculated based on the receive vector, and a path metric of the corresponding path is calculated using the log-likelihood ratio of each decoded bit in each path. The parameters are input into a pre-trained neural network model to perform path selection for the polar code, wherein the neural network model extracts features from the path metric values ​​and determines the length of the path corresponding to the path metric value. The paths are classified in response to the path lengths. Based on the classification result, path selection is performed for the polar code to obtain a path selection result. Determining the path selection result to obtain a determination result, and inputting the determination result into a decoder; The polar code is decoded by the decoder.

2. The method according to claim 1, characterized in that Inputting the parameters into a preset neural network model to select a path for the polar code further includes: Calculating the loss of the neural network model based on the classification result; The neural network model is optimized using the loss.

3. The method according to claim 1, characterized in that The determining the result of the path selection and inputting the determination result to the decoder includes: Determining the result of the path selection by a preset threshold value; In response to determining that the result of the path selection is less than the preset threshold value, inputting the result of the path selection to the decoder; In response to determining that the result of the path selection is greater than the preset threshold value, the result of the path selection is discarded.

4. The method according to claim 3, characterized in that Determining the result of the path selection and inputting the result of the determination to the decoder further includes: in response to determining that the result of the determination is less than the path requirement of the decoder, completing the result of the determination.

5. The method according to claim 1, wherein The neural network model includes an input layer, an output layer and a hidden layer, wherein the number of neurons in the input layer and the output layer is the same as the scale of the decoder.

6. A polar code decoder based on intelligent path selection, characterized in that: include: Parameter setting module, which sets the parameters of the polarization code; a path selection module, determining a code length of the polar code and calculating a codeword of the polar code based on the code length; The codeword is modulated to obtain a transmit vector of the codeword, and the transmit vector is input into a transmission channel to obtain a receive vector. A log-likelihood ratio of each decoded bit in each path of the polar code is calculated based on the receive vector, and a path metric of the corresponding path is calculated using the log-likelihood ratio of each decoded bit in each path. The parameters are input into a pre-trained neural network model to perform path selection for the polar code, wherein the neural network model extracts features from the path metric values ​​and determines the length of the path corresponding to the path metric value. The paths are classified in response to the path lengths; and the path selection for the polar code is performed based on the classification result. A path determination module determines the result of the path selection; The path completion module completes the result of the determination in response to determining that the result of the determination is less than the path requirement of the decoder.

7. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the program, the method according to any one of claims 1 to 5 is implemented. 8 . A non-transitory computer-readable storage medium storing computer instructions, wherein the computer instructions are configured to cause a computer to execute the method according to claim 1 .

Images