Power imbalance multilevel decoding method and system for sparse code multiple access

Abstract

The invention discloses a power imbalance multistage decoding method and system for sparse code multiple access, and relates to the technical field of communication, and the method comprises the steps: mapping information bits sent by a plurality of users into multi-dimensional sparse code words through a preset codebook; constructing a factor graph matrix by adopting each codebook, grading each user according to the factor graph matrix under a preset user limiting condition, and determining # imgabs0 user levels; the # imgabs 1 # is a positive integer greater than 1; based on a preset total transmitting power, performing power imbalance distribution on each user according to # imgabs2 # user levels by adopting a progressive hierarchical power optimization algorithm, and determining a local optimal power vector; transmitting a transmitting signal corresponding to each multi-dimensional sparse code word according to the local optimal power vector; and sequentially performing power-oriented decoding on the receiving signals corresponding to the transmitting signals associated with each user level, and outputting a decoding bit sequence of # imgabs3 # user levels. According to the scheme, power imbalance and multi-level decoding are combined, and the bit error rate of the SCMA system is reduced.

Description

Technical Field

[0001] The present invention relates to the field of communication technology, and in particular to a power imbalance multi-stage decoding method and system for sparse code multiple access. Background Art

[0002] With the rapid development of mobile communication technology and the popularization of smart devices, spectrum resources are becoming increasingly scarce. In order to meet this challenge, the fifth generation of mobile communication systems (5G) has introduced the multiple access technology of non-orthogonal multiple access (NOMA), among which sparse code multiple access (SCMA) is particularly eye-catching. First, SCMA has significant advantages in spectrum efficiency, anti-interference capability and large-scale connection support. It can access more users under limited spectrum resources and meet the requirements of 5G massive connections to solve the problem of tight spectrum resources; secondly, SCMA has strong anti-interference capability and can still maintain stable communication quality in high interference environments; finally, SCMA supports large-scale connections and can meet the needs of 5G application scenarios such as the Internet of Things and large-scale sensor networks.

[0003] For SCMA systems, channel coding can effectively resist inter-user interference and improve the reliability of data transmission. A structured LDPC code called Protograph LDPC (PLDPC) code is considered to be an ideal channel coding scheme for SCMA systems. The block diagram of an uplink PLDPC coded SCMA system is shown in the figure. Figure 1 As shown: We consider a Individual users, In an SCMA system with uplink PLDPC coding of orthogonal resources, each user is independently PLDPC coded and equipped with one antenna. In order to achieve overloaded multi-user communication, , and define the overload factor as , where each user can only occupy resources to communicate, and each resource is user multiplexing; given a Users and The SCMA system with resources can be represented by a corresponding factor graph matrix It can clearly reveal the allocation relationship between users and resources; at the transmitter, first, the user's information bit sequence can be expressed as ,in , which is input into the PLDPC encoder to obtain a coded bit sequence , where the encoding rate ; Then send the coded bit sequence to the bit-level random interleaver After that, we can get an interleaved coded bit sequence after interleaving. , interleaved coded bit sequence Zhongmei The bits are mapped by the SCMA mapper into a dimensional sparse codewords, For users Codebook is a column vector in , and the size of the codebook is , is also the modulation order. Due to the sparsity of SCMA, in a sparse codeword In, only elements are non-zero elements, and the remaining The first element is the '0' element, and then each user The sparse codewords are superimposed as the transmitted signal; at the receiver, the received signal vector As the formula As shown, Indicates user The channel gain, Indicates a The additive complex Gaussian white noise vector of Represents the noise power spectral density; at the receiver, the interleaved log-likelihood ratios (LLRs) corresponding to the interleaved coded bit sequences of all users are calculated by using the log-domain Message Passing Algorithm (log-MPA) , after being processed by the deinterleaver, the LLRs sequence is obtained , and decoded by the PLDPC decoder using the belief propagation (BP) algorithm, and finally the user's decoded bit sequence is detected.

[0004] However, in the above-mentioned traditional PLDPC coded SCMA scheme, each user is allocated equal transmission power during resource allocation, and the decoding processes of different users are independent of each other, which cannot effectively eliminate interference between users, resulting in a high bit-error-rate (BER) of the SCMA system. Summary of the invention

[0005] The present invention provides a power imbalance multi-stage decoding method and system for sparse code multiple access, which solves the technical problem that in the SCMA scheme of traditional PLDPC coding, each user has equal transmission power allocation when allocating resources, and the decoding processes of different users are independent of each other, and interference between users cannot be effectively eliminated, which leads to a high bit error rate of the SCMA system.

[0006] The first aspect of the present invention provides a power imbalance multi-stage decoding method for sparse code multiple access, comprising:

[0007] Mapping the information bits sent by multiple users into multi-dimensional sparse codewords through a preset codebook;

[0008] The factor graph matrix is ​​constructed by using each codebook, and each user is graded according to the factor graph matrix under a preset user restriction condition to determine User level; is a positive integer greater than 1;

[0009] Based on the preset total transmit power, a progressive hierarchical power optimization algorithm is used. Performing unbalanced power allocation on each of the users at each user level to determine a local optimal power vector;

[0010] Transmitting a transmission signal corresponding to each of the multidimensional sparse codewords according to the local optimal power vector;

[0011] The received signals corresponding to the transmitted signals associated with each user level are sequentially power-guided decoded and output A decoded bit sequence of the user level.

[0012] Optionally, the user restriction condition includes:

[0013] The number of users included in each user level must be equal;

[0014] In each user level, each resource is reused, and each resource is reused by the same number of users.

[0015] Optionally, the total transmit power is preset based on a progressive hierarchical power optimization algorithm. The method further comprises: performing unbalanced power allocation for each user at each user level to determine a local optimal power vector, including:

[0016] After initializing the initial power of each user with equal power based on the preset total transmission power, the mutual information is calculated based on the initial power of each user using the original model external information transfer algorithm;

[0017] Performing an average operation on each of the mutual information to output an initial average mutual information, and assigning the initial average mutual information to the old average mutual information;

[0018] According to the power unit step Conduct the During the power distribution, Initial power reduction for user level , and for Initial power increase for user level , determine the User level and the The monotonic initialization power of the user level; where, is a positive number;

[0019] Based on the original model external information transfer algorithm, the monotonic initialization power is used to determine the User level and the The monotonically initialized average mutual information corresponding to the user level;

[0020] When the monotone initialized average mutual information is greater than the old average mutual information, assigning the monotone initialized average mutual information to the old average mutual information;

[0021] For the said Monotonic initialization power reduction at user level , for the said Monotonic initialization power increase at user level , determine the User level and the Local optimization power at user level and count the number of allocations ;

[0022] Based on the original model external information transfer algorithm, the local optimization power is used to determine the User level and the The local optimized average mutual information corresponding to the user level;

[0023] When the local optimized average mutual information is less than or equal to the old average mutual information, The local optimization power is taken as the User level and the Local optimal power at user level;

[0024] like If it is an even number, the power unit step is The next power allocation is carried out until Secondary power allocation, determine The local optimal power of each user level is combined into a local optimal power vector;

[0025] like If it is an odd number, the power unit step is The next power allocation is carried out until After the second power distribution, Initial power reduction for user level , and increase the local optimal power of the first user level , output the The monotonic initialization power of the first user level and the new monotonic initialization power of the first user level are used to determine the corresponding local optimal power. The local optimal powers of the user levels constitute a local optimal power vector.

[0026] Optionally, it also includes:

[0027] When the local optimized average mutual information is greater than the old average mutual information, the allocation times are determined. Is it less than the maximum number of cycles? ;

[0028] If so, the local optimization power and the local optimization average mutual information are used as the new monotone initialization power and the new old average mutual information, and the first Monotonic initialization power reduction at user level , for the said Monotonic initialization power increase at user level , until the local optimized average mutual information is less than or equal to the old average mutual information;

[0029] If not, then The local optimization power is taken as the User level and the Local optimal power at user level.

[0030] Optionally, the power-oriented decoding is performed on the received signals corresponding to the transmitted signals associated with each user level in turn, and the output is The user-level decoded bit sequence comprises:

[0031] Calculating an initial log-likelihood ratio sequence of a received signal corresponding to a transmitted signal associated with each of the user levels;

[0032] Calculating an outer interleaved log-likelihood ratio sequence using each of the initial log-likelihood ratio sequences through a log-domain message passing algorithm;

[0033] After deinterleaving the first-level user-level external interleaved log-likelihood ratio sequence, decoding is performed using a belief propagation algorithm to output first-level user-level decoded bits and a first-level external log-likelihood ratio sequence;

[0034] Interleaving the first-level external log-likelihood ratio sequences to obtain new initial log-likelihood ratio sequences of the first-level user level, and calculating new external interleaved log-likelihood ratio sequences using a log-domain message passing algorithm based on each initial log-likelihood ratio sequence;

[0035] After deinterleaving the external interleaved log-likelihood ratio sequence of the next user level, it is decoded by the belief propagation algorithm until the output The decoded bits of each user level are combined into a decoded bit sequence.

[0036] Optionally, the calculating an initial log-likelihood ratio sequence of a received signal corresponding to a transmitted signal associated with each user level includes:

[0037] Performing a logarithmic operation on the reciprocal of the modulation order of the transmission signal associated with each of the user levels to determine an initial log-likelihood ratio of the received signal of each of the transmission signals;

[0038] An initial log-likelihood ratio sequence is constructed using each of the initial log-likelihood ratios.

[0039] A second aspect of the present invention provides a power imbalance multi-stage decoding system for sparse code multiple access, comprising:

[0040] The transmitter is used to map the information bits sent by multiple users into multi-dimensional sparse codewords through a preset codebook; construct a factor graph matrix using each of the codebooks, classify each of the users according to the factor graph matrix under preset user restriction conditions, and determine User level; is a positive integer greater than 1; based on the preset total transmit power, a progressive hierarchical power optimization algorithm is used according to Performing unbalanced power allocation for each user at each user level to determine a local optimal power vector;

[0041] A Rayleigh fading channel, used for transmitting a transmission signal corresponding to each of the multidimensional sparse codewords according to the local optimal power vector;

[0042] The receiver is used to perform power-oriented decoding on the received signals corresponding to the transmitted signals associated with each user level in turn, and output A decoded bit sequence of the user level.

[0043] A third aspect of the present invention provides a computer device, comprising a memory and a processor, wherein a computer program is stored in the memory, and when the computer program is executed by the processor, the processor executes the steps of the power unbalanced multi-stage decoding method for sparse code multiple access as described in any one of the above items.

[0044] A fourth aspect of the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein when the computer program is executed, the power imbalance multi-stage decoding method for sparse code multiple access as described in any one of the above items is implemented.

[0045] A fifth aspect of the present invention provides a computer program product, comprising a computer program / instruction, which, when executed by a processor, implements the power imbalance multi-stage decoding method for sparse code multiple access as described in any one of the above items.

[0046] It can be seen from the above technical solutions that the present invention has the following advantages:

[0047] The above scheme of the present invention provides a power imbalance multi-stage decoding method for sparse code multiple access, including: mapping information bits sent by multiple users into multi-dimensional sparse code words through a preset code book; using each code book to construct a factor graph matrix, classifying each user according to the factor graph matrix under a preset user restriction condition, and determining User level; is a positive integer greater than 1; based on the preset total transmit power, a progressive hierarchical power optimization algorithm is used according to The power is unbalancedly allocated to each user at each user level to determine the local optimal power vector; the transmission signal corresponding to each multi-dimensional sparse codeword is transmitted according to the local optimal power vector; the received signal corresponding to the transmission signal associated with each user level is power-guided decoded in turn, and the output is Based on the above scheme, the mapping relationship between users and power resources is clearly revealed through the factor graph matrix to regularly classify users. Under the premise of keeping the total transmission power unchanged, power imbalance is allocated to each level of users to obtain the local optimal power vector, and the power-oriented decoding is performed based on the characteristics of power imbalance. In this way, power imbalance is combined with multi-level decoding, which helps to further eliminate interference between users and thus reduce the bit error rate of the SCMA system. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0049] Figure 1 A structural block diagram of the SCMA scheme of the traditional PLDPC encoding provided by the present invention;

[0050] Figure 2 A flowchart of the steps of a power imbalance multi-stage decoding method for sparse code multiple access provided by an embodiment of the present invention;

[0051] Figure 3 A block diagram of a power-oriented MLD receiver provided in an embodiment of the present invention;

[0052] Figure 4 AVE-BER curves of the PI-MLD-SCMA system provided by the embodiment of the present invention and the traditional system in simulation experiments;

[0053] Figure 5 A structural block diagram of a power imbalanced multi-stage decoding system for sparse code multiple access provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0054] The embodiment of the present invention provides a power-unbalanced multi-stage decoding method and system for sparse code multiple access, which is used to solve the technical problem that in the SCMA scheme of traditional PLDPC coding, each user has equal transmission power allocation when allocating resources, and the decoding processes of different users are independent of each other, and interference between users cannot be effectively eliminated, which leads to a high bit error rate in the SCMA system.

[0055] In order to make the purpose, features and advantages of the present invention more obvious and easy to understand, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the embodiments described below are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0056] See also Figure 2 , Figure 2 A flowchart of the steps of a power imbalance multi-stage decoding method for sparse code multiple access provided by an embodiment of the present invention.

[0057] The present invention provides a power imbalance multi-stage decoding method for sparse code multiple access, comprising:

[0058] Step 101: Map information bits sent by multiple users into multi-dimensional sparse codewords through a preset codebook.

[0059] It should be noted that, in the transmitter of the SCMA system, the user's information bits are processed by PLDPC coding and mapped to sparse multidimensional complex constellation points, namely, multidimensional sparse codewords. These multidimensional sparse codewords can be selected from a pre-designed codebook, and each user has his own codebook. This direct mapping method improves the spectrum efficiency and achieves a large constellation shaping gain. At the same time, these codewords are sparse in the frequency domain, that is, each codeword of each user only occupies part of the subcarrier for communication data transmission.

[0060] Step 102: construct a factor graph matrix using each codebook, classify each user according to the factor graph matrix under the preset user restriction condition, and determine User level; is a positive integer greater than 1.

[0061] User restrictions include:

[0062] The number of users included in each user level must be equal;

[0063] In each user level, each resource is reused, and each resource is reused by the same number of users.

[0064] It should be noted that for a given For the SCMA system, we use its corresponding factor graph matrix Perform user classification, where is the number of orthogonal resources, is the number of users; for example, we give a The SCMA system and a The factor graph matrix corresponding to the SCMA system is:

[0065] ;

[0066] In a factor graph matrix, each row vector represents an orthogonal resource, and each column vector represents an independent user. In the first column of the matrix, two elements '1' indicate that user 1 occupies the second and fourth resources for communication transmission, and '0' elements indicate that they are not occupied, and the number of resources occupied by a single user is , the number of times a single resource is reused by users Similarly, in the matrix In the second column, the two elements '1' represent that user 2 occupies the second resource and the sixth resource for communication transmission respectively;

[0067] Before classification, we impose the following user restrictions: (i) For a given SCMA system, users can be classified into user levels, the number of users included in each user level must be equal, that is, (ii) For a given SCMA system, all All resources are reused, and each resource is reused by the same number of users;

[0068] Based on user constraints (i) and (ii), we give an example of common SCMA system classification: for a factor graph matrix of The SCMA system can divide the six users into three levels according to the user restriction conditions. User 1 and user 2 are divided into the first level, user 3 and user 4 are divided into the first level, and user 5 and user 6 are divided into the first level. For a factor graph matrix of In the SCMA system, all users can be divided into three levels: user 1, user 2, and user 3 are divided into level 1, user 4, user 5, and user 6 are divided into level 1, and user 7, user 8, and user 9 are divided into level 1. In addition, the user restriction conditions can also include (iii) for users with an overload factor For SCMA systems, systems with the same overload factor can make the same number of classifications, e.g. SCMA system and The overload factor of the SCMA system is 1.5. The SCMA system can refer to The SCMA system is classified according to the number of grades; the common SCMA system grades are shown in Table 1:

[0069] Table 1 Common SCMA system classification

[0070]

[0071] 103. Based on the preset total transmission power, a progressive hierarchical power optimization algorithm is used according to The power is unbalancedly allocated to each user at each user level to determine the local optimal power vector.

[0072] Step 103 includes the following sub-steps:

[0073] After initializing the initial power of each user with equal power based on the preset total transmission power, the original model external information transfer algorithm is used to calculate the mutual information based on the initial power of each user;

[0074] Perform an average operation on each mutual information to output the initial average mutual information, and assign the initial average mutual information to the old average mutual information;

[0075] According to the power unit step Conduct the During the power distribution, Initial power reduction for user level , and for Initial power increase for user level , determine the User level and The monotonic initialization power of the user level; where, is a positive number;

[0076] Based on the original model external information transfer algorithm, the monotonic initialization power is used to determine the User level and The monotonically initialized average mutual information corresponding to the user level;

[0077] When the monotone initialized average mutual information is greater than the old average mutual information, the monotone initialized average mutual information is assigned to the old average mutual information;

[0078] For Monotonic initialization power reduction at user level , for Monotonic initialization power increase at user level , determine the User level and Local optimization power at user level and count the number of allocations ;

[0079] Based on the original model external information transfer algorithm, the local optimization power is used to determine the User level and The local optimized average mutual information corresponding to the user level;

[0080] When the local optimized average mutual information is less than or equal to the old average mutual information, The local optimization power is taken as the User level and Local optimal power at user level;

[0081] like If it is an even number, the power unit step is The next power allocation is carried out until Secondary power allocation, determine The local optimal power of each user level forms a local optimal power vector;

[0082] like If it is an odd number, the power unit step is The next power allocation is carried out until After the second power distribution, Initial power reduction for user level , and increase the local optimal power of the first user level , output The monotonic initialization power of the first user level and the new monotonic initialization power of the first user level are used to determine the corresponding local optimal power. The local optimal power vector is composed of the local optimal power at each user level.

[0083] Also includes:

[0084] When the monotone initialized average mutual information is less than or equal to the old average mutual information, the monotone initialized power is used as the first User level and Local optimal power at user level.

[0085] Also includes:

[0086] When the local optimized average mutual information is greater than the old average mutual information, the number of allocations is determined. Is it less than the maximum number of cycles? ;

[0087] If so, the local optimized power and the local optimized average mutual information are used as the new monotone initialization power and the new old average mutual information, and the Monotonic initialization power reduction at user level , for Monotonic initialization power increase at user level , until the local optimized average mutual information is less than or equal to the old average mutual information;

[0088] If not, then The local optimization power is taken as the User level and Local optimal power at user level.

[0089] It should be noted that after classifying all users in the SCMA system, we proposed a progressive hierarchical power optimization algorithm to implement power-imbalanced allocation (PIA) for each user level, and analyzed the average mutual information (AMI) convergence behavior of users with the help of the protograph extrinsic information transfer (PEXIT) algorithm, a very effective system performance analysis tool, to determine whether the PIA scheme has reached the local optimum;

[0090] Before introducing the progressive hierarchical power optimization algorithm, the following five power constraints are proposed:

[0091] (i) Under the premise that the total transmission power preset by the entire SCMA system remains unchanged, the power of each level is normalized to equal power (that is, the power of all users is 1);

[0092] (ii) The transmission power of different users in the same user class is equal, while the transmission power of different users in different user classes is unequal;

[0093] (iii) Assume that a SCMA system consists of Users and resources, divided into level, then the initial power vector corresponding to all levels is , then the local optimal power vector of The elements must satisfy ,and ,in ;

[0094] (iv) When When is an even number, the number of user levels with low power (power < 1) is equal to the number of user levels with high power (power > 1), which is ( represents rounding down), and when When is an odd number, the number of low-power user levels is one more than the number of high-power user levels, that is, the number of low-power levels is ( Indicates rounding up to an integer);

[0095] (v) The minimum power unit step size of PIA is set to It is understandable that the power unit step size can be set adaptively; it should be noted that before implementing PIA, all user levels can be sorted arbitrarily. The specific pseudo code of the progressive hierarchical power optimization algorithm is shown in Table 2:

[0096] Table 2 Pseudo code of the progressive hierarchical power optimization algorithm

[0097]

[0098] It can be seen from Table 2 that each time the power allocation operation is performed on the power of two user levels, the power of other levels remains unchanged during each power allocation operation, and the power allocation operation is performed on one level, that is, the same power allocation operation is performed on all users of this level; for example, in the During the power distribution operation, the last level, i.e. Initial power of the stage and the initial power of the first stage Perform power distribution operation; when When it is an even number, a total of Secondary power distribution operation, where Indicates Secondary power distribution operation, when When it is an odd number, a total of Second power distribution operation, this time Indicates The following is the detailed process of the progressive hierarchical power optimization algorithm:

[0099] Step 1: Initialize the initial power of all user levels to 1, that is , by using the PEXIT algorithm to calculate the MI (mutual information) using the initial power of each user, and then averaging the MI of each user to obtain the initial average mutual information (AMI) , initialize the old average mutual information , used to store The value of

[0100] Step 2: Assume this is the The second power distribution operation Initial power of the stage reduce , while at the same time Level power Add the same , and the corresponding initialization power is determined to be monotonically decreasing, which is to ensure the monotonically decreasing characteristics of the power limitation condition (iii); For example, the first power allocation operation, we need to Initial power of the stage reduce Get the corresponding monotonic initialization power, and at the same time the power of the first stage Increase Get the corresponding monotonic initialization power; use the PEXIT algorithm based on the The monotonic initialization power of the first level and the initial power of the remaining user levels calculate the new That is, the average mutual information is monotonically initialized, and then the process jumps to step 4. It can be understood that in order to simplify the algorithm process, the step of calculating the average mutual information of monotonically initialized can also be omitted in step 2, and the process jumps directly to step 3;

[0101] Step 3: Continue to Monotonic initialization power reduction of the stage , for The monotonic initialization power of the stage increases by the same , get the corresponding local optimal power, and then use the PEXIT algorithm to calculate the new That is, local optimization of the average mutual information and counting the number of allocations ;

[0102] Step 4: Compare and If the size , the monotonically initialized average mutual information or the locally optimized average mutual information The value of the old average mutual information , and then repeat step 3 until the ; Assume that the monotone initialization average mutual information calculated in step 2 is Less than or equal to the old average mutual information , then directly use the monotonic initialization power allocated in step 2 as the first User level and The local optimal power of the user level; assuming that The monotonic initialization power of the first The monotonic initialization power of the stage is executed After step 3, , then take The power allocation of step 3 is The value of AMI after iterations has reached its maximum value, which proves and The power allocation has reached a local optimum. At this time, combining steps 2 to 4, we can get Value in the power distribution operation: Level power , No. Level power ,by For example, Level power , the power of the first stage If you execute After step 3, it is still , then take the The power allocation of step 3; it should be noted that In the secondary power distribution operation, The values ​​of may be different, and step 3 can be repeated at most times, while at the same time needing to satisfy , this is to meet the power limitation condition (iii);

[0103] Step 5: When When it is an even number, a total of The last power distribution operation is the The second time, in Level and Between levels, for example The last power allocation operation is performed on the third and second levels, and then the local optimal power vector is obtained. , the optimization process of PIA is completed; and when When it is an odd number, a total of Second power distribution operation, while the previous Sub-power distribution operation and The process is the same when it is an even number (the power allocation operation of steps 2 to 4 is also performed, but the specific power may not be the same). Sub-power distribution operation and When it is an even number, it is similar. Level and The last power distribution operation is the The second power distribution operation is Initial power and the local optimal power of the first stage The operation is performed between step 3 and step 4. It should be noted that since the power of the first level of the first power allocation operation has increased, at this time, the local optimal power Continue to increase on the basis of After the power allocation operation, the local optimal power vector can also be obtained , the optimization process of PIA is now completed: As an example, we first perform step 2 to step 4 power allocation operations on the third stage and the first stage, and then perform step 4 on the second stage and the first stage. The secondary power distribution operation, the operation steps are step 3 to step 4.

[0104] Step 104: transmit a transmission signal corresponding to each multi-dimensional sparse codeword according to the local optimal power vector.

[0105] It should be noted that the multi-dimensional sparse codewords of each user will be superimposed into a transmission signal in the transmitter. Therefore, after completing the transmission power allocation for each user, each transmission signal can be transmitted to the transmitter through the Rayleigh fading channel as a medium according to the local optimal power vector.

[0106] Step 105: Power-directed decoding is performed on the received signals corresponding to the transmitted signals associated with each user level in turn, and the output is A user-level decoded bit sequence.

[0107] Step 105 includes the following sub-steps:

[0108] S1. Calculate the initial log-likelihood ratio sequence of the received signal corresponding to the transmitted signal associated with each user level;

[0109] S2, calculating an outer interleaved log-likelihood ratio sequence using each initial log-likelihood ratio sequence through a log-domain message passing algorithm;

[0110] S3, after deinterleaving the first-level user-level external interleaved log-likelihood ratio sequence, decoding it through a belief propagation algorithm, and outputting the first-level user-level decoded bits and the first-level external log-likelihood ratio sequence;

[0111] S4, interleaving the first-level external log-likelihood ratio sequences to obtain new initial log-likelihood ratio sequences of the first-level user level, and calculating new external interleaved log-likelihood ratio sequences using a log-domain message passing algorithm based on each initial log-likelihood ratio sequence;

[0112] S5, after deinterleaving the external interleaved log-likelihood ratio sequence of the next user level, decode it through the belief propagation algorithm until the output The decoded bits of each user level are combined into a decoded bit sequence.

[0113] Optionally, sub-step S1 includes:

[0114] Performing a logarithmic operation on the reciprocal of the modulation order of the transmitted signal associated with each user level to determine an initial log-likelihood ratio of the received signal of each transmitted signal;

[0115] The initial log-likelihood ratio sequence is constructed using each initial log-likelihood ratio.

[0116] It should be noted that Figure 3The block diagram of the MLD receiver based on power steering designed in this embodiment is shown. It can be understood that the multi-level decoding (MLD) widely used in the multi-data stream system is a sequential decoding scheme. The currently decoded data stream can improve the decoding performance of the subsequent data stream. The SCMA system, as a multi-user non-orthogonal multiple access technology, naturally has the characteristics of multiple data streams and is very suitable for combination with MLD. The problem of interference between users can be solved by power-oriented decoding, that is, decoding different user levels from high power to low power:

[0117] exist Figure 3 middle, Indicates the received signal. and They represent the maximum number of iterations of the SCMA detector and the PLDPC decoder, respectively. The dotted line indicates all The decoding order of users of different levels; when the SCMA detector receives the received signal, it calculates the initial log-likelihood ratio of the received signal corresponding to the transmitted signal associated with each user level. Since this embodiment uses the logarithmic domain message passing algorithm and assumes that each user is equally likely to choose a multidimensional sparse codeword as the transmitted signal, in one implementation, the channel initialization method is used to calculate the received signal. ,in is the modulation order, which can be understood as the size of the codebook. Since the size of the codebook of each user in the same SCMA system is unified, the modulation order is also equal; then the corresponding initial log-likelihood ratio is formed into an initial log-likelihood ratio sequence (initial LLRs sequence) according to the user level. , and then use the log-MPA to calculate the outer interleaved log-likelihood ratio sequence (outer interleaved LLR sequence) using each initial LLRs sequence , ;

[0118] In the designed receiver, after the SCMA detector completes the first round of detection, the first-level user with the highest power performs decoding to obtain the decoded bits, first through the deinterleaver The outer interleaved log-likelihood ratio sequence for the first user level After deinterleaving, the PLDPC decoder uses the belief propagation algorithm (BP) to decode and output the first-level user-level decoded bits and the first-level external log-likelihood ratio sequence. , then through the interleaver Interleave the first-level external log-likelihood ratio sequence to obtain a new initial LLRs sequence at the first-level user level ;

[0119] Because there is no gain from the previous iteration of the receiver in the first SCMA detection, compared with users at other levels, users at the highest power level can obtain a more accurate initial LLRs sequence by decoding first. Therefore, the new initial LLRs sequence obtained after the first level decoding is compared with the initial LLRs sequences of other levels ( ) are sent to the SCMA detector for the second round of detection, and the log-MPA algorithm is used again to calculate the new external interleaved LLRs sequence of users at all levels. ; Then, the user of the second highest power user level performs BP decoding, and similarly, the outer interleaved LLRs sequence of the second level After deinterleaving and decoding, the second-level decoded bits and the second-level external LLRs sequence are output. , interleave the second-level external LLRs sequence to obtain the second-level new initial LLRs sequence , and and other levels of initial LLRs sequences ( ) Use the log-MPA algorithm to calculate the new outer interleaved LLRs sequence According to the above process, when the user with the lowest power level completes BP decoding, the output The decoding bits of the user levels are combined into a decoding bit sequence, and the work of the receiver is completed. Since the first-level user with the highest power obtains the performance gain brought by the power increase, the other levels ( ) all users obtain the significant performance gain brought by the above power-oriented MLD receiver scheme.

[0120] In order to clearly illustrate the technical effect of the power imbalance multi-stage decoding scheme for sparse code multiple access provided by this embodiment, the PLDPC coded power imbalance multi-stage decoding SCMA system (PI-MLD-SCMA system) formed by the scheme of this embodiment and the traditional PLDPC coded SCMA system are simulated:

[0121] For the traditional PLDPC coded SCMA system, we do not perform classification, the power of all users is set to 1, and the Figure 1 The conventional receiver shown in the figure; both systems are encoded using the punctured PLDPC code designed by us, the basic matrix of the PLDPC code As shown below:

[0122] ;

[0123] We puncture the second column with the highest degree distribution in the basic matrix. The information bit length of each user is 1200 and the code rate is 1 / 2. The simulation of the two SCMA systems is carried out under Rayleigh fading channels. The SCMA systems used include and Two scales, The SCMA system uses SCMA codebook-1 and codebook-2, and The SCMA system uses SCMA codebook-3 and codebook-4, and the modulation order is ;

[0124] After searching with the progressive hierarchical power optimization algorithm, we obtain the local optimal power vector of codebook-1 , the local optimal power vector of codebook-2 , the local optimal power vector of codebook-3 , the local optimal power vector of codebook-4 is ; The SCMA detector uses the log-MPA algorithm with a maximum number of iterations of , PLDPC decoding uses BP algorithm, and the maximum number of iterations is ;

[0125] Figure 4 The following table shows the and In the SCMA system, codebooks -1 to -4 and their corresponding local optimal power vectors ( ) of the PI-MLD-SCMA system and the traditional SCMA system, the average BER (average BER, AVE-BER) performance comparison, it can be observed that and In the simulation of two-scale SCMA systems with a total of four codebooks, the PI-MLD-SCMA system with a local optimal power vector achieved a gain of more than 1.6 dB compared with the traditional one, which has a significant improvement in system performance.

[0126] In an embodiment of the present invention, the mapping relationship between users and power resources is clearly revealed through a factor graph matrix to regularly classify users. Under the premise of keeping the total transmission power unchanged, power is unbalancedly allocated to each level of users to obtain a local optimal power vector, and power-oriented decoding is performed based on the characteristics of power imbalance. In this way, power imbalance is combined with multi-level decoding, which helps to further eliminate interference between users, thereby reducing the bit error rate of the PLDPC-coded SCMA system.

[0127] See also Figure 5 , Figure 5 A structural block diagram of a power imbalanced multi-stage decoding system for sparse code multiple access provided by an embodiment of the present invention.

[0128] The present invention provides a power imbalance multi-stage decoding system for sparse code multiple access, comprising:

[0129] The transmitter 501 is used to map the information bits sent by multiple users into multi-dimensional sparse codewords through a preset codebook; construct a factor graph matrix using each codebook, classify each user according to the factor graph matrix under a preset user restriction condition, and determine User level; is a positive integer greater than 1; based on the preset total transmit power, a progressive hierarchical power optimization algorithm is used according to Perform unbalanced power allocation on each user level and determine the local optimal power vector;

[0130] A Rayleigh fading channel 502, used to transmit a transmission signal corresponding to each multi-dimensional sparse codeword according to a local optimal power vector;

[0131] The receiver 503 is used to perform power-oriented decoding on the received signals corresponding to the transmitted signals associated with each user level in turn, and output A user-level decoded bit sequence.

[0132] Optionally, user restrictions include:

[0133] The number of users included in each user level must be equal;

[0134] In each user level, each resource is reused, and each resource is reused by the same number of users.

[0135] Optionally, based on the preset total transmit power, a progressive hierarchical power optimization algorithm is used according to The power is unbalancedly allocated to each user at the user level to determine the local optimal power vector, including:

[0136] After initializing the initial power of each user with equal power based on the preset total transmission power, the original model external information transfer algorithm is used to calculate the mutual information based on the initial power of each user;

[0137] Perform an average operation on each mutual information to output the initial average mutual information, and assign the initial average mutual information to the old average mutual information;

[0138] According to the power unit step Conduct the During the power distribution, Initial power reduction for user level , and for Initial power increase for user level , determine the User level and The monotonic initialization power of the user level; where, is a positive number;

[0139] Based on the original model external information transfer algorithm, the monotonic initialization power is used to determine the User level and The monotonically initialized average mutual information corresponding to the user level;

[0140] When the monotone initialized average mutual information is greater than the old average mutual information, the monotone initialized average mutual information is assigned to the old average mutual information;

[0141] For Monotonic initialization power reduction at user level , for Monotonic initialization power increase at user level , determine the User level and Local optimization power at user level and count the number of allocations ;

[0142] Based on the original model external information transfer algorithm, the local optimization power is used to determine the User level and The local optimized average mutual information corresponding to the user level;

[0143] When the local optimized average mutual information is less than or equal to the old average mutual information, The local optimization power is taken as the User level and Local optimal power at user level;

[0144] like If it is an even number, the power unit step is The next power allocation is carried out until Secondary power allocation, determine The local optimal power of each user level forms a local optimal power vector;

[0145] like If it is an odd number, the power unit step is The next power allocation is carried out until After the second power distribution, Initial power reduction for user level , and increase the local optimal power of the first user level , output The monotonic initialization power of the first user level and the new monotonic initialization power of the first user level are used to determine the corresponding local optimal power. The local optimal power vector is composed of the local optimal power at each user level.

[0146] Optionally, it also includes:

[0147] When the monotone initialized average mutual information is less than or equal to the old average mutual information, the monotone initialized power is used as the first User level and Local optimal power at user level.

[0148] Optionally, it also includes:

[0149] When the local optimized average mutual information is greater than the old average mutual information, the number of allocations is determined. Is it less than the maximum number of cycles? ;

[0150] If so, the local optimized power and the local optimized average mutual information are used as the new monotone initialization power and the new old average mutual information, and the Monotonic initialization power reduction at user level , for Monotonic initialization power increase at user level , until the local optimized average mutual information is less than or equal to the old average mutual information;

[0151] If not, then The local optimization power is taken as the User level and Local optimal power at user level.

[0152] Optionally, power-directed decoding is performed on the received signals corresponding to the transmitted signals associated with each user level in turn, and the output is A user-level decoded bit sequence, including:

[0153] Calculating an initial log-likelihood ratio sequence of received signals corresponding to the transmitted signals associated with each user level;

[0154] Using each initial log-likelihood ratio sequence to calculate an outer interleaved log-likelihood ratio sequence through a log-domain message passing algorithm;

[0155] After deinterleaving the first-level user-level external interleaved log-likelihood ratio sequence, decoding is performed using a belief propagation algorithm to output first-level user-level decoded bits and a first-level external log-likelihood ratio sequence;

[0156] Interleaving the first-level external log-likelihood ratio sequences to obtain new initial log-likelihood ratio sequences of the first-level user level, and calculating new external interleaved log-likelihood ratio sequences using a log-domain message passing algorithm based on each initial log-likelihood ratio sequence;

[0157] After deinterleaving the external interleaved log-likelihood ratio sequence of the next user level, it is decoded by the belief propagation algorithm until the output The decoded bits of each user level are combined into a decoded bit sequence.

[0158] Optionally, calculating an initial log-likelihood ratio sequence of a received signal corresponding to a transmitted signal associated with each user level includes:

[0159] Performing a logarithmic operation on the reciprocal of the modulation order of the transmitted signal associated with each user level to determine an initial log-likelihood ratio of the received signal of each transmitted signal;

[0160] The initial log-likelihood ratio sequence is constructed using each initial log-likelihood ratio.

[0161] An embodiment of the present invention further provides a computer device, including a memory and a processor, wherein a computer program is stored in the memory; when the computer program is executed by the processor, the processor executes the steps of the power imbalance multi-stage decoding method for sparse code multiple access as described in any of the above embodiments.

[0162] An embodiment of the present invention further provides a computer-readable storage medium having a computer program / instruction stored thereon. When the computer program / instruction is executed by a processor, the steps of the power imbalance multi-stage decoding method for sparse code multiple access as in any of the above embodiments are implemented.

[0163] An embodiment of the present invention further provides a computer program product, including a computer program / instruction, which, when executed by a processor, implements the steps of the power imbalance multi-stage decoding method for sparse code multiple access as in any of the above embodiments.

[0164] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems and units described above can refer to the corresponding processes in the aforementioned method embodiments, and will not be repeated here.

[0165] In the several embodiments provided in the present application, it should be understood that the disclosed systems and methods can be implemented in other ways. For example, the system embodiments described above are only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of systems or units, which can be electrical, mechanical or other forms.

[0166] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0167] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0168] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc. Various media that can store program codes.

[0169] As described above, the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features thereof may be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A power imbalance multi-stage decoding method for sparse code multiple access, characterized in that: include: Mapping the information bits sent by multiple users into multi-dimensional sparse codewords through a preset codebook; The factor graph matrix is ​​constructed by using each codebook, and each user is graded according to the factor graph matrix under a preset user restriction condition to determine User level; is a positive integer greater than 1; Based on the preset total transmit power, a progressive hierarchical power optimization algorithm is used. Performing unbalanced power allocation on each of the users at each user level to determine a local optimal power vector; Transmitting a transmission signal corresponding to each of the multidimensional sparse codewords according to the local optimal power vector; The received signals corresponding to the transmitted signals associated with each user level are sequentially power-guided decoded and output A decoded bit sequence of the user level.

2. The power imbalance multi-stage decoding method for sparse code multiple access according to claim 1, characterized in that: The user restriction conditions include: The number of users included in each user level must be equal; In each user level, each resource is reused, and each resource is reused by the same number of users.

3. The power imbalance multi-stage decoding method for sparse code multiple access according to claim 1, characterized in that: The total transmission power is preset based on the progressive hierarchical power optimization algorithm. The method further comprises: performing unbalanced power allocation for each user at each user level to determine a local optimal power vector, including: After initializing the initial power of each user with equal power based on the preset total transmission power, the mutual information is calculated based on the initial power of each user using the original model external information transfer algorithm; Performing an average operation on each of the mutual information to output an initial average mutual information, and assigning the initial average mutual information to the old average mutual information; According to the power unit step Conduct the During the power distribution, Initial power reduction for user level , and for Initial power increase for user level , determine the User level and the The monotonic initialization power of the user level; where, is a positive number; Based on the original model external information transfer algorithm, the monotonic initialization power is used to determine the User level and the The monotonically initialized average mutual information corresponding to the user level; When the monotone initialized average mutual information is greater than the old average mutual information, assigning the monotone initialized average mutual information to the old average mutual information; Regarding the User-level monotonic initialization power reduction , for the said Monotonic initialization power increase at user level , determine the User level and the Local optimization power at user level and count the number of allocations ; Based on the original model external information transfer algorithm, the local optimization power is used to determine the User level and the The local optimized average mutual information corresponding to the user level; When the local optimized average mutual information is less than or equal to the old average mutual information, The local optimization power is taken as the User level and the Local optimal power at user level; like If it is an even number, the power unit step length is The next power allocation is carried out until Secondary power allocation, determine The local optimal power of each user level is combined into a local optimal power vector; like If it is an odd number, the power unit step is The next power allocation is carried out until After the second power distribution, Initial power reduction for user level , and increase the local optimal power of the first user level , output the The monotonic initialization power of the first user level and the new monotonic initialization power of the first user level are used to determine the corresponding local optimal power. The local optimal powers of the user levels constitute a local optimal power vector.

4. The power imbalance multi-stage decoding method for sparse code multiple access according to claim 3, characterized in that: Also includes: When the local optimized average mutual information is greater than the old average mutual information, the allocation times are determined. Is it less than the maximum number of cycles? ; If so, the local optimization power and the local optimization average mutual information are used as the new monotone initialization power and the new old average mutual information, and the first Monotonic initialization power reduction at user level , for the said Monotonic initialization power increase at user level , until the local optimized average mutual information is less than or equal to the old average mutual information; If not, then The local optimization power is taken as the User level and the Local optimal power at user level.

5. The power imbalance multi-stage decoding method for sparse code multiple access according to claim 1, characterized in that: The power-oriented decoding is performed on the received signals corresponding to the transmitted signals associated with each user level in turn, and the output is The user-level decoded bit sequence comprises: Calculating an initial log-likelihood ratio sequence of a received signal corresponding to a transmitted signal associated with each of the user levels; Calculating an outer interleaved log-likelihood ratio sequence using each of the initial log-likelihood ratio sequences through a log-domain message passing algorithm; After deinterleaving the first-level user-level external interleaved log-likelihood ratio sequence, decoding is performed using a belief propagation algorithm to output first-level user-level decoded bits and a first-level external log-likelihood ratio sequence; Interleaving the first-level external log-likelihood ratio sequences to obtain new initial log-likelihood ratio sequences of the first-level user level, and calculating new external interleaved log-likelihood ratio sequences using a log-domain message passing algorithm based on each initial log-likelihood ratio sequence; After deinterleaving the external interleaved log-likelihood ratio sequence of the next user level, it is decoded by the belief propagation algorithm until the output The decoded bits of each user level are combined into a decoded bit sequence.

6. The power imbalance multi-stage decoding method for sparse code multiple access according to claim 5, characterized in that: The calculating of an initial log-likelihood ratio sequence of a received signal corresponding to a transmitted signal associated with each user level comprises: Performing a logarithmic operation on the reciprocal of the modulation order of the transmission signal associated with each of the user levels to determine an initial log-likelihood ratio of the received signal of each of the transmission signals; An initial log-likelihood ratio sequence is constructed using each of the initial log-likelihood ratios.

7. A power imbalance multi-stage decoding system for sparse code multiple access, characterized in that: include: A transmitter, used to map information bits sent by multiple users into multi-dimensional sparse codewords through a preset codebook; The factor graph matrix is ​​constructed by using each codebook, and each user is graded according to the factor graph matrix under a preset user restriction condition to determine User level; is a positive integer greater than 1; based on the preset total transmit power, a progressive hierarchical power optimization algorithm is used according to Performing unbalanced power allocation for each user at each user level to determine a local optimal power vector; A Rayleigh fading channel, used for transmitting a transmission signal corresponding to each of the multidimensional sparse codewords according to the local optimal power vector; The receiver is used to perform power-oriented decoding on the received signals corresponding to the transmitted signals associated with each user level in turn, and output A decoded bit sequence of the user level.

8. A computer device, characterized in that: It comprises a memory and a processor, wherein a computer program is stored in the memory, and when the computer program is executed by the processor, the processor executes the steps of the power imbalance multi-stage decoding method for sparse code multiple access as described in any one of claims 1-6.

9. A computer-readable storage medium having a computer program / instruction stored thereon, characterized in that: When the computer program / instructions are executed by a processor, the steps of the power imbalance multi-stage decoding method for sparse code multiple access as described in any one of claims 1-6 are implemented.

10. A computer program product comprising a computer program / instructions, characterized in that When the computer program / instructions are executed by a processor, the steps of the power imbalance multi-stage decoding method for sparse code multiple access as described in any one of claims 1-6 are implemented.

Images