Construction method, coding method and decoding method of P element group EA code
By constructing an EA code that satisfies USPM constraints on GF(pm), the problem of expanding binary sources to p-primary sources is solved, and the multiple access system support for multi-primary sources is realized, and communication transmission is improved spectrum efficiency.
Patent Information
- Application Number
- CN202411071699.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-06
- Publication Date
- 2024-12-03
AI Technical Summary
The prior art is difficult to extend binary sources to more general p-region sources, resulting in difficulty in distinguishing finite domain virtual resource blocks of different users in the FFMA system, limiting the spectrum efficiency of communication transmission.
A method for constructing a P element element group EA code is proposed. By using the original element α to represent the element group on GF (pm), an EA code that satisfies the USPM constraints, including a double-coded EA code and a generalized AIEP code, using the base The decomposition method constructs channel encoding to realize the support of multiple access systems for p-digit sources.
It effectively improves the spectrum efficiency of communication transmission, and can support multi-access access to multi-price sources in massive users and short packet communication scenarios, achieving higher resource utilization.
Smart Images

Figure CN119070950A_ABST
Abstract
Description
Technical Field
[0001] The invention belongs to the field of communication technology and relates to a construction method, encoding and decoding method of an EA code. Background Art
[0002] Ultra-large-scale connection is one of the five typical scenarios of 6G, which has broad application prospects. The two key features of this scenario are small packet transmission and random user access. In order to support massive user access in the case of short packets, FFMA technology is proposed. This technology distinguishes users in a limited domain by using different element pairs (element-pairs, EPs). EP in GF (p m ) is constructed on the basis of the EP code, and the formed EP code is allocated to different users as a virtual resource block, forming four transmission modes: FF-TDMA, FF-CDMA, FF-CCMA, and FF-NOMA, and the mapping from the finite field to the complex field is realized through the 3ASK signal.
[0003] The information sequence of EP code is binary because of the influence of von Neumann's computer system. With the update of technology, ternary and even multi-binary logic is recognized by more and more people. Compared with binary logic, ternary logic can cover binary logic and has all the advantages of binary logic, achieving faster calculation and more compact storage. In the field of communication, there are many studies related to ternary systems, including ternary cyclic codes, ternary convolutional codes of ternary phase shift keying (TPSK) signals, Turbo codes of TPSK signals, ternary Hamming codes, TPSK and 9QAM signals are also widely used in Gaussian point-to-point ternary data transmission. The ternary system can provide higher spectrum efficiency for communication transmission.
[0004] Therefore, it is attractive to extend the binary source to a more general p-ary source, mainly a ternary source, and study the corresponding characteristics of the FFMA system. However, the EP code is composed of element pairs, which can only correspond to 0 and 1 in the binary system. For the p-ary information sequence, how to find the corresponding finite field virtual resource blocks that distinguish different users and how to apply it in the FFMA system are problems that need to be solved. Summary of the invention
[0005] The present invention aims to solve the problem that there is currently a lack of a method for finding finite field virtual resource blocks that distinguish different users in the process of expanding a binary information source to a more general p-ary information source, thereby limiting the communication transmission to provide higher spectrum efficiency.
[0006] A method for constructing a P-element group EA code, comprising:
[0007] Assume that α is GF(p m ) of the primitive element, GF(p m) can be uniquely represented by a vector of length m on GF(p), that is, represented by an m-tuple, and the primitive element α can represent the element group EA code;
[0008] For a ternary source, when p = 3, the possible output of the source is C T =(0,1,2), when the J user information is all 0, 1, 2, the corresponding generating matrices are formed
[0009] Each of the J users sends K data, and the sending sequence of the jth user is expressed as d j =(d j,0 ,d j,1 ,...,d j,k ,…d j,K-1 ), where d j,k ∈{0,1,2}, 1≤j≤J, 0≤k≤K-1; the user's data is converted through the conversion function F p2q Converted to GF(3 m ) on the EA codeword sequence u j =(u j,0 ,u j,1 ,…,u j,k ,…u j,K-1 ), which can be specifically expressed as
[0010]
[0011] The finite field and pattern block FFSP corresponding to each user is: Where d[k] is the set of the k-th bits of data sent by J users, is the EA code generation matrix corresponding to d[k];
[0012] USPM constraint: For a finite field GF(p m ) is used to support the EA code of the limited area multiple access system of J users Φ={C1,C2,…,C J}, J≤m, d[k] and FFSP block ω k When is a one-to-one mapping, This EA code is the only translatable EA, namely the UD-EA code;
[0013] The EA code of the limited area multiple access system supporting J users is constructed according to the USPM constraint.
[0014] Further, the process of constructing an EA code for a limited area multiple access system supporting J users according to satisfying the USPM constraint includes:
[0015] EA code is Form and satisfaction The EA code is the double codeword EA code, and the double codeword EA code is expressed as The double codeword EA code that meets the USPM constraint is the double codeword UD-EA code;
[0016] use Respectively represent the code words corresponding to the data; if Where p is an m-tuple whose elements are all p, then the code is called a generalized AIEP code. If the multi-user code formed by the generalized AIEP code is uniquely decodable, it is a generalized UD-AIEP code. The generator matrices corresponding to the generalized UD-AIEP code with all 1s and all 2s are respectively recorded as and satisfy P is a matrix of size J×m with p elements;
[0017] In GF(3 m ) is used to form a dual-codeword UD-EA code, and the generator matrix G on GF(3) lc The corresponding generalized additive inverse matrix is is the equal sign that limits the element domain calculation, indicating that the value in the corresponding element domain is obtained after calculation; the generator matrix of the dual codeword UD-EA code is designed as Then a double codeword UD-EA code is constructed.
[0018] or,
[0019] The process of constructing the EA code of the limited area multiple access system supporting J users according to the USPM constraint includes:
[0020] EA code is Form and satisfaction The EA code is the double codeword EA code, and the double codeword EA code is expressed as The double codeword EA code that meets the USPM constraint is the double codeword UD-EA code;
[0021] In the finite field GF(p m ) there exists a J-dimensional subspace, and a basis set of the subspace is Decompose this basis set into N d Subset Each subset is and The number of elements in each subset is the same and the intersection is 0. Based on these subsets, N d The independent code rate is J / (N d ·m) channel coding; when N d=2, that is, decomposing the basis set into two subsets to construct two channel codes C lc,1 and C lc,2 , C lc,1 The generator matrix of is the generator matrix corresponding to all 1s, C lc,2 The generator matrix of is used as the generator matrix corresponding to all 2, and then a dual codeword UD-EA code for use by J / 2 users is constructed.
[0022] A decoding method for a P-element group EA code, comprising:
[0023] A double codeword UD-EA code constructed by a construction method of a P-element group EA code has
[0024]
[0025] FFSP Block
[0026] At the sending end, the data block d[k] is processed by a generator matrix corresponding to all 1s, and the receiving end solves and recovers the data sequence according to the generator matrix corresponding to all 1s.
[0027] or,
[0028] For a dual codeword UD-EA code constructed by a construction method of a P-element group EA code, the parallel generation matrix of the dual codeword EA code is defined as
[0029]
[0030] All 1s generator matrix and the all-2 generator matrix They are located in the upper and lower segments of the parallel matrix, respectively, so that a matrix of size 2J′×m is obtained, where J′ represents the number of supported users, corresponding to J and J / 2 in a construction method of a P-element group EA code;
[0031] Since the length of the data block d[k] is 1×J′, the mapping function is used to convert d[k] into a 1×2J′ vector to adapt the parallel generation matrix and define the parallel data block. in That is, the ternary element d in d[k] j,k Unique mapping to binary vector The mapping function is written as
[0032]
[0033] The binary vector represents the corresponding generator matrix, which is used to select G M,pll First half or lower half The FFSP block is represented as
[0034] ω k =a[k]·G M,pll
[0035] Solve the parallel data block a[k] according to the parallel generation matrix, and then inversely map it back to ternary data.
[0036] A method for constructing a P-element group EA code, comprising:
[0037] Assume that α is GF(p m ) of the primitive element, GF(p m ) can be uniquely represented by a vector of length m on GF(p), that is, represented by an m-tuple, and the primitive element α can represent the element group EA code;
[0038] Construct a p-ary EA code for p-ary system, where p is a prime number; the codeword of the p-ary EA code is expressed as:
[0039]
[0040] First, reduce the p-ary dimension to 3-ary: For p-ary transmission, let d j,k is the p-ary data of the j-th user, where p is a prime number greater than 3; through the p-ary to ternary transformation function F p2T Convert to 3-ary form,
[0041]
[0042] Where N d It is the length of a multi-base symbol after dimensionality reduction to base 3. The subscript T indicates base 3. is the round-up function; then the converted parallel data block is
[0043]
[0044] a[k] is a string of length J·N d The ternary vector of ;
[0045] In the finite field GF(p m ) there exists a J-dimensional subspace, and a basis set of the subspace is Decompose this basis set into N d Subset Each subset is and The number of elements in each subset is the same and the intersection is 0. Based on these subsets, N d The independent code rate is J / (N d m) channel coding;
[0046] Take a[k] as the basis set Design a J.N d Dimensional base set Decompose into N d subsets, each of which is J-dimensional, corresponding to N d The parallel generation matrix is Then the P-element group EA code is designed.
[0047] A decoding method for a P-element group EA code, for a P-element group EA code constructed by the aforementioned method for constructing a P-element group EA code, the FFSP block of the encoded parallel data block is represented as ω k =a[k]·G M,pll At this time, the data are all on GF(3). The decoding method of the P-ary element group EA code described in claim 5 is used to decode the 3-ary data to restore the parallel data block a[k], and then the dimension is upgraded to multi-ary to restore the multi-ary symbol sent by the user.
[0048] A method for constructing a P-element group EA code, comprising:
[0049] Assume that α is GF(p m ) of the primitive element, GF(p m ) can be uniquely represented by a vector of length m on GF(p), that is, represented by an m-tuple, and the primitive element α can represent the element group EA code;
[0050] Construct a p-ary EA code for p-ary system, where p is a prime number; the codeword of the p-ary EA code is expressed as:
[0051]
[0052] First, reduce the p-ary dimension to binary: For p-ary transmission, let d j,k is the p-ary data of the j-th user, where p is a prime number greater than 3; through the p-ary to binary transformation function F p2D Convert to binary form, that is
[0053]
[0054] Where N d It is the length of a multi-base symbol after dimensionality reduction to binary. The subscript D indicates binary. is the round-up function; then the converted parallel data block is
[0055]
[0056] a[k] is a string of length J·N d A binary vector of ;
[0057] According to the binary vector, the source EP code design method is used to construct it.
[0058] A decoding method for a P-ary element group EA code. For a P-ary element group EA code constructed by the aforementioned construction method for a P-ary element group EA code, after encoding, the data are all on GF(2). The binary data is decoded using the GF(2) decoding method, and then the dimension is increased to multi-ary to recover the multi-ary symbols sent by the user.
[0059] A method for constructing an EA code of a P-element group is provided. For a finite field non-orthogonal multiple access scheme, namely, for FF-NOMA, the number of users carried is J>m, and the EA code used to support FF-NOMA is called NO-EA code. The construction process of NO-EA code includes the following steps:
[0060] Assume that α is GF(p m ) of the primitive element, GF(p m ) can be uniquely represented by a vector of length m on GF(p), that is, represented by an m-tuple, and the primitive element α can represent the element group EA code;
[0061] Construct a p-ary NO-EA code for p-ary system, where p is a prime number; the codeword of the p-ary NO-EA code is expressed as
[0062] The generator matrix of NO-EA code is expressed as:
[0063]
[0064] in, They represent the corresponding generator matrices of the multi-ary system, Elements in Then the NO-EA code used for user J is “no” means non-orthogonal, where each user’s EA codeword is in the form of a dual-codeword EA code; NO-EA is designed in the complex domain of mapping; the corresponding mapping function is
[0065]
[0066] In the formula, F F2C Represents the corresponding function mapped to the complex domain, is the form of NO-EA codeword corresponding to the complex field; since the codeword is obtained by sending symbol mapping, it can be further written as
[0067]
[0068] Among them, F p2q Indicates converting the user's data into GF(3m ) on the EA code conversion function;
[0069] Based on the corresponding mapping relationship between the above-mentioned transmitted symbols and the complex domain signal, its complex domain and pattern block CFSP are defined as follows:
[0070] Complex domain USPM constraint: For a p-element source, the transmitted digital block must be uniquely decoded, and the one-to-one mapping between the symbol block d[k] sent by the user and the complex domain and mode CFSP block r[k] is 0≤k<K, in this case, the transmitted symbol block has a unique and pattern-mapped structural characteristic in the complex domain;
[0071] Based on the structural characteristics that the transmitted symbol block has unique and pattern mapping in the complex domain, the EA code of the finite domain multiple access system supporting J users is constructed according to the complex domain USPM constraint, and then the NO-EA code is constructed using the method based on non-orthogonal EP code.
[0072] Beneficial effects:
[0073] The present invention proposes a method for distinguishing finite domain virtual resource blocks of different users by using a p-ary signal source, which can effectively improve communication transmission and provide higher spectrum efficiency. BRIEF DESCRIPTION OF THE DRAWINGS
[0074] Figure 1 Schematic diagram of the relationship between data blocks, parallel data blocks and FFSP blocks.
[0075] Figure 2 The figure is a comparison chart of different transmitter methods in different FFMA modes.
[0076] Figure 3 Figure 2 is the bit error rate performance diagram for different transmitter modes.
[0077] Figure 4 This is the Shannon limit diagram under the BB method and the TT method.
[0078] Figure 5 PUER performance diagram of different NOMA systems. DETAILED DESCRIPTION
[0079] The present invention belongs to the potential technology in the field of next generation communication technology, mainly provides a new solution for large-scale user multi-access, and provides a solution for multi-access communication of multi-level information sources. The specific technology involves GF(p m), the construction and decoding methods of EA codewords (codeword EA, CWEA) based on AIEP and basis decomposition, and the finite-field multiple-access (FFMA) transmission scheme for multi-base information sources. The main work of the present invention is to design FFMA coding and transmission schemes for communication systems with p-base information sources, which are mainly suitable for short packet communication between massive users, ultra-large-scale connection, multi-user random access, information transmission of multi-base communication systems and other related scenarios. Specific implementation method one:
[0081] This embodiment is a method for constructing a P-element group EA code, including:
[0082] Element Group (EA) Code:
[0083] Assume that α is GF(p m ) is the primitive element, then α -∞ =0, Can express GF(p m ), GF(p m ) on each element α j can be expressed as α 0 =1,…,α m-1 A linear combination of:
[0084] α j =a j,0 +a j,1 α+a j,2 α 2 +…+a j,m-1 α m-1
[0085] Where j = -∞, 0,…, p m -2 and coefficient a j,0 、a j,1 , …, a j,m-1 is an element of GF(p);
[0086] From the above formula, we can see that α j It can be obtained by the vector (a) of length m on GF(p) j,0 ,a j,1 ,...,a j,m-1 ) only indicates.
[0087] For a p-ary signal source, there are p possible information bits, namely 0, 1, 2, ..., p-1. In this case, the element group EA code can be expressed as
[0088]
[0089] Among them, 0≤l j,0 ,l j,1 ,...,l j,p-1 ≤p m -2, Subscript j represents the jth EA code, and subscripts 0, 1, ..., p-1 correspond to the information bits of the multi-bit source. For a user, the p codewords in its EA code are different, corresponding to p source symbols, and the receiving end can determine the sent data based on the received codewords. However, when the EA code is used in a multi-user system, the codewords of different users will interfere with each other, so it is necessary to consider how to avoid interference between different users and distinguish different information sent by different users based on the received codewords.
[0090] The EA code of J element groups designed for J users can be expressed as:
[0091]
[0092] The Cartesian product of the EA code of the element group is expressed as This can be done in GF(p m ) to form an EA code for J users. These code words need to meet certain conditions so that they can be uniquely deciphered by the receiving end.
[0093] This implementation manner is first described from the EA code construction of a ternary signal source, that is, the case of p=3, and then extended to the p-ary system.
[0094] (I) EA code construction and decoding of ternary source
[0095] When p=3, the possible output of the source is C T =(0,1,2), the corresponding EA code is The codeword corresponding to each information bit can be represented by an m-tuple. Then when all J user information bits are 0, a generator matrix corresponding to all 0s can be formed, which is recorded as
[0096]
[0097] Similarly, is the generator matrix corresponding to all 1s, is the generator matrix corresponding to all 2. Unless otherwise specified, EA codes refer to the 3-ary source case.
[0098] (I) Concept of double CWEA code:
[0099] For a ternary signal source (ternary source), since there are three possible symbols of the ternary signal source, designing a corresponding codeword for each symbol will increase the complexity of codeword design and decoding. In order to reduce the complexity, this embodiment proposes a dual-codeword EA code based on codeword construction, hereinafter referred to as a dual-codeword EA code. In GF(3 m ) on the double codeword EA code satisfy That is, the double codeword EA code can be expressed as The form of , where 0 is a tuple of length m, and the subscript cw represents the codeword. Then, the Cartesian product The dual codeword EA code of J users is formed. Since the codeword corresponding to the dual codeword EA code is 0 when the information bit is 0, the corresponding generator matrix (referred to as the all-0 generator matrix, the same below) when the data sent by J users is all 0 is a 0 matrix of size J×m, that is, It is only necessary to design the generator matrix of all 1s and all 2s, which reduces the difficulty in designing codewords.
[0100] For example, to find the value of GF(3 4 ) constructs a dual codeword EA code for 4 users We only need to design the generator matrices of all 1 and all 2 respectively:
[0101]
[0102] Then the corresponding 4-user EA code can be obtained as
[0103]
[0104] It can be found that the code words corresponding to different users are different, but when the transmitted data is 0, the corresponding code words are all 0, which reduces the difficulty of design. In order to ensure that the received code words are uniquely decodable, the conditions that need to be met by the generation matrix will be introduced later.
[0105] In multi-user coding, the orthogonal dual-codeword EA code is a special case of the dual-codeword EA code. Its all-0 generator matrix is an m×m zero matrix, and the all-1 and all-2 generator matrices are
[0106]
[0107] The codewords of different users of this EA code are orthogonal in the time domain, and different users are essentially distinguished by time slots. Since the number of users supported by the system is m and the finite field code element length used is m, its loading factor is 1. The loading factor refers to the ratio of the number of users to the number of occupied resources. The higher this value is, the higher the resource utilization of multi-user coding is.
[0108] However, orthogonal EA code is only a special case. The following introduces the design method of J user EA code for universal cases.
[0109] (II) Construction of double codeword EA code:
[0110] For a ternary source (ternary source), assuming there are J users, each user sends K data, then the transmission sequence of the jth user can be expressed as d j =(d j,0 ,d j,1 ,...,d j,k ,…d j,K-1 ), where d j,k ∈{0,1,2}, 1≤j≤J, 0≤k≤K-1. The user’s data is transformed through the function F p2q Converted to GF(3 m ) on the EA codeword sequence u j =(u j,0 ,u j,1 ,...,u j,k ,...u j,K-1 ), which can be specifically expressed as
[0111]
[0112] Among them, d j,k ⊙C j Indicates that according to different d j,k The value of is used to select different code words.
[0113] Assuming that the reuse index of each user in the system is 1, that is, the weight of each user's information is the same, the finite-field sum-pattern (FFSP) is obtained as follows:
[0114]
[0115] Where d[k] is the set of the kth bit of data sent by J users, that is, d[k] = (d 1,k ,d 2,k ,...,d J,k ), is the EA code generation matrix corresponding to d[k]. Therefore, J users each send K data bits, and the multiplexing result is a FFSP block sequence w=(ω1,ω2,...,ω K-1 ). What the receiving end needs to do is to determine the transmission data d[k] based on the received FFSP block. Therefore, when designing the J user EA code, it is necessary to make sure that the FFSP blocks corresponding to different transmission data are different.
[0116] This condition can be summarized as the USPM constraint.
[0117] USPM constraint: For a finite field GF(p m ) used to support J users of the FFMA system EA code Φ = {C1, C2, ..., C J}, J≤m, to get FFSP block ω k is the only translatable one, then we need to input d[k] and FFSP block ω k is a one-to-one mapping, then the generator matrix corresponding to d[k] is It should be of full row rank, that is The EA codeword constructed according to this constraint can be used in the FFMA system. This EA code is called a uniquely-decodable (UD) EA, namely, a UD-EA code.
[0118] There are many situations for the codeword generator matrix, and it only needs to satisfy the USPM constraint. The dual-codeword EA code that satisfies the USPM constraint is the dual-codeword UD-EA code. This embodiment provides two dual-codeword EA code construction methods, namely, a construction method based on additive inverse EP (AIEP) code and a construction method based on basis decomposition.
[0119] (A) Coding based on additive inverse EP (AIEP) codes:
[0120] The concept of additive inverse element pair (AIEP) code is mentioned in the FFMA related patents. Its structure is as follows: Represent the code words corresponding to data 0 and 1 respectively. If Where p is an m-tuple whose elements are all p, then the code is called a generalized AIEP code. If the multi-user code formed by the generalized AIEP code is uniquely decodable, it is a generalized UD-AIEP code. Its all-one and all-two generator matrices are and satisfy P is a matrix of size J×m, with each element being p.
[0121] In GF(3 m ) can form a dual-codeword UD-EA code. In this case, the all-1 generator matrix of the dual-codeword UD-EA code is the same as the all-1 generator matrix of the generalized UD-AIEP code, and the all-2 generator matrix is the same as the all-0 generator matrix of the generalized UD-AIEP code, that is, The dual codeword EA code formed in this way satisfies the USPM constraints and is therefore uniquely translatable.
[0122] According to this principle, it is also possible to combine it with channel coding to construct a dual codeword UD-EA code. lc is a linear block code constructed on GF(2), then its generator matrix Glc The size is J×m. By expanding the elements to GF(3), we can get the generalized additive inverse matrix of the generator matrix: Then the generator matrix of the dual codeword UD-EA code is The equal sign is used to limit the calculation of the element domain, indicating that the value in the corresponding element domain is obtained after calculation. For example, if the element is 2, the value obtained by multiplying by 2 should be 4, but it is 1 in GF(3).
[0123] (B) Coding based on basis decomposition:
[0124] The coding method based on AIEP will be affected by AIEP coding. Here is a more general coding method, namely basis decomposition method. m ) there exists a J-dimensional subspace, and a basis set of the subspace is Decompose this basis set into N d Subset Each subset is
[0125] and The number of elements in each subset is the same and the intersection is 0. Based on these subsets, N d The independent code rate is J / (N d ·m) channel coding. d The value of can be chosen arbitrarily, mainly depending on the actual needs. For example, when N d =2, the basis set can be decomposed into two subsets to construct two channel codes C lc,1 and C lc,2 , C lc,1 The generator matrix is the all-1 generator matrix, C lc,2 The generator matrix of is used as the full 2 generator matrix, and the dual codeword UD-EA code for J / 2 users is constructed. The code rate of the code constructed by this method is low, but it can solve the problem of limited block length in the transmission process. At the same time, the number of generator matrices constructed by this method is N d It is easy to generalize to polyradical by choosing any one.
[0126] (III) Decoding method of ternary EA code:
[0127] (1) In general, the decoding scheme is determined by its generator matrix, each of which corresponds to a received FFSP block, i.e. Based on this one-to-one correspondence, we can find the corresponding generator matrix, and then determine d[k] based on the generator matrix. However, for EA codes, if there are J users and each user has 3 possible data to send, then the number of generator matrices is 3. JWhen the number of users is large, from 3 J Finding the generator matrix corresponding to the received data from the generator matrix will make the decoding complexity very high. This section gives some methods to reduce the decoding complexity.
[0128] If the dual codeword UD-EA code is constructed by the AIEP-based method, according to the analysis in the previous method (A), we have Then there is
[0129]
[0130] It can be found that the switching function d j,k ⊙C j It becomes a scalar multiplication of vectors. So its FFSP block becomes
[0131]
[0132] At the transmitting end, the data block d[k] is processed by an all-1 generator matrix, so the receiving end only needs to solve it according to the all-1 generator matrix. This solution can restore the data sequence using any ternary decoding algorithm. The simplest method is to find the inverse of the all-1 generator matrix.
[0133] (2) Next, we introduce another method to reduce the complexity, which is called the parallel generation matrix method. The parallel generation matrix of the dual codeword EA code is defined as
[0134]
[0135] All 1s generator matrix and the all-2 generator matrix They are located in the upper and lower segments of the parallel matrix respectively, thus obtaining a matrix of size 2J×m. It should be noted that: J here represents the number of supported users. When this method is applied to the decoding of the aforementioned (A) coding based on additive inverse EP (AIEP) code, since (A) supports J users, the matrix here is 2J×m. When it is applied to the decoding of the coding based on basis decomposition (B), since (B) supports J / 2 users, the matrix here is J×m. In other words, the above (B) is expressed as the number of supported J users, and it is still expressed as a 2J×m matrix here. In fact, J is not limited to the limitation of the representation form, as long as it represents the number of supported J users.
[0136] Since the length of the data block d[k] is 1×J, it needs to be converted into a 1×2J vector through a mapping function to adapt the parallel generation matrix, which is equivalent to mapping 1 number in the original data block to 2 numbers. Since the elements in d[k] are ternary, the parallel data block is defined as in That is, the ternary element d in d[k] j,k Unique mapping to binary vector The mapping function can be written as
[0137]
[0138] The binary vector is equivalent to selecting the corresponding generator matrix. For example, when a user sends 1, it is mapped to (1, 0). The corresponding value is 1, so select G M,pll First half The codeword of the corresponding position of the user is used as the codeword of the user, and the lower part The corresponding value is 0, so the generator matrix of the lower half is not selected. Overall, the codeword of the generator matrix of all 1 is still used. The FFSP block of this method can be expressed as
[0139] ω k =a[k]·G M,pll
[0140] According to the parallel generator matrix, the parallel data block a[k] can be solved, and then inversely mapped back to ternary data. This method can be used for methods based on AIEP and basis decomposition, but since the parallel generator matrix constructed based on the AIEP method is not a row full rank matrix, this method is particularly suitable for decoding dual codeword EA codes constructed based on basis decomposition.
[0141] (II) Construction and decoding of EA codes for p-ary sources
[0142] (I) Construction of p-ary EA code:
[0143] The previous content gives the design method of EA code when the source is 3-ary. Now it is extended to p-ary, where p is a prime number, to construct p-ary EA code. The codeword of p-ary EA code can be expressed as:
[0144]
[0145] However, when the value of p is large, the decoding complexity is very high. In order to reduce the complexity, we can design a low-dimensional parallel generation matrix like the previous basis decomposition. This requires reducing the multi-base to binary or ternary. This implementation mainly introduces the reduction to ternary, and the reduction to binary is the same method.
[0146] For p-element transmission, let d j,k is the p-ary data of the jth user, where p is a prime number greater than 3. Then, it is transformed into its ternary form through the p-ary to ternary transformation function, that is,
[0147]
[0148] Where N d It is the length of a multi-base symbol after dimensionality reduction to base 3. The subscript T indicates base 3. When the dimension is reduced to binary, the logarithm base is 2. is a round-up function. Then the converted parallel data block is
[0149]
[0150] a[k] is a string of length J·N d The ternary vector of .
[0151] According to the basis decomposition method, a J·N d Dimensional base set Can be decomposed into N d subsets, each of which is J-dimensional, corresponding to N d generator matrices. Then, the parallel generator matrix is
[0152]
[0153] When the dimension is reduced to binary, a[k] is a length of J·N d The binary vector can be constructed by using the source EP code design method in the prior art. The present invention will not be described in detail. It needs to be additionally explained that the binary design can also be constructed by the basis decomposition method. When the corresponding decoding is performed, the binary source EP code decoding of the prior art can be used, and then the multi-base symbol sent by the user can be restored by upgrading the dimension to multi-base.
[0154] (II) Decoding of p-ary EA code:
[0155] Then, the FFSP block of the encoded parallel data block can be expressed as ω k =a[k]·G M,pll At this time, the data are all on GF(3). The parallel data block a[k] can be recovered by using the 3-ary decoding algorithm, and then the multi-ary symbols sent by the user can be recovered by upgrading the dimension to multi-ary.
[0156] The multi-base EA code based on AIEP coding cannot guarantee unique decipherability. To ensure unique decipherability, it is necessary to design AIEP code in a higher dimension, which brings more coding complexity and resource waste.
[0157] (III) EA codes of FFMA systems under different transmission modes
[0158] In the related invention of EP code, four FFMA modes are introduced under the condition of binary source, namely FF-TDMA, FF-CDMA, FF-CCMA, and FF-NOMA.
[0159] FFMA (finite field multiple access) refers to limited field multiple access.
[0160] FF-TDMA (time multiple access in finite-field) refers to finite-field time division multiple access technology.
[0161] FF-CDMA (code division multiple access in finite-field) refers to finite-field code division multiple access technology.
[0162] FF-CCMA (channel codeword multiple access in finite-field) refers to finite-field channel coding multiple access technology.
[0163] FF-NOMA (non-orthogonal multiple access in finite-field) refers to finite field non-orthogonal multiple access technology.
[0164] It should be noted that, in the present invention, the elements of the EA codeword corresponding to the ternary source are not necessarily ternary, which is different from the elements of the codeword mentioned in the previous patent application, and the corresponding modes such as FF-CDMA will be quite different.
[0165] The present invention generalizes this to the case of a ternary signal source.
[0166] (a) For FF-TDMA, similar to the EP code method, an orthogonal EA code is used to support this mode. The orthogonal EA code can be written as Among them C T =(0,1,2), the loading factor is also 1. Compared with FF-TDMA of EP code, the use of orthogonal EA code is similar to it, except that the transmission symbol is extended from GF(2) to GF(3). The principle is to send data of different users through different time slots and distinguish user data by time slots.
[0167] (b) For FF-CCMA, the dual codeword UD-EA code can be constructed by using the AIEP-based or basis decomposition-based methods described above. This code can be used in the FF-CCMA mode. It is worth noting that although the information bits generated by the source are on GF(3), the constructed generator matrix can still be on GF(2 m ), which is equivalent to mapping the information bits on GF(3) to GF(2 m). Users in this mode are distinguished by EA codes constructed based on channel coding.
[0168] (c) For FF-CDMA, the use of EA code will have different phenomena. As mentioned above, the generalized UD-EP code can form a double codeword UD-EA code. For the FF-CDMA mode of binary source transmission, the generalized UD-EP code is constructed on GF(3 m ), the number of code words is 2 m , less than 3 m , so there is a spread spectrum gain, but for the ternary signal source, the number of codewords required is also 3 m , there is no spreading gain. Therefore, if the double codeword UD-EA code is constructed in GF(3 m ), then the FF-CDMA mode does not exist, and we need to find a higher-dimensional double codeword UD-EA code, such as GF(5 m ). This mode uses EA code as the spreading code to distinguish users.
[0169] (d) For FF-NOMA, the number of users that this mode can carry is J>m, and the EA code used to support this mode is called NO-EA code. Obviously, the generator matrix of NO-EA code does not satisfy the USPM constraint. Therefore, its generator matrix should be designed to satisfy the alternative constraint. The generator matrix of NO-EA code can be expressed as:
[0170]
[0171] Among them, taking the ternary system as an example, The superscript can be 1 or 2, representing the matrix generated by all 1s and all 2s, respectively. Then the NO-EA code used for user J is "no" means non-orthogonal, where each user's EA codeword is in the form of a dual-codeword EA code In fact, multi-base is also possible. We just need to expand the ternary system to multi-base. It only needs to satisfy the USPM constraints of the complex domain. There is no other change, only the number of corresponding generating matrices increases.
[0172] Since NO-EA does not satisfy the USPM constraint, to make it uniquely translatable, it needs to be designed on the complex domain of the mapping.
[0173] The corresponding mapping function is
[0174]
[0175] In the formula is the form of NO-EA codeword corresponding to the complex field. Since the codeword is obtained by sending symbol mapping, it can be further written as
[0176]
[0177] It indicates that there is a corresponding mapping relationship between the transmitted symbol and the signal in the complex domain. Then define its complex domain and mode block
[0178] (CFSP) is
[0179] If the transmitted symbol block d[k] can have a one-to-one correspondence with the CFSP, then it can also be uniquely deciphered at the receiving end. The following is a universal theorem:
[0180] Complex domain USPM constraint: For a p-element source, to uniquely decode the transmitted digital block, it should be a one-to-one mapping between the symbol block d[k] sent by the user and the complex domain sum pattern (CFSP) block r[k], that is, 0≤k<K, which means that the transmitted symbol block has a unique and pattern-mapped structural characteristic in the complex domain.
[0181] For example, the NO-EA code is constructed using a method based on non-orthogonal EP codes, and its all-1 generator matrix is Specifically, it can be written as:
[0182]
[0183] In the formula, p is used to replace 0 as an adjustment code to assist decoding. p can be set to 3. It can be found that the generated matrix does not satisfy the USPM constraint. Therefore, when modulating, this p is mapped to a special point in the complex domain to ensure that the mapping result of the NO-EA codeword satisfies the complex domain USPM constraint. This specific value can be flexibly designed to pursue improved demodulation and decoding performance.
[0184] This constraint is applicable to p-dimensional information sources. For higher-dimensional information sources, we only need to ensure that they satisfy this constraint and then design their generator matrices accordingly.
[0185] Example
[0186] Embodiment 1: Constructing a dual codeword EA code based on AIEP
[0187] First, a generalized AIEP code for two users is designed based on the ternary orthogonal matrix T o (2,2) to generate the generalized AIEP code, and then according to the previous method, let the all-1 generator matrix of the dual codeword EA code be T o (2,2), the all-2 generating matrix is its additive inverse, then
[0188]
[0189] There are 9 types of FFSP blocks obtained according to this encoding, namely 00(0,0) (the data in brackets are sent by two users, the same below), 11(1,0), 21(0,1), 22(2,0), 12(0,2), 02(1,1), 20(1,2), 10(2,1), 01(2,2). The loading factor of this EA code is 1.
[0190] As mentioned above, linear block codes can be combined with EA codes to construct codewords. Such codewords are used in FFMA-CCMA and can take advantage of the performance gain brought by local channel codes. Suppose there is a (7, 4) linear cyclic block code, whose generating polynomial is g(X) = 1 + X 2 +X 3 , then its all-1 and all-2 generating matrices can be expressed as
[0191]
[0192] There are also 9 types of FFSP blocks, namely 0000000 (0, 0), 1011000 (1, 0), 0101110 (0, 1), 2022000 (2, 0), 0202200 (0, 2), 1112100 (1, 1), 1210200 (1, 2), 2120100 (2, 1), 2221200 (2, 2). The transmitter modulates and sends according to the FFSP block, and the receiver decodes according to the one-to-one correspondence between the FFSP block and the transmitted data after demodulation. The loading factor of the EA code is 2 / 7.
[0193] Embodiment 2: Constructing a dual codeword EA code based on the basis decomposition method
[0194] Same as Example 1, assume that there is a generating polynomial g(X)=1+X 2 +X 3 The (7, 4) linear cyclic block code, then its basis set It can be divided into two subsets, namely According to the basis decomposition method, the generator matrix of the corresponding double codeword EA code can be obtained as
[0195]
[0196] Different from the AIEP-based method, the EA code in this case has no domain restriction, that is, the codeword can be on GF(2) or GF(3), and the two cases correspond to different FFSP blocks.
[0197] If FFSP is calculated on GF(2), the possible values of the resulting FFSP block are 0000000(0,0), 1011000(1,0), 0101100(0,1), 1110100(1,1), 0010110(2,0), 0001011(0,2), 1010011(1,2), 0111010(2,1), 0011101(2,2).
[0198] If FFSP is calculated over GF(3), the possible values of the FFSP block are 0000000(0,0), 1011000(1,0), 0101100(0,1), 1112100(1,1), 0010110(2,0), 0001011(0,2), 1012011(1,2), 0111210(2,1), 0011121(2,2). The loading factor of the EA code is 2 / 7, which is half the code rate of the linear cyclic block code.
[0199] Another interesting phenomenon is that the above generator matrix can also be constructed by combining the additive inverse method, for example, using the additive inverse matrix of the above all-2 generator matrix to replace the matrix, that is,
[0200]
[0201] Then the possible cases of its FFSP block are 0000000(0,0), 1011000(1,0), 0101100(0,1), 1112100(1,1), 0020220(2,0), 0002022(0,2), 1010022(1,2), 0121020(2,1), 0022212(2,2).
[0202] Therefore, the method based on basis decomposition has more optional types of generated matrices and is more flexible.
[0203] Example 3: Application of Parallel Matrix Decoding Method
[0204] Assuming there are two users in the FFMA system and the information source is ternary, the data sent by the two users may be d[k]∈{00,01,02,10,11,12,20,21,22}. For the two-user dual codeword EA code, when the parallel matrix method is used for decoding, the parallel data block corresponding to the sent data is
[0205] a[k]={0000,0100,0001,1000,1100,1001,0010,0110,0011}
[0206] The first and third bits in the corresponding quaternary codeword represent the data sent by the first user, and the second and fourth bits represent the data sent by the second user.
[0207] Using the EA code construction method using the AIEP generation matrix in Example 1, the parallel generation matrix of the dual codeword EA code is
[0208]
[0209] The rank of this generator matrix is 2, and its FFSP block ω k =a[k]·G M,pll The types are 00, 21, 12, 11, 02, 20, 22, 10, and 01. It can be found that the results are consistent with those given in Example 1, which shows the feasibility of this method.
[0210] If this method is applied to the generator matrix of another dual-codeword EA code in Example 1, the parallel generator matrix can be obtained as
[0211]
[0212] The rank of this generator matrix is also 2. It is not a full-rank matrix. It is an equivalent generator matrix based on the user's sent data, which is convenient for decoding. In this way, the generator matrix of the dual-codeword UD-EA code can be equivalent to a parallel matrix, reducing the complexity of decoding. The output codewords, i.e., FFSP blocks, have 9 possibilities: 0000000, 0101100, 0202200, 1011000, 1112100, 1210200, 2022000, 2120100, and 2221200.
[0213] The dual codeword EA code constructed according to the basis decomposition method is more suitable for this decoding method because its parallel generation matrix is
[0214]
[0215] The rank of this matrix is 4, which is a row full rank matrix. Such a matrix has more advantages in decoding and is applicable to more methods. According to the method of parallel matrix generation, it is easy to get that its FFSP block on GF(2) is 0000000, 0101100, 0001011, 1011000, 1110100, 1010011, 0010110, 0111010, 0011101; the FFSP block on GF(3) is 0000000, 0101100, 0001011, 1011000, 1112100, 1012011, 0010110, 0111210, 0011121.
[0216] It can be seen that this method can be used for decoding EA codes, thereby reducing the decoding complexity. Figure 1 The relationship between user data and the parallel generation matrices of different coding methods is summarized in .
[0217] Embodiment 4: Design method of EA code for multi-base information source transmission system
[0218] Assume that the source is 5-base, and a 5-base EA code for two users is to be constructed. Since p = 5, the converted It means that each 5-base symbol can be decomposed into two 3-base symbols, namely So the conversion function is expressed as
[0219]
[0220] Since the number of users is 2, we can list all possible combinations of the input digital block d[k] and its corresponding parallel data block a[k] as
[0221]
[0222] The parallel generation matrix can be expressed as
[0223]
[0224] There are 5 2 There are three FFSP blocks: 0000000, 0001011, 0002022, 0101100, 0102111, 0010110, 0011121, 0012102, 0111210, 0112221, 0020220, 0021201, 0022212, 0121020, 0122001, 1011000, 1012011, 1010022, 1112100, 1110111, 1021110, 1022121, 1020102, 1122210, 1120221.
[0225] Example 5: FFMA uplink system performance simulation and analysis
[0226] First, let the loading factors of the four modes of FFMA, namely FF-TDMA, FF-CDMA, FF-CCMA, and FF-NOMA, be The influence of global channel coding is not considered (its code rate is set to 1). There are three transmission modes for FFMA transmitters, namely binary source BPSK signal (BB), binary source 3ASK signal (BT) and ternary source 3ASK signal. The spectral efficiency SE, spatial utilization SUR and Shannon limit of the three modes are listed in Figure 2 .
[0227] The BB mode only supports two FFMA modes, namely FF-TDMA and FF-CCMA. In order to expand more modes, 3ASK signal, namely BT mode, is used. This mode uses a binary to ternary conversion function to convert binary data into ternary data, and then uses 3ASK signal for modulation. This mode supports four FFMA modes. The TT mode supports FF-TDMA, FF-CCMA and FF-NOMA. As FF-CDMA has been analyzed above, it has no spread spectrum gain, so it is similar to FF-TDMA.
[0228] Then, the bit error rate performance of several modes is obtained through simulation, such as Figure 3 As shown in the figure, for the BB mode and the TT mode, the SER deteriorates as the number of users J increases, but the difference in SER is small when J = 300 and J = 2. The SER of the BB mode is close to the SER of the BPSK (or QPSK) signal, and the SER of the TT mode is close to the SER of the 3ASK (or 9QAM) signal.
[0229] The relationship between the number of users J and the Shannon limit is placed Figure 4 In the figure, the expansion factor m = 100. For a given number of users J, we find that the TT method requires a larger Eb / N0 than the BB method because the TT method can provide a higher SE than the BB method. In addition, the Shannon limit of both classical TDMA and power domain (PD) NOMA is on the curve of the BB method, corresponding to LF = 1 and SE = 2 in FF-TDMA, and LF = 2, SE = 4.
[0230] For FF-NOMA, the simulation diagram of the per-user error probability (PUER) is shown in Figure 5 For FF-NOMA in TT mode, the adjustment code in the complex domain is set to It can be found that for the FF-NOMA system, the performance of the adjustment code set to i is better than that of The performance is better when the power ratio is 1:4. Therefore, using adjustment code i and power ratio 1:4 may be a good choice for FF-NOMA.
[0231] The above calculation examples of the present invention are only used to explain the calculation model and calculation process of the present invention in detail, and are not intended to limit the implementation methods of the present invention. For ordinary technicians in the relevant field, other different forms of changes or modifications can be made based on the above description. It is impossible to list all the implementation methods here. All obvious changes or modifications derived from the technical solution of the present invention are still within the scope of protection of the present invention.
Claims
1. A method for constructing a P-element group EA code, It is characterized in that include: Assume that α is GF(p m ) of the primitive element, GF(p m ) can be uniquely represented by a vector of length m on GF(p), that is, represented by an m-tuple, and the primitive element α can represent the element group EA code; For a ternary source, when p = 3, the possible output of the source is C T =(0,1,2), when the J user information is all 0, 1, 2, the corresponding generating matrices are formed Each of the J users sends K data, and the sending sequence of the jth user is expressed as d j =(d j,0 ,d j,1 ,...,d j,k ,…d j,K-1 ), where d j,k ∈{0,1,2}, 1≤j≤J, 0≤k≤K-1; the user's data is converted through the conversion function F p2q Converted to GF(3 m ) on the EA codeword sequence u j =(u j,0 ,u j,1 ,...,u j,k ,...u j,K-1 ), which can be specifically expressed as The finite field and pattern block FFSP corresponding to each user is: Where d[k] is the set of the k-th bits of data sent by J users, is the EA code generation matrix corresponding to d[k]; USPM constraint: For a finite field GF(p m ) used to support J users of the limited area multiple access system EA code Φ = {C 1 ,C 2 ,...,C J }, J≤m, d[k] and FFSP block ω k When is a one-to-one mapping, This EA code is the only translatable EA, namely the UD-EA code; The EA code of the limited area multiple access system supporting J users is constructed according to the USPM constraint.
2. A method for constructing a P-element group EA code according to claim 1, It is characterized in that The process of constructing the EA code of the limited area multiple access system supporting J users according to the USPM constraint includes: EA code is Form and satisfaction The EA code is the double codeword EA code, and the double codeword EA code is expressed as The double codeword EA code that meets the USPM constraint is the double codeword UD-EA code; use Respectively represent the code words corresponding to the data; if Where p is an m-tuple whose elements are all p, then the code is called a generalized AIEP code. If the multi-user code formed by the generalized AIEP code is uniquely decodable, it is a generalized UD-AIEP code. The generator matrices corresponding to the generalized UD-AIEP code with all 1s and all 2s are respectively recorded as and satisfy P is a matrix of size J×m with p elements; In GF(3 m ) is used to form a dual-codeword UD-EA code, and the generator matrix G on GF(3) lc The corresponding generalized additive inverse matrix is is the equal sign that limits the element domain calculation, indicating that the value in the corresponding element domain is obtained after calculation; the generator matrix of the dual codeword UD-EA code is designed as Then a double codeword UD-EA code is constructed.
3. The method for constructing a P-element group EA code according to claim 1, It is characterized in that The process of constructing the EA code of the limited area multiple access system supporting J users according to the USPM constraint includes: EA code is Form and satisfaction The EA code is the double codeword EA code, and the double codeword EA code is expressed as The double codeword EA code that meets the USPM constraint is the double codeword UD-EA code; In the finite field GF(p m ) there exists a J-dimensional subspace, and a basis set of the subspace is Decompose this basis set into N d Subset Each subset is and The number of elements in each subset is the same and the intersection is 0. Based on these subsets, N d The independent code rate is J / (N d ·m) channel coding; when N d =2, that is, decomposing the basis set into two subsets to construct two channel codes C lc,1 and C lc,2 , C lc,1 The generator matrix of is the generator matrix corresponding to all 1s, C lc,2 The generator matrix of is used as the generator matrix corresponding to all 2, and then a dual codeword UD-EA code for use by J / 2 users is constructed.
4. A decoding method for a P-element group EA code, It is characterized in that include: The double codeword UD-EA code constructed by the method for constructing a P-element group EA code according to claim 2 has FFSP Block At the sending end, the data block d[k] is processed by a generator matrix corresponding to all 1s, and the receiving end solves and recovers the data sequence according to the generator matrix corresponding to all 1s.
5. A decoding method for a P-element group EA code, It is characterized in that include: For the dual codeword UD-EA code constructed by the construction method of a P-element group EA code according to claim 2 or 3, the parallel generation matrix of the dual codeword EA code is defined as All 1s generator matrix and the all-2 generator matrix They are located in the upper and lower segments of the parallel matrix, respectively, thus obtaining a matrix of size 2J′×m, where J′ represents the number of supported users; Since the length of the data block d[k] is 1×J′, the mapping function is used to convert d[k] into a 1×2J′ vector to adapt the parallel generation matrix and define the parallel data block. in That is, the ternary element d in d[k] j,k Unique mapping to binary vector The mapping function is written as The binary vector represents the corresponding generator matrix, which is used to select G M,pll First half or lower half The FFSP block is represented as oh k =a[k]·G M,pll Solve the parallel data block a[k] according to the parallel generation matrix, and then inversely map it back to ternary data.
6. A method for constructing a P-element group EA code, It is characterized in that include: Assume that α is GF(p m ) of the primitive element, GF(p m ) can be uniquely represented by a vector of length m on GF(p), that is, represented by an m-tuple, and the primitive element α can represent the element group EA code; Construct a p-ary EA code for p-ary system, where p is a prime number; the codeword of the p-ary EA code is expressed as: First, reduce the p-ary dimension to 3-ary: For p-ary transmission, let d j,k is the p-ary data of the j-th user, where p is a prime number greater than 3; through the p-ary to ternary transformation function F p2T Convert to 3-ary form, Where N d It is the length of a multi-base symbol after dimensionality reduction to base 3. The subscript T indicates base 3. is the round-up function; then the converted parallel data block is a[k] is a string of length J·N d The ternary vector of ; In the finite field GF(p m ) there exists a J-dimensional subspace, and a basis set of the subspace is Decompose this basis set into N d Subset Each subset is and The number of elements in each subset is the same and the intersection is 0. Based on these subsets, N d The independent code rate is J / (N d m) channel coding; Take a[k] as the basis set Design a J.N d Dimensional base set Decompose into N d subsets, each of which is J-dimensional, corresponding to N d The parallel generation matrix is Then the P-element group EA code is designed.
7. A decoding method for a P-element group EA code, It is characterized in that For the P-element group EA code constructed by the construction method of the P-element group EA code according to claim 6, the FFSP block of the encoded parallel data block is represented by ω k =a[k]·G M,pll At this time, the data are all on GF(3). The decoding method of the P-ary element group EA code described in claim 5 is used to decode the 3-ary data to restore the parallel data block a[k], and then the dimension is upgraded to multi-ary to restore the multi-ary symbol sent by the user.
8. A method for constructing a P-element group EA code, It is characterized in that include: Assume that α is GF(p m ) of the primitive element, GF(p m ) can be uniquely represented by a vector of length m on GF(p), that is, represented by an m-tuple, and the primitive element α can represent the element group EA code; Construct a p-ary EA code for p-ary system, where p is a prime number; the codeword of the p-ary EA code is expressed as: First, reduce the p-ary dimension to binary: For p-ary transmission, let d j,k is the p-ary data of the j-th user, where p is a prime number greater than 3; through the p-ary to binary transformation function F p2D Convert to binary form, that is Where N d It is the length of a multi-base symbol after dimensionality reduction to binary. The subscript D indicates binary. is the round-up function; then the converted parallel data block is a[k] is a string of length J·N d A binary vector of ; According to the binary vector, the source EP code design method is used to construct it.
9. A method for decoding a P-element group EA code, It is characterized in that According to the P-ary element group EA code constructed by the construction method of the P-ary element group EA code described in claim 8, the data are all on GF(2) after encoding. The binary data is decoded using the GF(2) decoding method, and then the dimension is increased to multi-ary to restore the multi-ary symbols sent by the user.
10. A method for constructing a P-element group EA code, It is characterized in that For the finite field non-orthogonal multiple access scheme, i.e., for FF-NOMA, the number of carried users J>m, the EA code used to support FF-NOMA is called NO-EA code; The construction process of NO-EA code includes the following steps: Assume that α is GF(p m ) of the primitive element, GF(p m ) can be uniquely represented by a vector of length m on GF(p), that is, represented by an m-tuple, and the primitive element α can represent the element group EA code; Construct a p-ary NO-EA code for p-ary system, where p is a prime number; the codeword of the p-ary NO-EA code is expressed as The generator matrix of NO-EA code is expressed as: in, They represent the corresponding generator matrices of the multi-ary system, Elements in Then the NO-EA code used for user J is "no" means non-orthogonal, where each user's EA codeword is in the form of a dual-codeword EA code; NO-EA is designed in the complex domain of mapping; the corresponding mapping function is In the formula, F F2C Represents the corresponding function mapped to the complex domain, is the form of NO-EA codeword corresponding to the complex field; since the codeword is obtained by sending symbol mapping, it can be further written as Among them, F p2q Indicates converting the user's data into GF(3 m ) on the EA code conversion function; Based on the corresponding mapping relationship between the above-mentioned transmitted symbols and the complex domain signal, the complex domain and pattern block CFSP are defined as follows: Complex domain USPM constraint: For a p-element source, the transmitted digital block must be uniquely decoded, and the one-to-one mapping between the symbol block d[k] sent by the user and the complex domain and mode CFSP block r[k] is 0≤k<K, in this case, the transmitted symbol block has a unique and pattern-mapped structural characteristic in the complex domain; Based on the structural characteristics that the transmitted symbol block has unique and pattern mapping in the complex domain, the EA code of the limited domain multiple access system supporting J users is constructed according to the complex domain USPM constraint, and then the NO-EA code is constructed using the method based on non-orthogonal EP code.
