Signal detection method and device based on Gaussian approximation interference and parallel architecture

Abstract

The invention discloses a signal detection method and device based on Gaussian approximation interference and parallel architecture, relates to the field of digital signal processing, and solves the problem of complex calculation of a traditional E-MPA algorithm. The technical scheme is characterized in that a received signal vector is modeled to obtain a signal model; modeling interference terms in the received signal model as a Gaussian random variable distribution model, and determining to obtain Gaussian information; updating the function node to obtain a logarithm domain function node message; executing a cyclic shift gating operation according to the control signal; receiving a logarithm domain function node message, and carrying out variable node updating to obtain a normalized probability domain variable node message; the probability domain variable node message is returned for iterative updating until a preset number of times is reached; after the last iteration is completed, calculation is carried out, and the log-likelihood ratio of the bit stream is output; and the effects of reducing the calculation complexity, optimizing the in-chip storage access scheduling and greatly reducing the resource overhead are achieved.

Description

Technical Field

[0001] The present invention relates to the field of digital signal processing, and more particularly, to a signal detection method and device based on Gaussian approximate interference and parallel architecture. Background Art

[0002] Non-orthogonal multiple access (NOMA), with its overload access and scheduling-free transmission characteristics, has become a candidate multiple access technology for next-generation communication systems. Sparse code multiple access (SCMA), a form of code-domain NOMA, achieves significant shaping gain through its sparse spread spectrum characteristics. Receiver design is a key issue in SCMA systems and is crucial for meeting the large-scale access and fast transmission requirements of future communication systems.

[0003] The SCMA system performs signal detection based on the Belief Propagation (BP) criterion. The Message Passing Algorithm (MPA), as a representative example, implements Bayesian posterior probability inference through iterative updates and can approach maximum likelihood (ML) detection performance. In frequency-selective fading channels, as the number of multipath signals increases, the Message Passing Algorithm (MPA) is unable to distinguish multipath signals, which will have a fatal impact on actual communication systems. Single-carrier schemes face the technical bottleneck of severe inter-symbol interference (ISI) and multi-user interference (MUI) in frequency-selective channels. Traditional single-carrier frequency domain equalization (SC-FDE) is ineffective for single-carrier SCMA systems (SC-SCMA) because it is impossible to achieve joint equalization of multi-user channels with a single tap coefficient. To this end, the Extended Factor Graph (EFG) is used to accurately characterize the received signal composition and improve receiver signal detection accuracy. The EFG-based MPA algorithm (E-MPA) achieves good computational accuracy, but its computational complexity increases exponentially in high-profile systems or complex multipath scenarios, leading to computational disasters. Given that the high computational complexity of E-MPA exceeds practical hardware processing capabilities, the EFG-based Gaussian Approximate Interference-based Belief Propagation (E-GAIBP) algorithm can reduce the complexity from exponential to linear. However, the E-GAIBP algorithm still requires a large number of nonlinear computations, and using lookup tables (LUTs) to perform these computations consumes significant resources. Furthermore, in high-profile systems or complex multipath scenarios, the variable storage requirements of the fully parallel iterative update scheme can be significant. In summary, existing E-GAIBP receivers place high demands on hardware resources, hindering their practical application.

[0004] Therefore, there is an urgent need for a signal detection method that can solve the above technical problems. Summary of the Invention

[0005] In order to address the deficiencies in the prior art, the purpose of the present invention is to provide a signal detection method and device based on Gaussian approximate interference and parallel architecture. By modeling the interference term as a Gaussian random variable distribution, and passing messages between the function node update and the variable node update through multiple cycles of function node update and variable node update, as well as cyclic shift gating operation control, the data is converted to the logarithmic domain, the multiplication operation is converted into an accumulation operation, and finally the reliability information of the bit sent by each user is calculated, forming a partial parallel architecture solution, which reduces the exponential complexity of the traditional E-MPA algorithm based on extended factor graph message passing to the linear level, reduces the computational complexity, realizes nonlinear optimization, avoids exponential operations and lookup table (LUT) dependence, optimizes on-chip storage access scheduling, and greatly reduces resource overhead.

[0006] The above technical objectives of the present invention are achieved through the following technical solutions:

[0007] In a first aspect, a signal detection method based on Gaussian approximate interference and a parallel architecture is provided, comprising the following steps:

[0008] S1. Under the condition of frequency selective fading channel, the received signal vector is modeled to obtain a signal model;

[0009] S2. Modeling the interference term in the received signal model as a Gaussian random variable distribution model, and determining Gaussian information according to the Gaussian random variable distribution model, where the Gaussian information includes a mean value and a variance of the interference term;

[0010] S3. Update the function node according to the interference term mean, the interference term variance, the received signal vector, the channel response, and the codebook data to obtain a logarithmic domain function node message;

[0011] S4. Perform a cyclic shift gating operation according to the control signal to match the variable node information with the received signal vector, channel response, and codebook data of the functional node to calculate and update the logarithmic domain functional node information;

[0012] S5. Update the variable node according to the logarithmic domain function node message to obtain the normalized probability domain variable node message and updated Gaussian information;

[0013] S6. Return the probability domain variable node message and the updated Gaussian information to S3, and iteratively execute S3-S5 according to the received signal, channel response, and codebook data until a preset number of times is reached;

[0014] S7. After the last iteration is completed, calculation is performed based on the latest logarithmic domain function node message obtained after iterative update, and the log-likelihood ratio of the bit stream is output.

[0015] Furthermore, the calculation formula of the signal model in step S1 is:

[0016]

[0017] Among them, y j (n) is the received signal vector at the nth symbol position in the jth time slot; h j,d (p) is the channel response of the lth signal of the dth user at the nth symbol position in the jth time slot, x j,d (n-L+p) is the symbol at the n-L+pth symbol position of the dth user in the jth time slot; h j,k (n, l) represents the channel response of the lth signal of the kth user at the nth symbol position in the jth time slot, For h j,k Estimated value of (n,l); For the symbol x j,k The mean of (n-L+l); w j (n) is the noise vector in the jth time slot.

[0018] Furthermore, step S2 includes the following steps:

[0019] The interference term mean is obtained by inputting the channel response estimation value and the symbol mean into the Gaussian random variable distribution model, and the calculation formula is:

[0020]

[0021] Among them, μ j,d (n,p) is the expected interference term between the jth function node and the dth variable node at the nth symbol position of the pth path signal; x j,k (n-L+1) is the symbol at the n-L+1th symbol position of the kth user in the jth functional node; Ω j is the set of users carried by the j-th time slot.

[0022] The interference term mean is obtained by inputting the channel response estimate, symbol variance, and noise power into the Gaussian random variable distribution model, and is calculated as follows:

[0023]

[0024] in, is the variance of the interference term between the jth function node and the dth variable node at the nth symbol position of the pth path signal; Var[x j，k (n-L+l)] is the symbol x j,k The variance of (n-L+l); N0 is the noise power.

[0025] Furthermore, the calculation formula for the logarithmic domain function node message in step S3 is:

[0026]

[0027] in, F j Passed to V k The m-th order logarithmic domain information of the l-th path signal; V k is the kth variable node; F j is the jth functional node; n is the symbol position; L is the number of multipath signals; l is the multipath signal index; is the channel response estimate of the lth signal at the nth symbol position of the dth user in the jth time slot; is x j,d The m-th order symbol of (n-L+l).

[0028] Furthermore, step S4 includes the following steps:

[0029] Generate address information that needs to be read by the variable information storage unit corresponding to the function node when the function node is updated through the control signal, and calculate and update the logarithmic domain function node message in combination with the received signal vector, channel response and codebook data of the function node;

[0030] When the control signal is updated, the updated processing windows of all functional nodes are shifted, and a circular shift operation is performed on the storage units so that the first storage unit is moved to the last one.

[0031] Furthermore, step S5 includes the following steps:

[0032] Calculating updated Gaussian information according to the codebook storage data and the logarithmic domain function node message, and storing the Gaussian information in a matching storage unit;

[0033] The control signal controls the addresses of all storage units storing updated Gaussian information;

[0034] The control signal controls the matching between the variable node update and the storage unit through cyclic shift gating.

[0035] Furthermore, step S5 determines that the probability domain variable node message includes:

[0036] The logarithmic domain information is obtained by calculation in the logarithmic domain. The calculation formula is:

[0037]

[0038] in, From V k Pass to F zThe m-th order logarithmic domain information of the p-th path signal; p is the multipath signal index.

[0039] Perform exponential operation on the logarithmic domain information to calculate the probability domain variable node message. The calculation formula is:

[0040]

[0041] in, From V k Pass to F z The m-th order probability domain variable node message of the p-th path signal.

[0042] Furthermore, the log-likelihood ratio calculation formula of the bit stream in step S7 is:

[0043]

[0044] Among them, LLR k,b is the log-likelihood ratio of the b-th bit of the k-th user; M1 and M0 are the modulation order sets corresponding to when the b-th bit is 1 and 0, respectively; is the sum of the logarithmic domain information under the M1 set; It is the sum of the logarithmic domain information under the M0 set.

[0045] In a second aspect, a signal detection device based on Gaussian approximate interference and a parallel architecture is provided, comprising:

[0046] A receiving signal module is used to model a received signal vector in a single-carrier sparse code division multiple access communication system under frequency selective fading channel conditions to obtain a signal model;

[0047] A data module, configured to store received signals, channel responses, and codebook data, as well as probability domain variable node messages returned to the functional node update module in each iteration;

[0048] A Gaussian modeling module, configured to model the interference term in the received signal model as a Gaussian random variable distribution model, and determine the mean and variance of the interference term according to the Gaussian random variable distribution model;

[0049] a function node updating module, configured to update the function node according to the interference term mean, the interference term variance, the received signal vector, the channel response, and the codebook data to obtain a logarithmic domain function node message;

[0050] a cyclic shift gating module, configured to perform a cyclic shift gating operation according to a control signal, matching the variable node information with the received signal vector, channel response, and codebook data of the functional node to calculate and update the logarithmic domain functional node information;

[0051] A variable node updating module, configured to update the variable node according to the logarithmic domain function node message to obtain a normalized probability domain variable node message and updated Gaussian information;

[0052] an iterative detection module, configured to return the probability domain variable node message to the function node update module, and perform iterative updates according to the received signal, channel response, and codebook data until a preset number of times is reached;

[0053] The output module is used to calculate and output the log-likelihood ratio of the bit stream according to the latest logarithmic domain function node message obtained after iterative update after the last iteration is completed.

[0054] According to a third aspect, an electronic device is provided, comprising the signal detection device based on Gaussian approximation interference and parallel architecture as claimed in claim 8.

[0055] Compared with the prior art, the present invention has the following beneficial effects:

[0056] 1. The present invention models the interference term as a Gaussian random variable distribution, performs multiple cycles of function node updates and variable node updates, and transmits messages between the function node updates and variable node updates controlled by cyclic shift gating operations. Furthermore, the present invention converts data into a logarithmic domain and converts multiplication operations into accumulation operations. Finally, the reliability information of each user's transmitted bit is calculated, forming a partially parallel architecture solution. This reduces the exponential complexity of the traditional extended factor graph-based message passing E-MPA algorithm to a linear level, thus lowering computational complexity, achieving nonlinear optimization, avoiding exponential operations and lookup table (LUT) dependence, optimizing on-chip memory access scheduling, and significantly reducing resource overhead.

[0057] 2. The present invention achieves dynamic binding of updated output data of functional nodes with physical storage addresses in multi-symbol scenarios by cyclically shifting storage units and synchronously shifting the update window of functional nodes. Through the cyclic shift operation, the output of each functional node within the update window is always aligned with the first position of the shifted storage unit, ensuring that the positional relationship between the data processing unit and the storage unit of each control signal remains consistent.

[0058] 3. The present invention proposes a detection device that includes a partially parallel architecture design scheme, which takes into account both throughput rate and area constraints, forms an efficient hardware solution, greatly improves its application scenarios, and provides a reliable SC-SCMA signal detection solution for ultra-dense access systems. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] The drawings described herein are used to provide a further understanding of the embodiments of the present invention, constitute a part of this application, and do not constitute a limitation of the embodiments of the present invention. In the drawings:

[0060] Figure 1 This is a flowchart of Example 1 of the present invention;

[0061] Figure 2 This is a schematic diagram of the system in Example 1 of the present invention;

[0062] Figure 3 This is a system block diagram in Example 2 of the present invention;

[0063] Figure 4 This is the overall framework diagram of the detector in Example 2 of the present invention;

[0064] Figure 5 This is a functional node update module diagram in Example 2 of the present invention;

[0065] Figure 6 This is a diagram of a cyclic shift gating module in embodiment 2 of the present invention;

[0066] Figure 7 This is a diagram of a variable node update module in Example 2 of the present invention. DETAILED DESCRIPTION

[0067] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with examples and drawings. The exemplary embodiments of the present invention and their descriptions are only used to explain the present invention and are not intended to limit the present invention.

[0068] Example 1: Provide a signal detection method based on Gaussian approximation interference and parallel architecture, such as Figure 1 As shown, the following steps are included:

[0069] S1. Under the condition of frequency selective fading channel, the received signal vector is modeled to obtain a signal model;

[0070] S2. Modeling the interference term in the received signal model as a Gaussian random variable distribution model, and determining Gaussian information based on the Gaussian random variable distribution model, where the Gaussian information includes the mean and variance of the interference term;

[0071] S3. Update the function node according to the interference term mean, interference term variance, received signal vector, channel response, and codebook data to obtain a logarithmic domain function node message;

[0072] S4. Perform a cyclic shift gating operation according to the control signal to match the variable node information with the received signal vector, channel response, and codebook data of the functional node to calculate and update the logarithmic domain functional node information;

[0073] S5. Update the variable node according to the logarithmic domain function node message to obtain the normalized probability domain variable node message and the updated Gaussian information;

[0074] S6. Return the probability domain variable node message and the updated Gaussian information to S3, and iterate S3-S5 according to the received signal, channel response, and codebook data until a preset number of times is reached;

[0075] S7. After the last iteration is completed, the log-likelihood ratio of the bit stream is outputted by performing calculations based on the latest log-domain function node message obtained after the iterative update.

[0076] In the existing SCMA system, the SCMA codebook configuration scheme is commonly used, and its technical characteristics are: the multiplexing degree of each physical resource element (RE) is 3, and the multiplexing degree of each user is 2. It should be noted that this scheme is based on the SC-SCMA transmission system, in which the time slot (TS) is the minimum scheduling unit. Therefore, the physical resource unit corresponds to a TS, and each TS transmits symbols, that is, it contains symbol positions (SP).

[0077] like Figure 2 As shown, the present invention relates to an uplink SC-SCMA system, wherein the basic system unit includes J time slots TS and K users. For the kth user, the SCMA encoder maps log2M coded bits into a J-dimensional codeword x k (n) = [x 1，k (n), x 2，k (n), ..., x J，k (n)] T , where x j,k (n) represents the symbol of the nth SP of the kth user in the jth time slot TS. Under the quasi-static multipath fading channel condition with a memory length of L, the system of the present invention receives the signal vector y(n) = [y1(n), y2(n), ..., y J (n)] T Can be modeled as

[0078] Where y(n) is the received signal vector; w(n) is the noise vector; is the multipath signal vector of the kth user.

[0079]

[0080] And x k (nl)=[x 1，k (nl), x 2，k (nl),...,x J，k (nl)] T ;

[0081] At the same time, the channel response matrix It can be expressed as:

[0082]

[0083] Among them, h 1，k (n, l) represents the channel response of the lth signal of the kth user in the nth SP of the jth TS, and the channel responses of different users are independent.

[0084] If the target signal is the p-th path signal of the d-th user, based on the central limit theorem, the interference signal received by the n-th SP in the j-th time slot TS can be modeled as approximately obeying the Gaussian random variable distribution, so the received signal y j (n) can be expressed as the calculation formula of the signal model in step S1:

[0085]

[0086] Among them, y j (n) is the received signal vector at the nth symbol position in the jth time slot; h j，d (p) is the channel response of the lth signal of the dth user at the nth symbol position in the jth time slot, x j,d (n-L+p) is the symbol at the n-L+pth symbol position of the dth user in the jth time slot; h j,k (n, l) represents the channel response of the lth signal of the kth user at the nth symbol position in the jth time slot, For h j,k The estimated value of (n,l); E[x j,k (n-L+l)] is the symbol x j,k The mean of (n-L+l); w j (n) is the noise vector in the jth time slot; Z j,d (n,p) represents the interference term and obeys

[0087] The basic principle of FN updating is to calculate the marginal conditional probability of each user based on the received signal.

[0088] Step S2 includes the following steps: the interference term mean is obtained by inputting the channel response estimation value and the symbol mean into the Gaussian random variable distribution model, and the calculation formula is:

[0089]

[0090] Among them, μ j,d(n,p) is the expected interference term between the jth function node and the dth variable node at the nth symbol position of the pth path signal; x j,k (n-L+l) is the symbol of the k-th user in the j-th functional node at the n-L+l-th symbol position; E[x j,k (n-L+l)] is the symbol x j,k The mean of (n-L+l); h j,k (n, l) represents the channel response of the lth signal of the kth user at the nth symbol position in the jth time slot; Ω represents the channel response estimate of the lth signal of the kth user at the nth symbol position in the jth time slot; j is the set of users carried by the j-th time slot.

[0091] The interference term mean is obtained by inputting the channel response estimate, symbol variance, and noise power into the Gaussian random variable distribution model, and is calculated as follows:

[0092]

[0093] in, is the variance of the interference term between the jth function node and the dth variable node at the nth symbol position of the pth path signal; Var[x j,k (n-L+1)] is the symbol x j,k The variance of (n-L+l); N0 is the noise power.

[0094]

[0095] Var[x j,k (n-L+l)]=E[|x j,k (n-L+l)| 2 ]-E 2 [x j,k (n-L+l)].

[0096] Then, using μ j,d (n,p) and σ 2 j,d (n,p) can be obtained from F j Passed to V d Therefore, in step S3, the calculation formula of the logarithmic domain function node message is:

[0097]

[0098] in, F j Passed to V k The m-th order logarithmic domain information of the l-th path signal; Vk is the kth variable node; F j is the jth functional node; n is the symbol position; L is the number of multipath signals; l is the multipath signal index; is the channel response estimate of the lth signal at the nth symbol position of the dth user in the jth time slot; is x j,d The m-th order symbol of (n-L+l).

[0099] Step S4 includes the following steps: generating address information that needs to be read by the variable information storage unit corresponding to the function node when the function node is updated through the control signal, and calculating and updating the logarithmic domain function node message in combination with the received signal vector, channel response and codebook data of the function node;

[0100] When the control signal is updated, the updated processing windows of all functional nodes are shifted, and a circular shift operation is performed on the storage units so that the first storage unit is moved to the last one.

[0101] Step S5 includes the following steps: calculating updated Gaussian information based on the codebook storage data and the logarithmic domain function node message, and storing the Gaussian information in the matching storage unit; controlling the address of all storage units storing the updated Gaussian information with a control signal; and matching the control variable node update with the storage unit through a cyclic shift gating control.

[0102] There are many consecutive multiplication calculations in the normalization operation in the variable node VN update, and as the number of multipath signals increases, the number of multipliers will increase exponentially, which will occupy a large amount of computing resources. Therefore, the present invention optimizes these nonlinear calculations, converts the data to the logarithmic domain, and converts the consecutive multiplication operations into accumulation operations, which will greatly reduce resource overhead.

[0103] Specifically, step S5 determines the probability domain variable node information, including: obtaining logarithmic domain information by calculation in the logarithmic domain, and the calculation formula is:

[0104]

[0105] in, From V k Pass to F z The m-th order logarithmic domain information of the p-th path signal; p is the multipath signal index.

[0106] in,

[0107]

[0108] Perform exponential operation on the logarithmic domain information to calculate the probability domain variable node message. The calculation formula is:

[0109]

[0110] in, From V k Pass to F z The m-th order probability domain variable node message of the p-th path signal.

[0111] The present invention will explain the derivation process of this formula in detail below.

[0112] for In terms of cumulative factors Yes Take the logarithm, and the latter involves exponential operation, so the exponential operation can be omitted in the function node FN update, thereby avoiding the use of the lookup table. We define:

[0113]

[0114] for For , logarithmic addition is performed using log-exponential and factorization identities.

[0115] f(x,y)@ln(e x +e y )=max(x,y)+ln(1+e -|x-y| ).

[0116] Its core function is to transform the logarithm of the sum of two exponential terms into a combination of a maximum function and a correction term, thereby improving the stability of numerical calculations. So far, we have completed the transformation of the continuous multiplication calculation in the probability domain into the accumulation calculation in the logarithmic domain. Finding the exponent can get the probability domain information

[0117] After completing multiple FN updates and VN updates, the logarithmic likelihood ratio (LLR) of the b-th bit of the k-th user in step S7 is calculated as follows:

[0118]

[0119] Among them, LLR k,b is the log-likelihood ratio of the b-th bit of the k-th user; M1 and M0 are the modulation order sets corresponding to when the b-th bit is 1 and 0, respectively; and are the sum of the logarithmic domain information under the M1 and M0 sets respectively. The definitions of the two are the same. We use For example

[0120]

[0121] Among them, Len() is to obtain the length of the data set; Ω k is the number of time slots occupied by the kth user.

[0122] Example 2:

[0123] In order to meet the signal detector throughput requirements and area constraints, the SCMA detector adopts a partially parallel hardware architecture design. Based on the computational optimization of a large number of nonlinear operations such as multiplication, division, and exponential in the original scheme in Example 1, the resource overhead can be further reduced by optimizing the on-chip storage access scheduling to complete the partially parallel architecture design. Because, for the hardware implementation of the E-GAIBP detector, the full parallel architecture will cause huge resource overhead, and the serial architecture will not be able to meet the system throughput requirements, therefore, the partially parallel architecture design proposed in the present invention takes into account both throughput rate and area constraints, forming an efficient hardware solution, greatly improving its application scenarios, and providing a reliable SC-SCMA signal detection solution for ultra-dense access systems.

[0124] This embodiment provides a signal detection device based on Gaussian approximation interference and parallel architecture, such as Figure 3As shown, it includes a receiving signal module, a data module, a Gaussian modeling module, a function node update module, a cyclic shift gating module, a variable node update module, an iterative detection module, and an output module. Among them, the receiving signal module is used to model the received signal vector under the frequency selective fading channel condition in a single-carrier sparse code division multiple access communication system to obtain a signal model; the data module is used to store the received signal storage RxSigMem, the channel response storage HchMem and the codebook data CBMem, as well as the probability domain variable node message returned to the function node update module in each iteration; the Gaussian modeling module is used to model the interference term in the received signal model as a Gaussian random variable distribution model, and determine the interference term mean and interference term variance according to the Gaussian random variable distribution model; the function node update module is used to update the function node according to the interference term mean, interference term variance, the received signal vector, the channel response and the codebook data to obtain a logarithmic domain function node message; The shift gating module is used to perform a cyclic shift gating operation according to the control signal, match the variable node information with the received signal vector, channel response and codebook data of the functional node, so as to calculate and update the logarithmic domain functional node information; the variable node update module is used to update the variable node according to the logarithmic domain functional node message, and obtain the normalized probability domain variable node message and the updated Gaussian information; the iterative detection module is used to return the probability domain variable node message to the functional node update module, and iteratively update according to the received signal, channel response and codebook data until a preset number of times is reached; the output module is used to calculate according to the latest logarithmic domain functional node message obtained after the iterative update after the last iteration, and output the log-likelihood ratio of the bit stream. The schematic diagram is referenced. Figure 4 .

[0125] In some embodiments, as Figure 5As shown, in the functional node update module, at the data input, the data module transmits the received signal Rxsig_Info, the channel response Hch_Info, and the codebook data CB_Info, while simultaneously reading the Gaussian information Gauss_Info from the matching RAM memory. At the data output, the logarithmic domain functional node message lnq is calculated based on the Gaussian information Gauss_Info, CB_Info, H_Info, and Rxsig_Info. At the control end, the control signal State_cnt transmitted by the cyclic shift gate Ctr_unit module controls the iteration number, symbol scheduling, multipath scheduling, and operating state of the functional node FN update module. The control signal State_cnt contains the Iter, Symbol, Tap, and State signals. The workflow is as follows: first, the control signal Tap generates the address information required to be read from the RAM corresponding to the functional node update module. Then, the control signal Symbol serves as the input of the cyclic shift gate, matching the RAM memory with the correct functional node update module. After reading all the required Gaussian information, the required logarithmic domain functional node message lnq is calculated.

[0126] In some embodiments, as Figure 6 As shown, the cyclic shift gating module operates as follows: when the current control signal symbol is n, n to n+L-1 function node update modules are updated. Assume that these function node update modules correspond to RAM1 through RAML, in order. When the control signal symbol changes to n+1, all function node update modules are shifted, updating n+1 to n+L function node update modules. Simultaneously, all RAM memories are cyclically shifted, becoming RAM2, RAM3, ..., RAML, RAM1. This ensures that the function node update module FN_Block_(N+M) always matches the RAM memory (1+M%L). It can be seen that the function node update module FN_Block_N+1 always matches the RAM memory RAM2.

[0127] In some embodiments, as Figure 7As shown, in the variable node update module, the data input reads CN_Info from the CBMem module and the logarithmic-domain function node message lnq from the function node update module. At the data output, the updated Gaussian information Gauss_Info and the bitstream's log-likelihood ratio (LLR) are calculated. At the control end, the control signal State_cnt controls the number of iterations, symbol scheduling, multipath scheduling, and operating state of the variable node update module. The control signal State_cnt includes the Iter, Symbol, Tap, and State signals. The workflow is as follows: the variable node update module receives the logarithmic-domain function node message lnq from the function node update module and the codebook data CB_info from the codebook storage CBMem. After calculation, it obtains the updated Gaussian information and stores it in the matching RAM memory. The control signal Symbol controls the addresses of all RAM memory locations for storing the updated Gaussian information and, through a cyclic shift gate, matches the variable node update module with the RAM memory.

[0128] Embodiment 3: An electronic device is provided, which includes the signal detection device based on Gaussian approximation interference and parallel architecture as described in Embodiment 2.

[0129] Working principle: The present invention models the interference term as a Gaussian random variable distribution, undergoes multiple cycles of function node updates and variable node updates, and transmits messages between the function node updates and variable node updates controlled by cyclic shift gating operations. At the same time, the data is converted to the logarithmic domain, and the continuous multiplication operation is converted into an accumulation operation. Finally, the reliability information of the bits sent by each user is calculated, forming a partial parallel architecture solution. The exponential complexity of the traditional E-MPA algorithm based on extended factor graph message passing is reduced to the linear level, the computational complexity is reduced, nonlinear optimization is achieved, exponential operations and lookup table (LUT) dependence are avoided, on-chip storage access scheduling is optimized, and resource overhead is reduced.

[0130] Those skilled in the art will appreciate that the embodiments of the present application can be provided as methods, systems, or computer program products. Therefore, the present application can adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application can adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) that contain computer-usable program code.

[0131] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the steps in the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0132] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0133] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0134] The specific implementation methods described above further illustrate the objectives, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above description is only a specific implementation method of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A signal detection method based on Gaussian approximate interference and parallel architecture, characterized in that: The steps include: S1. Under the condition of frequency selective fading channel, the received signal vector is modeled to obtain a signal model; S2. Modeling the interference term in the received signal model as a Gaussian random variable distribution model, and determining Gaussian information according to the Gaussian random variable distribution model, where the Gaussian information includes a mean value and a variance of the interference term; S3. Update the function node according to the interference term mean, the interference term variance, the received signal vector, the channel response, and the codebook data to obtain a logarithmic domain function node message; S4. Perform a cyclic shift gating operation according to the control signal to match the variable node information with the received signal vector, channel response, and codebook data of the functional node to calculate and update the logarithmic domain functional node information; S5. Update the variable node according to the logarithmic domain function node message to obtain the probability domain variable node message and the updated Gaussian information; S6. Return the probability domain variable node message and the updated Gaussian information to S3, and iteratively execute S3-S5 according to the received signal, channel response, and codebook data until a preset number of times is reached; S7. After the last iteration is completed, calculation is performed based on the latest logarithmic domain function node message obtained after iterative update, and the log-likelihood ratio of the bit stream is output.

2. The signal detection method based on Gaussian approximate interference and parallel architecture according to claim 1, characterized in that: The calculation formula of the signal model in step S1 is: Among them, y j (n) is the received signal vector at the nth symbol position in the jth time slot; h j,d (n,p) is the channel response of the p-th signal of the d-th user at the n-th symbol position in the j-th time slot; x j,d (n-L+p) is the symbol at the n-L+pth symbol position of the dth user in the jth time slot; x j,k (n-L+1) is the symbol at the n-L+1th symbol position of the dth user in the jth time slot; Ω j is the set of users carried by the jth time slot; h j,k (n, l) represents the channel response of the lth signal of the kth user at the nth symbol position in the jth time slot; w j (n) is the nth noise in the jth time slot.

3. The signal detection method based on Gaussian approximate interference and parallel architecture according to claim 1, characterized in that: Step S2 includes the following steps: The interference term mean is obtained by inputting the channel response estimation value and the symbol mean into the Gaussian random variable distribution model, and the calculation formula is: Among them, μ j,d (n,p) is the expected interference term of the p-th path signal of the d-th user at the n-th symbol position in the j-th time slot; x j,k (n-L+l) is the symbol at the n-L+lth symbol position of the kth user in the jth time slot; E[x j,k (n-L+l)] is the symbol x j,k The mean of (n-L+l); Ω represents the channel response estimate of the lth signal of the kth user at the nth symbol position in the jth time slot; j is the set of users carried by the j-th time slot. The interference term variance is obtained by inputting the channel response estimate, symbol variance and noise power into the Gaussian random variable distribution model, and the calculation formula is: in, is the variance of the interference term of the p-th path signal of the d-th user at the n-th symbol position in the j-th time slot; Var[x j，k (n-L+l)] is the symbol x j,k The variance of (n-L+l); N0 is the noise power.

4. The signal detection method based on Gaussian approximate interference and parallel architecture according to claim 1, characterized in that: The calculation formula for the logarithmic domain function node message in step S3 is: in, F j Passed to V k The m-th order logarithmic domain information of the l-th path signal; V k is the kth variable node; F j is the jth functional node; n is the symbol position; L is the number of multipath signals; l is the multipath signal index; is the channel response estimate of the lth signal at the nth symbol position of the dth user in the jth time slot; is x j,d The m-th order symbol of (n-L+l).

5. The signal detection method based on Gaussian approximate interference and parallel architecture according to claim 1, characterized in that: Step S4 includes the following steps: Generate address information that needs to be read by the variable information storage unit corresponding to the function node when the function node is updated through the control signal, and calculate and update the logarithmic domain function node message in combination with the received signal vector, channel response and codebook data of the function node; When the control signal is updated, the updated processing windows of all functional nodes are shifted, and a circular shift operation is performed on the storage units so that the first storage unit is moved to the last one.

6. The signal detection method based on Gaussian approximate interference and parallel architecture according to claim 1, characterized in that: Step S5 includes the following steps: Calculating updated Gaussian information according to the codebook storage data and the logarithmic domain function node message, and storing the Gaussian information in a matching storage unit; The control signal controls the addresses of all storage units storing updated Gaussian information; The control signal controls the matching between the variable node update and the storage unit through cyclic shift gating.

7. The signal detection method based on Gaussian approximate interference and parallel architecture according to claim 5, characterized in that: Step S5 determines the probability domain variable node message includes: The logarithmic domain information is obtained by calculation in the logarithmic domain. The calculation formula is: in, From V k Pass to F z The m-th order logarithmic domain information of the p-th path signal; p is the multipath signal index. Perform exponential operation on the logarithmic domain information to calculate the probability domain variable node message. The calculation formula is: in, From V k Pass to F z The m-th order probability domain variable node message of the p-th path signal.

8. The signal detection method based on Gaussian approximate interference and parallel architecture according to claim 1, characterized in that: The log-likelihood ratio calculation formula of the bit stream in step S7 is: Among them, LLR k,b is the log-likelihood ratio of the b-th bit of the k-th user; M1 and M0 are the modulation order sets corresponding to when the b-th bit is 1 and 0, respectively; is the sum of the logarithmic domain information under the M1 set; It is the sum of the logarithmic domain information under the M0 set.

9. A signal detection device based on Gaussian approximate interference and parallel architecture, characterized in that: include: A receiving signal module is used to model a received signal vector in a single-carrier sparse code division multiple access communication system under frequency selective fading channel conditions to obtain a signal model; A data module, configured to store received signals, channel responses, and codebook data, as well as probability domain variable node messages returned to the functional node update module in each iteration; A Gaussian modeling module, configured to model the interference term in the received signal model as a Gaussian random variable distribution model, and determine the mean and variance of the interference term according to the Gaussian random variable distribution model; a function node updating module, configured to update the function node according to the interference term mean, the interference term variance, the received signal vector, the channel response, and the codebook data to obtain a logarithmic domain function node message; a cyclic shift gating module, configured to perform a cyclic shift gating operation according to a control signal, matching the variable node information with the received signal vector, channel response, and codebook data of the functional node to calculate and update the logarithmic domain functional node information; A variable node updating module, configured to update the variable node according to the logarithmic domain function node message to obtain a normalized probability domain variable node message and updated Gaussian information; an iterative detection module, configured to return the probability domain variable node message to the function node update module, and perform iterative updates according to the received signal, channel response, and codebook data until a preset number of times is reached; The output module is used to calculate and output the log-likelihood ratio of the bit stream according to the latest logarithmic domain function node message obtained after iterative update after the last iteration is completed.

10. An electronic device, characterized in that: The electronic device includes the signal detection device based on Gaussian approximation interference and parallel architecture as claimed in claim 8.

Images