技术领域technical field
本发明属于无线通信技术领域，具体的说是涉及一种降低稀疏码多址(SparseCode Multiple Access,SCMA)系统译码复杂度的方法，更具体的说是基于动态因子图的降低SCMA 系统中消息传递(Message Passing Algorithm,MPA)复杂度的方法。The present invention belongs to the technical field of wireless communication, and in particular relates to a method for reducing the decoding complexity of a sparse code multiple access (SCMA) system, and more specifically, a method for reducing message complexity in an SCMA system based on a dynamic factor graph. A method of passing (Message Passing Algorithm, MPA) complexity.
背景技术Background technique
随着移动互联网和移动物联网的高速发展，5G需要满足更加多样化的场景和极致的性能 挑战，不仅需要大幅度提升系统频谱效率，而且还要具备支持海量设备连接的能力。传统的 正交多址技术已经不能适应未来移动通信的发展，比如频分多址、时分多址、码分多址等， 新型的非正交多址技术应运而生。SCMA技术作为一种面向5G无线移动通信的非正交空口 技术，它将低密度码(Low Density Signature,LDS)和调制相结合，根据码本设计原理选择 最优的码本结合并分配给特定的用户，然后不同用户根据所分配的码本进行信息传输。由于 SCMA采用非正交稀疏编码叠加技术，在相同的时频资源条件下，SCMA技术能够支持更多 的用户连接。同时，SCMA将多维调制和扩频技术结合在一起，这样能够支持海量连接和提 高频谱效率。相比最优的多用户检测算法，消息传递算法(Message PassingAlgorithm,MPA) 能够有效降低SCMA多用户检测算法复杂度，并且译码性能接近最优。然而，随着用户数量 和码本维度的增加，MPA算法的复杂度依旧呈指数增长，所以降低MPA算法的计算复杂度 尤为重要。With the rapid development of mobile Internet and mobile Internet of Things, 5G needs to meet more diverse scenarios and extreme performance challenges. It not only needs to greatly improve the system spectrum efficiency, but also has the ability to support massive device connections. The traditional orthogonal multiple access technology can no longer adapt to the development of future mobile communication, such as frequency division multiple access, time division multiple access, code division multiple access, etc., and new non-orthogonal multiple access technology emerges as the times require. As a non-orthogonal air interface technology for 5G wireless mobile communication, SCMA technology combines low density code (Low Density Signature, LDS) and modulation, and selects the optimal codebook combination according to the codebook design principle and assigns it to a specific users, and then different users transmit information according to the assigned codebook. Since SCMA adopts non-orthogonal sparse coding superposition technology, under the same time-frequency resource conditions, SCMA technology can support more user connections. At the same time, SCMA combines multi-dimensional modulation and spread spectrum technology, which can support massive connections and improve spectrum efficiency. Compared with the optimal multi-user detection algorithm, the Message Passing Algorithm (MPA) can effectively reduce the complexity of the SCMA multi-user detection algorithm, and the decoding performance is close to optimal. However, as the number of users and the dimension of the codebook increase, the complexity of the MPA algorithm still increases exponentially, so it is particularly important to reduce the computational complexity of the MPA algorithm.
发明内容Contents of the invention
本发明的目的是，针对上述问题，提供一种基于动态因子图(Dynamic FactorGraph,DFG) 的MPA算法来降低上行链路SCMA系统多用户检测的复杂度。根据在消息迭代更新过程中 动态地减少因子图中参与消息更新的边的数量，本发明可以大幅度降低SCMA系统的译码复 杂度。通过选择不同的停止消息更新的边的数量，本发明在SCMA多用户检测复杂度和误比 特率(Bit Error Rate,BER)性能之间进行折中。因此，本发明也能够自适应地满足5G不同 场景下的通信质量要求。The object of the present invention is to provide an MPA algorithm based on a Dynamic Factor Graph (DFG) to reduce the complexity of multi-user detection in an uplink SCMA system in view of the above problems. According to dynamically reducing the number of edges participating in message update in the factor graph during the message iterative update process, the invention can greatly reduce the decoding complexity of the SCMA system. By selecting different numbers of sides to stop message updates, the present invention makes a compromise between SCMA multi-user detection complexity and bit error rate (Bit Error Rate, BER) performance. Therefore, the present invention can also adaptively meet the communication quality requirements in different 5G scenarios.
本发明的技术方案是：Technical scheme of the present invention is:
一种降低SCMA系统译码复杂度的方法，该方法用于上行链路SCMA系统中的消息传递，预设该上行SCMA通信系统模型为，有J个用户共享K个时频资源，每个用户采用的星 座点数为M，用户的过载因子定义为：A method for reducing the decoding complexity of an SCMA system, the method is used for message transmission in an uplink SCMA system, the uplink SCMA communication system model is preset as, there are J users sharing K time-frequency resources, each user The number of constellation points used is M, and the user's overload factor is defined as:
其特征在于，包括以下步骤：It is characterized in that, comprising the following steps:
S1、设置SCMA系统参数，包括对用户数J、时频资源K、码本大小M、过载因子λ进 行赋值以及设置最大迭代次数为tmax；S1, setting the SCMA system parameters, including assigning values to the number of users J, time-frequency resources K, codebook size M, overload factor λ, and setting the maximum number of iterations as tmax ;
S2、初始化消息传递概率值为：S2. The initial message delivery probability value is:
其中，是指因子图中用户节点uj到资源节点rk的消息更新过程，其中t是迭代次数， xj＝(x1,j,x2,j…,xK,j)T∈CK表示第j个用户发送的码字；in, refers to the message update process from user node uj to resource node rk in the factor graph, where t is the number of iterations, xj =(x1,j ,x2,j …,xK,j )T ∈ CK means The code word sent by the jth user;
S3、设置迭代次数初始值t＝1；S3. Set the initial value of the number of iterations t=1;
S4、进入消息迭代更新，判断t是否超出步骤S1中设定的最大迭代次数，若t＝1，则进 入步骤S6，若1<t≤tmax，则进入步骤S5，若t>tmax，则终止消息迭代更新并进入步骤S8；S4. Enter message iterative update, judge whether t exceeds the maximum number of iterations set in step S1, if t=1, then enter step S6, if 1<t≤tmax , then enter step S5, if t>tmax , Then terminate the iterative update of the message and enter step S8;
S5、否则计算因子图中每条双向边的置信度，并得到置信度集合ΦK,J(t)，然后对ΦK,J(t) 中的置信度元素进行排序，将置信度较大的G条双向边设置成单向边，更新此次消息迭代所 需要的因子图F(t)；S5. Otherwise, calculate the confidence degree of each bidirectional edge in the factor graph, and obtain the confidence degree set ΦK, J (t), and then sort the confidence degree elements in ΦK, J (t), and set the confidence degree The G bidirectional edges of G are set as unidirectional edges, and the factor graph F(t) required for this message iteration is updated;
S6、根据更新后的因子图F(t)进行资源节点与用户节点的消息更新，具体为：S6. Update the message of the resource node and the user node according to the updated factor graph F(t), specifically:
S61、资源节点的消息更新，对所有资源节点rk计算：S61. The message update of resource nodes is calculated for all resource nodes rk :
其中，σ是高斯分布的标准差，yk是基站在第k个时频资源所接收到的信号，h为信道，是指因子图中用户节点up到资源节点rk的消息更新过程；Among them, σ is the standard deviation of the Gaussian distribution, yk is the signal received by the base station at the kth time-frequency resource, h is the channel, refers to the message update process from user node up to resource node rk in the factor graph;
S62、用户节点的消息更新，对所有用户节点uj计算：S62. The message update of the user node is calculated for all user nodes uj :
S7、迭代次数进行更新t＝t+1，回到步骤S4；S7, update the number of iterations t=t+1, return to step S4;
S8、利用更新后的概率分布，完成多用户检测，对用户进行软判决输出为：S8. Using the updated probability distribution, the multi-user detection is completed, and the soft-judgment output for the user is:
本发明的依据是，由于传统MPA算法在迭代过程中需要对因子图中所有的边进行更新， 即使因子图中部分边对应的概率分布已经收敛，此方案会造成巨大的计算冗余并且效率较低， 所以本发明提出一种基于动态因子图的消息传递算法(即DFG-MPA算法)来降低SCMA多 用户检测的计算复杂度。在每次迭代开始之前计算因子图中的每条双向边所对应的置信度， 将置信度较高的双向边设置成单向边，单向边在此次迭代过程中将停止从资源节点到用户节 点的消息更新，这样能够实现大幅度降低多用户检测的复杂度，并且保证BER性能基本没有 损失。The basis of the present invention is that since the traditional MPA algorithm needs to update all the edges in the factor graph during the iterative process, even if the probability distributions corresponding to some edges in the factor graph have converged, this scheme will cause huge computational redundancy and low efficiency. Low, so the present invention proposes a dynamic factor graph-based message passing algorithm (ie, DFG-MPA algorithm) to reduce the computational complexity of SCMA multi-user detection. Before the start of each iteration, calculate the confidence corresponding to each bidirectional edge in the factor graph, set the bidirectional edge with higher confidence as a unidirectional edge, and the unidirectional edge will stop from the resource node to the The message update of user nodes can greatly reduce the complexity of multi-user detection and ensure that BER performance is basically not lost.
本发明的有益效果是：能够大幅度降低SCMA系统译码复杂度，在每次迭代过程中通过 选择合适的停止更新的双向边的数量，译码性能几乎没有损失；由于每次迭代都能得到因子 图中每条双向边所对应的置信度，而每条双向边的置信度是由发送各个码字的概率分布所决 定，所以本发明能够充分利用码本中各个码字概率的分布特征来确定因子图中每条双向边的 置信度，双向边的置信度越大，其概率分布越稳定。为了降低计算冗余，将置信度较大的双 向边设置成单向边，并且双向边设置成单向边的数量可以自适应地选择，这样能够实现在检 测复杂度和译码性能之间进行折中，同时也能满足5G不同质量要求的通信场景。The beneficial effects of the present invention are: the decoding complexity of the SCMA system can be greatly reduced, and the decoding performance is hardly lost by selecting the appropriate number of bidirectional edges to stop updating in each iteration; since each iteration can obtain The confidence degree corresponding to each bidirectional edge in the factor graph, and the confidence degree of each bidirectional edge is determined by the probability distribution of each codeword sent, so the present invention can make full use of the distribution characteristics of each codeword probability in the codebook to Determine the confidence level of each two-way edge in the factor graph, the greater the confidence level of the two-way edge, the more stable its probability distribution. In order to reduce computational redundancy, the bidirectional edges with high confidence are set as unidirectional edges, and the number of bidirectional edges set as unidirectional edges can be adaptively selected, which can achieve a balance between detection complexity and decoding performance. At the same time, it can meet the communication scenarios with different quality requirements of 5G.
附图说明Description of drawings
图1为上行链路SCMA系统框图；Figure 1 is a block diagram of an uplink SCMA system;
图2为SCMA编码原理图；Fig. 2 is a schematic diagram of SCMA coding;
图3为因子图变化示意图；Figure 3 is a schematic diagram of factor graph changes;
图4为DFG-MPA算法流程图。Figure 4 is the flow chart of DFG-MPA algorithm.
具体实施方式Detailed ways
下面结合附图和实施例详细说明本发明的技术方案。The technical solution of the present invention will be described in detail below in conjunction with the drawings and embodiments.
虑一个上行链路SCMA通信系统模型，J个用户共享K个时频资源块，其原理框图如图1所示。SCMA系统根据码本设计原理生成相应的码本并分配给特定的用户，用户根据所要传输的二进制比特流选择码本中对应的码字进行传输，码本映射定义为：Blog2M→χj，其中M为每个码本包含的码字个数，χj为第j个用户码本且|χj|＝M，SCMA编码过程如图2所示。 在信号发送端，SCMA通过直接将二进制比特流映射为多维码字，将传统正交多址方案中的 调制与扩频结合在一起，编码方式更加高效。在接收端，基站接收到的信号为来自J个不同 用户发送码字的叠加，由于码字是稀疏的，基站端可以采用MPA算法进行译码。SCMA系 统的过载因子定义为Consider an uplink SCMA communication system model, where J users share K time-frequency resource blocks, and its functional block diagram is shown in Figure 1. The SCMA system generates the corresponding codebook according to the codebook design principle and assigns it to a specific user. The user selects the corresponding codeword in the codebook for transmission according to the binary bit stream to be transmitted. The codebook mapping is defined as: Blog2M → χj , Where M is the number of codewords contained in each codebook, χj is the jth user codebook and |χj |=M, and the SCMA encoding process is shown in Figure 2. At the signal sending end, SCMA combines the modulation and spread spectrum in the traditional orthogonal multiple access scheme by directly mapping the binary bit stream into multi-dimensional code words, and the coding method is more efficient. At the receiving end, the signal received by the base station is the superposition of codewords sent by J different users. Since the codewords are sparse, the base station can use the MPA algorithm for decoding. The overload factor of SCMA system is defined as
一般而言过载因子满足λ>1，即较少的时频资源实现更多的用户连接数，这也是SCMA 系统能够支持海量连接的原因所在。Generally speaking, the overload factor satisfies λ>1, that is, fewer time-frequency resources can achieve more user connections, which is why the SCMA system can support massive connections.
假设所有用户时间是同步，则基站端接收到的信号表示为Assuming that all user times are synchronized, the signal received by the base station is expressed as
式中y＝(y1,y2,…,yK)T表示基站端接收到的信号，xj＝(x1,j,x2,j…,xK,j)T表示第j个用户发 送的码字，hj＝(h1,j,h2,j…,hK,j)T表示基站与第j个用户之间的信道状态信息，n＝(n1,n2,…,nK)T表示加性高斯白噪声(additive white Gaussian noise,AWGN)，并且服从分布n～CN(0,σ2I)。In the formula, y=(y1 ,y2 ,…,yK )T represents the signal received by the base station, and xj =(x1,j ,x2,j ...,xK,j )T represents the jth The code word sent by the user, hj =(h1,j ,h2,j ...,hK,j )T represents the channel state information between the base station and the jth user, n=(n1 ,n2 , ...,nK )T represents additive white Gaussian noise (AWGN), and obeys the distribution n~CN(0,σ2 I).
SCMA码的总体结构可以用因子图(包含资源节点和用户节点)和因子图矩阵 F＝(F1,F2,…,FJ)表示，其中Fj＝(f1,j,f2,j,…,fK,j)。因子图与因子图矩阵存在一定的联系，用户节点和资源节点相连当且仅当fk,j＝1。在规则的SCMA系统中，每个用户占用相同数目的资源块并且每个资源块关联相同数目的用户。用表示用户j所占用的时频资 源且|ζj|＝du，表示资源块k所关联的用户且|ξk|＝dr。因此在第k个时频资源 所接收到的信号可以表示为：The overall structure of the SCMA code can be represented by a factor graph (including resource nodes and user nodes) and a factor graph matrix F=(F1 ,F2 ,...,FJ ), where Fj =(f1,j ,f2, j ,...,fK,j ). There is a certain relationship between the factor graph and the factor graph matrix, and the user node is connected to the resource node if and only if fk,j =1. In a regular SCMA system, each user occupies the same number of resource blocks and each resource block is associated with the same number of users. use Indicates the time-frequency resource occupied by user j and |ζj |=du , Denotes the user associated with resource block k and |ξk |=dr . Therefore, the signal received at the kth time-frequency resource can be expressed as:
SCMA技术将输入的比特流通过高维度的调制和扩频结合在一起直接映射成码本中多维 稀疏的码字，所以SCMA系统根据码字稀疏特征可以利用多用户检测算法译码出每个用户所 发送的比特信息。与此同时，SCMA通信系统利用码字的稀疏性，提出一种次优的MPA多 用户检测算法，MPA算法是变量节点和资源节点之间的消息传递，然后不断地迭代。是 指因子图中资源节点rk到用户节点uj的消息更新过程，是指因子图中用户节点uj到资源 节点rk的消息更新过程，其中t是迭代次数。通过对MPA算法复杂度的分析可知，原始MPA 算法的复杂度主要集中在资源节点的消息更新。因此，本发明主要降低原始MPA算法中资源 节点上的消息更新复杂度。原始MPA算法在每次迭代过程中需要更新因子图中所有的边，即 使因子图中的部分边在迭代过程中几乎不再变化(边的置信度较高)仍需要更新，这样会造 成大量的计算冗余，增加计算负担。本发明所提算法在每次迭代开始之前计算因子图中每条 边的置信度，选择置信度低的参与到消息更新过程，而置信度较高的边停止消息更新，这样 能够大幅度降低计算复杂度，使SCMA多用户检测算法更加高效。SCMA technology combines the input bit stream through high-dimensional modulation and spread spectrum to directly map the multi-dimensional sparse codewords in the codebook, so the SCMA system can use the multi-user detection algorithm to decode each user according to the sparseness of the codewords. The bit information sent. At the same time, the SCMA communication system uses the sparsity of codewords to propose a suboptimal MPA multi-user detection algorithm. The MPA algorithm is a message transfer between variable nodes and resource nodes, and then iteratively. refers to the message update process from resource node rk to user node uj in the factor graph, refers to the message update process from user node uj to resource node rk in the factor graph, where t is the number of iterations. The analysis of the complexity of the MPA algorithm shows that the complexity of the original MPA algorithm is mainly concentrated in the message update of the resource nodes. Therefore, the present invention mainly reduces the complexity of message update on resource nodes in the original MPA algorithm. The original MPA algorithm needs to update all the edges in the factor graph during each iteration. Even if some edges in the factor graph hardly change during the iteration process (the edges have higher confidence), they still need to be updated, which will cause a large number of Computational redundancy increases the computational burden. The algorithm proposed in the present invention calculates the confidence degree of each edge in the factor graph before each iteration, and selects the edge with a low confidence degree to participate in the message update process, while the edge with a higher confidence degree stops the message update, which can greatly reduce the calculation The complexity makes the SCMA multi-user detection algorithm more efficient.
所提DFG-MPA多用户检测算法将因子图中的边划分为两种类型：单向边和双向边，示 例如图3所示。单向边定义为：只进行从用户节点到资源节点的消息更新过程，而不能进行 从资源节点到用户节点的消息更新过程；双向边定义为：既能够进行从用户节点到资源节点 的消息更新过程，也能够进行从资源节点到用户节点的消息更新过程。在因子图中，单向边 与双向边可以表示为The proposed DFG-MPA multi-user detection algorithm divides the edges in the factor graph into two types: one-way edges and two-way edges. An example is shown in Figure 3. The one-way edge is defined as: only the message update process from the user node to the resource node can be performed, but the message update process from the resource node to the user node cannot be performed; the bidirectional edge is defined as: both the message update process from the user node to the resource node can be performed The process can also carry out the message update process from the resource node to the user node. In the factor graph, the one-way edge and the two-way edge can be expressed as
其中Bs(t)与Bd(t)分别表示第t次迭代因子图中的双向边集合与单向边集合。Among them, Bs (t) and Bd (t) represent the bidirectional edge set and unidirectional edge set in the t-th iteration factor graph, respectively.
显然地，单向边与双向边之间也存在一定的关系，可以表示为Obviously, there is also a certain relationship between one-way edges and two-way edges, which can be expressed as
|Bs(t)|+|Bd(t)|＝Kdr|Bs (t)|+|Bd (t)|＝Kdr
其中|Bs(1)|＝Kdr且|Bd(1)|＝0。where |Bs (1)|=Kdr and |Bd (1)|=0.
接下来将介绍如何区分单向边与双向边。在迭代开始之前，根据MPA算法上次迭代的码 字概率分布来确定此次迭代过程中双向边的置信度，因子图F(t)中的每条双向边的置信度定 义如下Next, we will introduce how to distinguish between one-way edges and two-way edges. Before the iteration starts, the confidence degree of the bidirectional edge in this iteration process is determined according to the codeword probability distribution of the last iteration of the MPA algorithm. The confidence degree of each bidirectional edge in the factor graph F(t) is defined as follows
其中与分别是码本中的码字，为第t次迭代双向边的置 信度。in and codebook codeword in is the confidence of the two-way edge at the t-th iteration.
用ΦK,J(t)表示第t次迭代因子图中的双向边对应的置信度集合，则可以得到：Using ΦK,J (t) to represent the confidence set corresponding to the bidirectional edge in the t-th iteration factor graph, we can get:
在获取因子图中双向边对应的置信度集合的情况下，对ΦK,J(t)进行排序，然后选择置信 度较大的G条双向边设置为单向边，更新此次迭代的因子图F(t)，其中参数G表示每次迭代 将双向边设置为单向边的数量，并且G满足以下条件In the case of obtaining the confidence set corresponding to the bidirectional edge in the factor graph, sort ΦK,J (t), and then select G bidirectional edges with higher confidence to set as unidirectional edges, and update the factor of this iteration Graph F(t), where the parameter G represents the number of bidirectional edges set to unidirectional edges in each iteration, and G satisfies the following conditions
G＝|Bs(t-1)|-|Bs(t)|G＝|Bs (t-1)|-|Bs (t)|
其中1<t≤tmax且G∈{0,1,…,Kdr}。G的取值将决定了SCMA系统多用户检测的复杂度，所 以本发明能够根据G值大小自适应地在复杂度和译码性能之间进行折中。特别地，当G＝0时 因子图中不存在单向边，所提DFG-MPA算法等效于原始MPA算法。where 1<t≤tmax and G∈{0,1,...,Kdr }. The value of G will determine the complexity of multi-user detection in the SCMA system, so the present invention can adaptively compromise between complexity and decoding performance according to the value of G. In particular, when G=0, there is no unidirectional edge in the factor graph, and the proposed DFG-MPA algorithm is equivalent to the original MPA algorithm.
MPA是一种消息更新传播算法，它是一种利用因子图模型来求解概率推理问题的实用技 术，适合在低密度因子图中进行迭代运算，MPA算法的计算复杂度主要集中在资源节点处的 运算操作。假设最大迭代次数为tmax，在迭代过程中原始MPA算法一共需要计算tmaxKdr次DFG-MPA算法一共需要计算次由于|Bs(t)|≤Kdr，所以DFG-MPA算 法能够降低计算复杂度。本发明通过减少从资源节点到用户节点消息更新的边的数量，从而 实现降低MPA检测器的计算复杂度。那么原始MPA和DFG-MPA两种多用户检测算法的复 杂度表达式如下所示：MPA is a message update propagation algorithm. It is a practical technique for solving probabilistic reasoning problems using factor graph models. It is suitable for iterative operations in low-density factor graphs. The computational complexity of the MPA algorithm is mainly concentrated in resource nodes. arithmetic operation. Assuming that the maximum number of iterations is tmax , the original MPA algorithm needs to calculate tmax Kdr times in the iterative process The DFG-MPA algorithm needs to calculate a total of Second-rate Since |Bs (t)|≤Kdr , the DFG-MPA algorithm can reduce the computational complexity. The present invention reduces the calculation complexity of the MPA detector by reducing the number of edges for message update from resource nodes to user nodes. Then the complexity expressions of the original MPA and DFG-MPA two multi-user detection algorithms are as follows:
基于上述所构建模型及定义，本发明提供了一种基于动态因子图的MPA算法对上行链路 SCMA系统多用户检测复杂度进行降低，SCMA系统利用DFG-MPA算法进行译码流程如图 4所示。Based on the above-mentioned constructed model and definition, the present invention provides a dynamic factor graph-based MPA algorithm to reduce the multi-user detection complexity of the uplink SCMA system, and the SCMA system uses the DFG-MPA algorithm to perform decoding as shown in Figure 4 Show.
实施例Example
本实施例采用Matlab仿真平台进行实验。In this embodiment, a Matlab simulation platform is used for experiments.
本例的目的通过如下步骤实现：The purpose of this example is achieved through the following steps:
S1、设置SCMA系统参数。本例中系统参数如下：用户数J＝6，时频资源数K＝4，仿真信道为AWGN，过载因子为λ＝150％，仿真次数为106，最大迭代次数为tmax，所采用的码本 如下表所示：S1. Setting SCMA system parameters. The system parameters in this example are as follows: the number of users J=6, the number of time-frequency resources K=4, the simulated channel is AWGN, the overload factor is λ=150%, the number of simulations is 106 , and the maximum number of iterations is tmax . The codebook is shown in the table below:
S2、初始化消息传递概率值为：S2. The initial message delivery probability value is:
其中，是指因子图中用户节点uj到资源节点rk的消息更新过程，其中t是迭代次数， xj＝(x1,j,x2,j…,xK,j)T∈CK表示第j个用户发送的码字；in, refers to the message update process from user node uj to resource node rk in the factor graph, where t is the number of iterations, xj =(x1,j ,x2,j …,xK,j )T ∈ CK means The code word sent by the jth user;
S3、设置迭代次数初始值t＝1；S3. Set the initial value of the number of iterations t=1;
S4、进入消息迭代更新，判断t是否超出步骤S1中设定的最大迭代次数，若t＝1，则进 入步骤S6，若1<t≤tmax，则进入步骤S5，若t>tmax，则终止消息迭代更新并进入步骤S8；S4. Enter message iterative update, judge whether t exceeds the maximum number of iterations set in step S1, if t=1, then enter step S6, if 1<t≤tmax , then enter step S5, if t>tmax , Then terminate the iterative update of the message and enter step S8;
S5、否则计算因子图中每条双向边的置信度，并得到置信度集合ΦK,J(t)，然后对ΦK,J(t) 中的置信度元素进行排序，将置信度较大的G条双向边设置成单向边，更新此次消息迭代所 需要的因子图F(t)；S5. Otherwise, calculate the confidence degree of each bidirectional edge in the factor graph, and obtain the confidence degree set ΦK, J (t), and then sort the confidence degree elements in ΦK, J (t), and set the confidence degree The G bidirectional edges of G are set as unidirectional edges, and the factor graph F(t) required for this message iteration is updated;
S6、根据更新后的因子图F(t)进行资源节点与用户节点的消息更新，具体为：S6. Update the message of the resource node and the user node according to the updated factor graph F(t), specifically:
S61、资源节点的消息更新，对所有资源节点rk计算：S61. The message update of resource nodes is calculated for all resource nodes rk :
其中，σ是高斯分布的标准差，yk是基站在第k个时频资源所接收到的信号，h为信道，是指因子图中用户节点up到资源节点rk的消息更新过程；Among them, σ is the standard deviation of the Gaussian distribution, yk is the signal received by the base station at the kth time-frequency resource, h is the channel, refers to the message update process from user node up to resource node rk in the factor graph;
S62、用户节点的消息更新，对所有用户节点uj计算：S62. The message update of the user node is calculated for all user nodes uj :
S7、迭代次数进行更新t＝t+1，回到步骤S4；S7, update the number of iterations t=t+1, return to step S4;
S8、利用更新后的概率分布，完成多用户检测，对用户进行软判决输出为：S8. Using the updated probability distribution, the multi-user detection is completed, and the soft-judgment output for the user is:
采用本发明所述方法进行仿真测试。首先比较原始MPA和DFG-MPA的BER性能，其次比较原始MPA和DFG-MPA的收敛性能，最后比较原始MPA和DFG-MPA的计算复杂度。 随着参数G的取值不断增加，GT-MPA降低计算复杂度的比率越来越大。综上所述，本发明 首次将动态因子图技术用于SCMA系统进行多用户检测，所提DFG-MPA能够大幅度降低计 算复杂度，并通过选择合适的参数G值保证BER性能基本没有损失。The simulation test is carried out by adopting the method of the present invention. Firstly compare the BER performance of original MPA and DFG-MPA, secondly compare the convergence performance of original MPA and DFG-MPA, and finally compare the computational complexity of original MPA and DFG-MPA. As the value of the parameter G increases, the ratio of GT-MPA to reduce the computational complexity becomes larger and larger. To sum up, the present invention uses the dynamic factor graph technology for the SCMA system for multi-user detection for the first time, and the proposed DFG-MPA can greatly reduce the computational complexity, and ensure that the BER performance is basically not lost by selecting the appropriate parameter G value.
本领域的技术人员将会意识到，这里所述的实例为了帮助读者理解本发明的原理，是以SCMA系统4×6的码本(J＝6，K＝4)为例阐述的，本方法可扩展为任意SCMA的码本形式的 应用中，且应被理解为本发明的保护范围并不局限于这样的特别陈述和实施案例；本领域的 普通技术人员可以根据本发明公开的这些技术启示做出各种不脱离本发明实质的其它各种具 体变形和组合，这些变形和组合仍然在本发明的保护范围内。Those skilled in the art will appreciate that the examples described here are set forth as an example with the codebook (J=6, K=4) of SCMA system 4×6 in order to help readers understand the principle of the present invention. It can be extended to the application of any SCMA codebook form, and it should be understood that the protection scope of the present invention is not limited to such special statements and implementation cases; those of ordinary skill in the art can Various other specific modifications and combinations can be made without departing from the essence of the present invention, and these modifications and combinations are still within the protection scope of the present invention.
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810019294.0ACN108199805A (en) | 2018-01-09 | 2018-01-09 | A kind of method for reducing Sparse Code multi-address system decoding complexity |
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810019294.0ACN108199805A (en) | 2018-01-09 | 2018-01-09 | A kind of method for reducing Sparse Code multi-address system decoding complexity |
| Publication Number | Publication Date |
|---|---|
| CN108199805Atrue CN108199805A (en) | 2018-06-22 |
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810019294.0APendingCN108199805A (en) | 2018-01-09 | 2018-01-09 | A kind of method for reducing Sparse Code multi-address system decoding complexity |
| Country | Link |
|---|---|
| CN (1) | CN108199805A (en) |
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109831281A (en)* | 2019-03-21 | 2019-05-31 | 西安电子科技大学 | A kind of low complex degree Sparse Code multiple access system multi-user test method and device |
| CN110098883A (en)* | 2019-05-09 | 2019-08-06 | 电子科技大学 | A kind of multi-user test method based on clipped noise auxiliary |
| CN112104582A (en)* | 2020-11-09 | 2020-12-18 | 电子科技大学 | I/Q domain modulation method, double domain modulation method and multiple access communication method |
| CN115514450A (en)* | 2021-06-22 | 2022-12-23 | 中国联合网络通信集团有限公司 | Decoding method and device for a sparse code division multiple access system |
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104798317A (en)* | 2012-11-16 | 2015-07-22 | 华为技术有限公司 | Systems and methods for sparse code multiple access |
| CN105554865A (en)* | 2015-11-09 | 2016-05-04 | 哈尔滨工业大学 | MIMO-SCMA system downlink design method based on STBC |
| CN105721106A (en)* | 2016-01-27 | 2016-06-29 | 电子科技大学 | Multiuser detection method based on serial strategy for SCMA (Sparse Code Multiple Access) uplink communication system |
| CN107181567A (en)* | 2017-05-12 | 2017-09-19 | 电子科技大学 | A kind of low complex degree MPA algorithms based on thresholding |
| CN107210807A (en)* | 2015-02-27 | 2017-09-26 | 华为技术有限公司 | Low complex degree SCMA/LDS detecting systems and method |
| CN107276725A (en)* | 2017-07-31 | 2017-10-20 | 北京交通大学 | Multi-user test method in SCMA systems |
| CN107483151A (en)* | 2017-08-11 | 2017-12-15 | 北京交通大学 | A Serial Multi-user Dynamic Iteration Method Based on SCMA System |
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104798317A (en)* | 2012-11-16 | 2015-07-22 | 华为技术有限公司 | Systems and methods for sparse code multiple access |
| CN107210807A (en)* | 2015-02-27 | 2017-09-26 | 华为技术有限公司 | Low complex degree SCMA/LDS detecting systems and method |
| CN105554865A (en)* | 2015-11-09 | 2016-05-04 | 哈尔滨工业大学 | MIMO-SCMA system downlink design method based on STBC |
| CN105721106A (en)* | 2016-01-27 | 2016-06-29 | 电子科技大学 | Multiuser detection method based on serial strategy for SCMA (Sparse Code Multiple Access) uplink communication system |
| CN107181567A (en)* | 2017-05-12 | 2017-09-19 | 电子科技大学 | A kind of low complex degree MPA algorithms based on thresholding |
| CN107276725A (en)* | 2017-07-31 | 2017-10-20 | 北京交通大学 | Multi-user test method in SCMA systems |
| CN107483151A (en)* | 2017-08-11 | 2017-12-15 | 北京交通大学 | A Serial Multi-user Dynamic Iteration Method Based on SCMA System |
| Title |
|---|
| XINYING,MA: "Low complexity Detection Based on Dynamic Factor Graph for SCMA System", 《IEEE COMMUNICATIONS LETTERS》* |
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109831281A (en)* | 2019-03-21 | 2019-05-31 | 西安电子科技大学 | A kind of low complex degree Sparse Code multiple access system multi-user test method and device |
| CN110098883A (en)* | 2019-05-09 | 2019-08-06 | 电子科技大学 | A kind of multi-user test method based on clipped noise auxiliary |
| CN112104582A (en)* | 2020-11-09 | 2020-12-18 | 电子科技大学 | I/Q domain modulation method, double domain modulation method and multiple access communication method |
| CN115514450A (en)* | 2021-06-22 | 2022-12-23 | 中国联合网络通信集团有限公司 | Decoding method and device for a sparse code division multiple access system |
| CN115514450B (en)* | 2021-06-22 | 2024-06-04 | 中国联合网络通信集团有限公司 | Decoding method and device for sparse code division multiple access system |
| Publication | Publication Date | Title |
|---|---|---|
| CN109716662B (en) | Method and device for encoding data using polar codes | |
| CN115378524B (en) | Optimized expected propagation detection method and signal detection device | |
| CN108199805A (en) | A kind of method for reducing Sparse Code multi-address system decoding complexity | |
| CN105811998A (en) | Density evolution based polarization code constructing method and polarization code coding and decoding system | |
| CN110326221A (en) | A method of for generating ordered sequence for polarization code | |
| CN110535475A (en) | A kind of layered self-adapting normalization Min-Sum decoding algorithm | |
| CN106130688B (en) | A low-complexity sparse code multiple access detection method | |
| CN107864029A (en) | A kind of method for reducing Multiuser Detection complexity | |
| CN107565978B (en) | BP decoding method based on Tanner graph edge scheduling strategy | |
| CN106330199A (en) | SCMA and LDPC joint detection and decoding algorithm and device based on factor graph | |
| JP2016512417A (en) | Low complexity receiver and method for low density signature modulation | |
| WO2016029405A1 (en) | Decoding method and device based on multi-objective genetic | |
| CN112564716A (en) | PC-SCMA system joint decoding method based on pruning iteration | |
| CN107508657A (en) | A kind of SCMA multi-user test methods based on weight factor message transmission | |
| CN102594367A (en) | Low-complexity dynamic asynchronous BP decoding method | |
| Chen et al. | Memory AMP for generalized MIMO: Coding principle and information-theoretic optimality | |
| CN108737022A (en) | Low complex degree SCMA coding/decoding methods based on quantum calculation and device | |
| CN109831281B (en) | Multi-user detection method and device for low-complexity sparse code multiple access system | |
| Miao et al. | A low complexity multiuser detection scheme with dynamic factor graph for uplink SCMA systems | |
| CN110430011A (en) | The BATS code encoding method of rule-based variable node degree distribution | |
| CN112003680B (en) | Low-complexity multi-user detection method in SCMA system | |
| Wang et al. | An improved SC flip decoding algorithm of polar codes based on genetic algorithm | |
| CN115913454B (en) | Punishment dual decomposition method-based depth expansion channel decoding method and punishment dual decomposition method-based depth expansion channel decoding device | |
| CN107181567A (en) | A kind of low complex degree MPA algorithms based on thresholding | |
| CN106877883A (en) | A LDPC decoding method and device based on a restricted Boltzmann machine |
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| WD01 | Invention patent application deemed withdrawn after publication | Application publication date:20180622 | |
| WD01 | Invention patent application deemed withdrawn after publication |