CN105721106A - Multiuser detection method based on serial strategy for SCMA (Sparse Code Multiple Access) uplink communication system - Google Patents

Abstract

The invention provides a multiuser detection method based on a serial strategy for an SCMA (Sparse Code Multiple Access) uplink communication system, and belongs to the field of signal detection of wireless communication systems. The method is characterized in that nodes in a conventional SCMA factor graph are divided into J groups (J is a quantity of user nodes); each group comprises one user node and all resource nodes connected with the user node; and all nodes (one user node and all the resource nodes which are connected with the user node) of each group are updated in sequence in each iteration process. According to the multiuser detection method, already-updated node messages are utilized in each iteration process, so that a utilization ratio of the already-updated node messages can be effectively increased. The BER (Bit Error Rate) performance of the method is far superior to the BER performance of a method in the prior art under the condition of less iteration times. The calculation complexity of the method is much lower than the calculation complexity of the method in the prior art under the condition of hardly any loss of the BER performance.

Description

Multi-user detection method of SCMA (Single chip multiple Access) uplink communication system based on serial strategy

Technical Field

The invention belongs to the field of signal detection of a wireless communication system, and relates to a Sparse Code Multiple Access (SCMA) uplink communication system multi-user detection method with high convergence rate and low complexity, in particular to an MPA detection method based on a serial strategy, which is used for solving the multi-user detection problem of an alternative multiple access technology of 5G (fifth generation mobile communication) mobile communication.

Background

Multiple access is one of the core technologies of the physical layer of wireless communication, which enables a base station to distinguish and simultaneously serve multiple end users. In order to meet the requirements of 5G (fifth generation mobile communication) on large capacity, mass connection, low delay access and the like, an invention patent 'system and method for user sparse code multiple access' with application number 201380059380.X proposes an SCMA (sparse code multiple access) technology, fig. 1 is an SCMA uplink communication system model, a part of zero elements are in SCMA spread spectrum code words, baseband data of each user is subjected to nonzero-bit spread spectrum modulation only on a small number of chips, each bit spread spectrum chip is subjected to nonzero-bit spread spectrum modulation only by a small number of users, an SCMA encoder selects a codebook for each data layer or user in a predefined codebook set, then directly maps data bits into corresponding code words based on the selected codebook, finally, non-orthogonal superposition is carried out on the code words of a plurality of data layers or users, and a receiving end carries out a multi-user detection technology based on a confidence propagation algorithm on a received signal, namely, the parallel MPA detection method, fig. 2 is an SCMA factor graph of the method, and the parallel MPA detection method updates all resource nodes first and then all user nodes in each iteration process.

The invention provides an MPA detection method based on a parallel strategy in a user sparse code multiple access system and a user sparse code multiple access method, which specifically comprises the following steps:

step 1, initialization: when iteration starts, the user has no prior information, so the user node u_jTo resource node c_kThe message of (1) is:j＝1,2,...,J，k＝1,2,...,K，u_jis the jth user node, c_kIs the k-th resource node, M is the number of codewords of the codebook,for the user node u at the start of the iteration_jTo resource node c_kThe message of (2);

step 2, setting the maximum iteration number as t_max；

Step 3, in the process of the t iteration, the messages of the resource nodes and the user nodes are updated in parallel, namely all the resource nodes c are updated firstly_kTo user node u_jOf a message $M_{c_k\→u_j}^t\left(x_j\right)=\underset{~x_j}{\mathrm{\Σ}}\left\{\exp \left(-\frac{1}{2{\mathrm{\σ}}^2}\left|\right|y_k-\underset{i\∈{\mathrm{\ϵ}}_k}{\mathrm{\Σ}}{h}_{k,i}{x}_{k,i}|{|}^2\×\underset{l\∈{\mathrm{\ϵ}}_k\backslash j}{\mathrm{\Π}}{M}_{u_l\→c_k}^{t-1}\left(x_l\right)\right)\right\},\∀j,\∀k;$ t＝1,2,...,t_maxIs the t (t ═ 1, 2.., t_max) Resource node c obtained by calculation in secondary iteration process_kTo user node u_jMessage of x_j＝(x_1,j,...,x_K,j)^ΤIs the SCMA codeword, x, of the jth user_l＝(x_1,l,...,x_K,l)^ΤIs the SCMA codeword for the l-th user, y_kFor the signal received at the kth resource in the received signal y,(F is the sparse spreading matrix of the transmitter SCMA encoder) to resource node c_kA set of user nodes; x is the number of_k,iK resource, h, of a codeword representing an i-th user_k,iIs h_i＝(h_1,i,h_2,i,...,h_K,i)^ΤThe k-th element of (1), h_iChannel vector for ith user; then all the user sections are updatedPoint u_jTo resource node c_kThe message of (2):completing one iteration;for connecting to user node u_jA set of resource nodes;

step 4, judging t is more than t_maxIf yes, executing step 5; if not, making t equal to t +1, returning to the step 3, and performing the next iteration;

step 5, passing t_maxAnd exiting the loop after the secondary iteration, and calculating the code word message:

$Q\left(x_j\right)=\underset{q\∈{\mathrm{\ζ}}_j}{\mathrm{\Π}}{M}_{c_q\→u_j}\left(x_j\right);$

step 6, soft information of data bits is obtained through calculation

${\mathrm{LLR}}_j=\log \frac{P\left(b_j=1\right)}{P\left(b_j=0\right)}=\log \frac{\underset{b_j=1}{\mathrm{\Σ}}Q\left(x_j\right)}{\underset{b_j=0}{\mathrm{\Σ}}Q\left(x_j\right)};$

b_jA data bit for a jth user;

step 7, data bit judgment (hard judgment) $\widehat{U_j}=\left\{\begin{array}{cc}1& {\mathrm{LLR}}_j>0\\ 0& {\mathrm{LLR}}_j\≤0\end{array}\right..$

In the MPA detection method based on the parallel strategy, the detection accuracy and the maximum iteration number t_maxAnd message propagation. However, in practical engineering applications, as the number of iterations increases, the hardware computation complexity of the communication system becomes higher and higher, and thus the requirement on hardware becomes higher and higher; and all resource node-to-user node messages in the method are updatedIs based on the message from the user node to the resource node obtained by the t-1 iterationUpdating of all user node to resource node messagesUpdated based on the current iterationAlthough it is used forMaking good use of current iteration updatesBut do notIs generated based on the last iterationThe message updated from the user node to the resource node in the current iteration is not utilized, so that the convergence rate is low. In summary, although the parallel MPA detection method can effectively improve the probability of correct decoding, it cannot meet the requirement of 5G on high speed and high efficiency due to its higher computational complexity and lower convergence rate.

Disclosure of Invention

The invention provides a serial strategy-based multi-user detection method for an SCMA uplink communication system, aiming at the defects in the background art. The nodes in the traditional SCMA factor graph are divided into J groups (J is the number of user nodes), each group is a user node and all resource nodes connected with the user node, and in each iteration process, all the nodes (one user node and all the resource nodes connected with the user node) in each group are updated in sequence; the invention utilizes the updated node message in each iteration process, can effectively improve the utilization rate of the updated node message, and has low computational complexity and excellent BER performance.

The technical scheme of the invention is as follows:

a SCMA uplink communication system multi-user detection method based on serial strategy includes initialization, resource node and user node message updating, data bit soft information calculation, data bit decision step; the method is characterized in that the updating of the information of the resource nodes and the user nodes is completed by multiple iterations, in each iteration process, one user node and all the resource nodes connected with the user node are taken as a group, the user nodes and the resource nodes are divided into J groups, and J is the number of the user nodes; and then, updating all the messages from the resource nodes to the user nodes and the messages from the user nodes to the resource nodes in each group in sequence, namely finishing the iterative process.

A multi-user detection method of an SCMA uplink communication system based on a serial strategy specifically comprises the following steps:

step 1, initialization: when the SCMA starts iteration, the user has no prior information, so the user node u_jTo resource node c_kThe message of (1) is:j＝1,2,...,J，k＝1,2,...,K，u_jis the jth user node, c_kIs the k-th resource node, M is the number of codewords of the codebook,for the user node u at the start of the iteration_jTo resource node c_kThe message of (2);

step 2, setting the maximum iteration number as t_max；

Step 3, in the process of the t iteration, firstly, taking the jth user node and all resource nodes connected with the jth user node as a group, and dividing the user node and the resource nodes into J groups, wherein J is 1,2₁,γ₂,...,γ_JGroup (d); then, the gamma is calculated₁All resource node to user node messages of a group $M_{c_k\&RightArrow;u_j}^t\left(x_j\right)=\underset{~x_j}{\mathrm{\&Sigma;}}\left\{\exp \left(-\frac{1}{2{\mathrm{\&sigma;}}^2}\left|\right|y_k-\underset{i\&Element;{\mathrm{\&epsiv;}}_k}{\mathrm{\&Sigma;}}{h}_{k,i}{x}_{k,i}|{|}^2\right)\&times;\underset{l<j}{\underset{l\&Element;{\mathrm{\&epsiv;}}_k\backslash j}{\mathrm{\&Pi;}}}{M}_{u_l\&RightArrow;c_k}^t\left(x_l\right)\underset{l>j}{\underset{l\&Element;{\mathrm{\&epsiv;}}_k\backslash j}{\mathrm{\&Pi;}}}{M}_{u_l\&RightArrow;c_k}^{t-1}\left(x_l\right)\right\}$ And user node to resource node messagesThen calculating gamma in turn₂,...,γ_JThe message from the resource node to the user node of the group and the message from the user node to the resource node are completed by one iteration; x is the number of_j＝(x_1,j,...,x_K,j)^ΤIs the SCMA codeword, x, of the jth user_l＝(x_1,l,...,x_K,l)^ΤIs the SCMA codeword for the l-th user, y_kFor the signal received at the kth resource in the received signal y,(F is the sparse spreading matrix of the transmitter SCMA encoder) to resource node c_kA set of user nodes; x is the number of_k,iK resource, h, of a codeword representing an i-th user_k,i∈h_i＝(h_1,i,h_2,i,...,h_K,i)^ΤHi is the channel vector for user i,for connecting to user node u_jA set of resource nodes;

step 4, judging t is more than t_maxIf yes, executing step 5; if not, making t equal to t +1, returning to the step 3, and performing the next iteration;

step 5, passing t_maxAnd exiting the loop after the secondary iteration, and calculating the code word message:

$Q\left(x_j\right)=\underset{q\&Element;{\mathrm{\&zeta;}}_j}{\mathrm{\&Pi;}}{M}_{c_q\&RightArrow;u_j}\left(x_j\right);$

step 6, calculating soft information of data bit

${\mathrm{LLR}}_j=\log \frac{P\left(b_j=1\right)}{P\left(b_j=0\right)}=\log \frac{\underset{b_j=1}{\mathrm{\&Sigma;}}Q\left(x_j\right)}{\underset{b_j=0}{\mathrm{\&Sigma;}}Q\left(x_j\right)};$

b_jA data bit for a jth user;

step 7, data bit judgment (hard judgment) $\widehat{U_j}=\left\{\begin{array}{cc}1& {\mathrm{LLR}}_j>0\\ 0& {\mathrm{LLR}}_j\&le;0\end{array}\right..$

The invention has the beneficial effects that: the invention provides a multi-user detection method of an SCMA uplink communication system based on a serial strategy, wherein in each iteration process, all nodes are grouped into J groups according to the mode that each user node and all resource nodes connected with the user node are taken as one group, and the nodes in each group are updated in sequence. The updated node message is utilized in each iteration process, so that the utilization rate of the updated node message can be effectively improved; under the condition of less iteration times, the BER performance of the method is far superior to that of the background technology method; the computational complexity of the present invention is much lower than that of the background art method, with little loss in BER performance.

Drawings

Fig. 1 is a SCMA uplink communication system model proposed in the background art;

FIG. 2 is a factor graph of an SCMA uplink system in the background art;

FIG. 3 is a factor graph of the SCMA uplink system of the present invention;

FIG. 4 is a graph comparing the BER performance of the method of the present invention with that of the background art;

FIG. 5 is a comparison of computational complexity between the method of the present invention and the prior art method.

Detailed Description

The technical scheme of the invention is detailed below by combining the accompanying drawings and the embodiment.

According to the transmission condition and the program of the detection process, the following parameters are initialized:

the number J of user nodes is 6, the number M of code words of the codebook is 4, the length K of the code words is 4, and the overload rate of the systemThe channel model is AWGN (white Gaussian noise) channel, and the sparse spreading matrix is $F=\left[\begin{array}{cccccc}0& 1& 1& 0& 1& 0\\ 1& 0& 1& 0& 0& 1\\ 0& 1& 0& 1& 0& 1\\ 1& 0& 0& 1& 1& 0\end{array}\right];$ User information bits are mapped into corresponding code words x through an SCMA encoder, all user code words are transmitted through a channel after being superposed, and a receiving signal of a receiving endx_j＝(x_1,j,x_2,j,...,x_K,j)^ΤIs the SCMA codeword for user j, h_j＝(h_1,j,h_2,j,...,h_K,j)^ΤChannel vector for user j, n-CN (0, σ)²I) is Gaussian noise; signal y received at kth resource receiving signal y_kExpressed as:wherein K1, 2, and K, J1, 2.

The codebook of each user at the sending end is as follows:

user 1's codebook is $CB1=\left[\begin{array}{cccc}0& 0& 0& 0\\ -0.1815-0.1318i& -0.6351-0.4615i& 0.6351+0.4615i& 0.1815+0.1318i\\ 0& 0& 0& 0\\ 0.7851& -0.2243& 0.2243& -0.7851\end{array}\right],$ User 2's codebook is $CB2=\left[\begin{array}{cccc}0.7851& -0.2243& 0.2243& -0.7851\\ 0& 0& 0& 0\\ -0.1815-0.1318i& -0.6351-0.4615i& 0.6351+0.4615i& 0.1815+0.1318i\\ 0& 0& 0& 0\end{array}\right]$ User 3's codebook is $CB3=\left[\begin{array}{cccc}-0.6351+0.4615i& 0.1815-0.1318i& -0.1815+0.1318i& 0.6351-0.4615i\\ 0.1392-0.1759i& 0.4873-0.6156i& -0.4873+0.6156i& -0.1392+0.1759i\\ 0& 0& 0& 0\\ 0& 0& 0& 0\end{array}\right]$ User 4's codebook is $CB4=\left[\begin{array}{cccc}0& 0& 0& 0\\ 0& 0& 0& 0\\ 0.7851& -0.2243& 0.2243& -0.7851\\ -0.0055-0.2242i& -0.0193-0.7848i& 0.0193+0.7848i& 0.0055+0.2242i\end{array}\right]$ User 5's codebook is $CB5=\left[\begin{array}{cccc}-0.0055-0.2242i& -0.0193-0.7848i& 0.0193+0.7848i& 0.0055+0.2242i\\ 0& 0& 0& 0\\ 0& 0& 0& 0\\ -0.6351+0.4615i& 0.1815-0.1318i& -0.1815+0.1318i& 0.6351-0.4615i\end{array}\right]$ User 6 has a codebook of $CB6=\left[\begin{array}{cccc}0& 0& 0& 0\\ 0.7851& -0.2243& 0.2243& -0.7851\\ 0.1392-0.1759i& 0.4873-0.6156i& -0.4873+0.6156i& -0.1392+0.1759i\\ 0& 0& 0& 0\end{array}\right]$

Fig. 3 is an SCMA factor graph of the multi-user detection method of the SCMA uplink communication system based on the serial policy of the present invention, which is different from the factor graph of the background art method in that: all resource nodes and user nodes can be divided into J groups, which are respectively marked as gamma₁,γ₂,...,γ_JGroups, during each iteration, sequentially updating gamma₁,γ₂,...,γ_JAll nodes in the group (i.e., each user node and all resource nodes connected to the user node).

The multi-user detection method of the SCMA uplink communication system based on the serial strategy at the receiver comprises the following steps:

step 1, iterative detection is started, and no prior user information exists, so that the probability that each user acquires any code word in the corresponding codebook is the same, and the information of the user node is

Step 2, entering an iterative loop, referring to a factor graph of the MPA detection method based on the serial strategy in FIG. 3, dividing all resource nodes and user nodes into J groups, and marking the serial number of each group as gamma₁,γ₂,...,γ_J；

Step 3, updating the first group gamma₁All resource nodes and user nodes in the network; then judging whether all the J groups of nodes are updated, if so, executing the step 4, otherwise, continuously updating the next group of nodes;

step 4, judging t is more than t_maxIf yes, executing step 5; if not, making t equal to t +1, returning to the step 2, and performing the next iteration;

step 5, according to the data output by iterationSource nodeComputing codeword messages $Q\left(x_j\right)=\underset{q\&Element;{\mathrm{\&zeta;}}_j}{\mathrm{\&Pi;}}{M}_{c_q\&RightArrow;u_j}\left(x_j\right);$

Step 6, calculating the likelihood value of the original data bit according to the code word information obtained in the step 5

${\mathrm{LLR}}_j=\log \frac{P\left(b_j=1\right)}{P\left(b_j=0\right)}=\log \frac{\underset{b_j=1}{\mathrm{\&Sigma;}}Q\left(x_j\right)}{\underset{b_j=0}{\mathrm{\&Sigma;}}Q\left(x_j\right)};$

Step 7, finally, hard-decision is made on the original information bit, if LLR is available_j> 0, data bits are decidedIf LLR_jLess than or equal to 0, and judging data bit

The parallel MPA method in the background art and the multi-user detection method of the SCMA uplink communication system based on the serial strategy in the embodiment of the invention are utilized to carry out simulation comparison analysis on BER error code performance and computational complexity in an AWGN channel by Matlab, and simulation results are shown in fig. 4 and fig. 5. As can be seen from FIG. 4, at E_b/N₀The BER for the background technique of 2 iterations is 1.2 × 10 at 12dB^-3The BER value of the present invention is 2.5 × 10^-4The BER performance of the invention is improved by one order of magnitude; secondly, the BER performance of 2 iterations in the method is close to the performance of 6 iterations in the background art method, which shows that the method can effectively reduce the complexity of decoding. Due to the invention and the background art IThe computational complexity of the iterations is the same, so it makes sense to analyze the computational complexity of the invention for 2 iterations and the computational complexity of the background art for 6 iterations fig. 5 BER 1.2 × 10^-3Under the condition of (1), namely under the condition that the detection performances of 2 iterations of the method are the same as those of 6 iterations of the background technology, the calculation complexity of the method of the background technology is 3 times that of the method of the invention, so that the method of the invention can realize the performances of 6 iterations of the background technology by adopting 2 iterations, and the calculation complexity of the system is greatly reduced.

Claims (2)

1. A SCMA uplink communication system multi-user detection method based on serial strategy includes initialization, resource node and user node message updating, data bit soft information calculation, data bit decision step; the method is characterized in that the updating of the information of the resource nodes and the user nodes is completed by multiple iterations, in each iteration process, one user node and all the resource nodes connected with the user node are taken as one group, the user nodes and the resource nodes are divided into J groups, J is the number of the user nodes, and then all the information from the resource nodes to the user nodes and the information from the user nodes to the resource nodes in each group are updated in sequence, namely, the process of one iteration is completed.

2. A multi-user detection method of an SCMA uplink communication system based on a serial strategy comprises the following steps:

step 1, initialization: when iteration starts, the user has no prior information, so the user node u_jTo resource node c_kThe message of (1) is:j＝1,2,...,J，k＝1,2,...,K，u_jis the jth user node, c_kThe number of the k resource nodes is M, and the number of the code words of the codebook is M;

step 2, setting the maximum iteration number as t_max；

Step 3, in the process of the t iteration, firstly, taking the jth user node and all resource nodes connected with the jth user node as a group, and dividing the user node and the resource nodes into J groups, wherein J is 1,2₁,γ₂,...,γ_JGroup (d); then, the gamma is calculated₁All resource node to user node messages of a group $M_{c_k\&RightArrow;u_j}^t\left(x_j\right)=\underset{~x_j}{\mathrm{\&Sigma;}}\left\{\exp (-\frac{1}{2{\mathrm{\&sigma;}}^2}||y_k-\underset{i\&Element;{\mathrm{\&epsiv;}}_k}{\mathrm{\&Sigma;}}{h}_{k,i}{x}_{k,i}|{|}^2)\underset{l<j}{\underset{l\&Element;{\mathrm{\&epsiv;}}_k\backslash j}{\mathrm{\&Pi;}}}{M}_{u_l\&RightArrow;c_k}^t\left(x_l\right)\underset{l>j}{\underset{l\&Element;{\mathrm{\&epsiv;}}_k\backslash j}{\mathrm{\&Pi;}}}{M}_{u_l\&RightArrow;c_k}^{t-1}\left(x_l\right)\right\}$ And user node to resource node messagesFinally, gamma is calculated in sequence₂,...,γ_JThe message from the resource node to the user node of the group and the message from the user node to the resource node are completed by one iteration; wherein,to connect to resource node c_kUser node set of h_k，i∈h_i＝(h_1，i，h_2，i，...，h_k，i)^T，h_iFor the channel vector of the user i,for connecting to user node u_jA set of resource nodes;

step 4, judging t is more than t_maxIf yes, executing step 5; if not, making t equal to t +1, returning to the step 3, and performing the next iteration;

step 5, passing t_maxAnd exiting the loop after the secondary iteration, and calculating the code word message:

step 6, calculating soft information of data bits:

${\mathrm{LLR}}_j=\log \frac{P(b_j=1)}{P(b_j=0)}=\log \frac{\underset{b_j=1}{\&Sigma;}Q\left(x_j\right)}{\underset{b_j=0}{\&Sigma;}Q\left(x_j\right)};$

b_ja data bit for a jth user;

step 7, data bit judgment $\widehat{U_j}=\left\{\begin{array}{cc}1& {\mathrm{LLR}}_j>0\\ 0& {\mathrm{LLR}}_j\&le;0\end{array}\right..$