CN102723975B - Signal detection method and device of MIMO (multiple input multiple output) system - Google Patents

Abstract

The present invention relates to the technical field of signal detection, and discloses a signal detection method and device for a MIMO system. The method includes: S1, using a known channel matrix to construct a search tree; S2, searching the Nth layer of the search tree; S21, Divide the K nodes into M groups; S22, according to the principle that the number of expanded child nodes of nodes in each group is equal, and the number of expanded child nodes of nodes in different groups decreases sequentially from M, expand the child nodes of the Nth layer node; S23 1. Sorting according to the Euclidean distance between the part of the received vector and the part of the sent vector corresponding to all expanded sub-nodes, and keeping K sub-nodes whose Euclidean distance is smaller than the threshold for the search of the next layer; S3, according to steps S21-S23 All layers are searched in the same way, and the path with the smallest Euclidean distance is found in the path composed of child nodes as the search result. The invention reduces the complexity of signal detection and improves the detection efficiency of received data.

Description

Signal detection method and device for MIMO system

technical field

The present invention relates to the technical field of signal detection, in particular to a signal detection method and device for a MIMO system.

Background technique

Multiple-input multiple-output (MIMO) technology can double the transmission rate without increasing the frequency band. Today, frequency resources are increasingly tight. Therefore, MIMO technology is considered to be the most important physical layer technology in the next generation of broadband wireless communication technology. one.

In a MIMO system, in order to improve the reliability of source information transmission, at the sending end, the signal to be transmitted is firstly subjected to channel coding that can provide error correction capabilities, and then space-time/space-frequency/space-time-frequency coding, and then multiple The transmitting antennas transmit at the same time or in a certain time sequence. At the receiving end, the signals from the sending end are received by multiple receiving antennas simultaneously or in a certain time sequence, and space-time/space-frequency/space-time-frequency decoding and channel decoding are performed in sequence, so that the decoding result is used as the signal to be sent original information.

The channel decoding performed by the receiving end is essentially the detection of the received signal, that is, the optimal signal is detected from the received signal as the decoding result. However, the signal detection technology of MIMO technology is facing a huge problem. Although the maximum likelihood (ML) algorithm is optimal in the sense of minimum error probability, its computational complexity is large, which is caused by the number of antennas and the modulation order. The exponential form of , for example, when 16QAM (16-order quadrature amplitude modulation) modulation is adopted and there are 5 antennas in total, the calculation complexity is 16 ⁵ =1048576. It is difficult for real-time systems to accept such a large computational complexity.

As an approximate simplification of the ML algorithm, the Sphere Decoding (SD) series of algorithms has attracted increasing attention because of its performance close to or equal to ML and its greatly reduced computational complexity compared to ML algorithms. Therefore, the sphere decoding method has become the preferred solution for MIMO signal detection at present.

In the sphere decoding method, there are three sub-algorithms: Depth First SD, DFSD, Metric First SD, MFSD, and Breadth First SD, BFSD. It is called the K-bestSD algorithm, and the industry-wide abbreviation K-best SD is used below. Among them, the breadth-first algorithm has great application prospects in real-time systems because of its fixed computational complexity and definite computational time. However, the computational complexity of the breadth-first algorithm is still very large, how to further reduce the complexity without affecting the performance is a hot issue in its application.

At present, there are some simplified methods for the K-best SD algorithm. Luis G.Barbero et al. proposed an FSD algorithm with a limited number of nodes expanded by each node (L.G.Barbero and J.S.Thompson, "A Fixed-Complexity MIMO Detector Based on the ComplexSphere Decoder, "2006IEEE 7th Workshop on Signal Processing Advances in Wireless Communications, Cannes, France: 2006, pp.1-5.), to a certain extent reduces the nodes visited during the decoding process, but only when the number of reserved paths is large Only when it is close to the performance of the K-best SD algorithm, and at this time, the computational complexity is relatively high.

Cong Xiong et al proposed an algorithm that reduces the number of reserved nodes in the FSD algorithm as the number of search layers increases (Cong Xiong, Xin Zhang, Kai Wu, and Dacheng Yang, "A simplified fixed-complexity sphere decoder for V-BLAST systems, "Communications Letters, IEEE, vo1.13, 2009, pp.582-584.), this algorithm does not have enough theoretical basis for FSD to reduce the number of reserved nodes, and the reduced number of visited nodes is relatively limited.

Contents of the invention

(1) Technical problems to be solved

The technical problem to be solved by the present invention is: how to provide a signal detection scheme for a MIMO system, so as to reduce the complexity of the signal detection algorithm and improve the detection efficiency of received data without substantially reducing the detection performance.

(2) Technical solution

In order to solve the above-mentioned technical problems, the present invention provides a signal detection method of a MIMO system, comprising the following steps:

S1. Construct a search tree by using the known channel matrix and possible transmission symbols;

S2. Search the Nth layer of the search tree according to steps S21 to S23, where N is the total number of layers of the search tree:

S21. Divide the currently reserved nodes into M groups, each node corresponds to a partial transmission vector composed of a part of elements of a transmission symbol, and the transmission symbol is expressed in the form of a transmission vector;

S22. According to the principle that the number of expanded child nodes of nodes in each group is equal, and the number of expanded child nodes of nodes in different groups decreases sequentially from M, expand the child nodes of the Nth layer node. The way to expand the child nodes is to Adding a new element behind the corresponding part of the sending vector until the entire sending vector is formed, and the child node corresponds to the new element;

S23. Sorting according to the partial Euclidean distances between the partial sending vectors and partial receiving vectors corresponding to all expanded child nodes, and retaining K child nodes with the smallest partial Euclidean distances for the search of the next layer, the partial receiving vectors It consists of a part of elements of a received symbol corresponding to the sent symbol, the received symbol is expressed in the form of a received vector, and N, K, and M are all positive integers;

S3. Search other layers of the search tree in the manner of steps S21-S23, and search for a path with the smallest Euclidean distance among paths composed of child nodes as a search result.

Preferably, step S1 specifically includes: sorting each row of the channel matrix, and then triangulating the channel matrix obtained after sorting by using the possible transmission symbols to form the search tree.

Preferably, the possible transmission symbols are selected from a set formed by all permutations and combinations of possible transmission signals of each antenna.

Preferably, the manner of triangulating the channel matrix is performing QR decomposition on the channel matrix.

Preferably, in step S1, the method of sorting each row of the channel matrix is: if the number of all nodes in the row is less than or equal to K, then sort in order of gain from small to large, otherwise in order of gain from large to small sorted in order.

Preferably, the initial value of M is the number of transmitted symbols, and the value of M decreases as the number of search layers increases.

Preferably, in step S4, the path is composed of subnodes: select a subnode from each layer and connect them in turn to form 2 ^N×m paths, and the selected subnodes of the jth layer are all from the previous layer Expanded by the selected child node, j=2, . . . , N, m is the modulation order of the transmitted symbol.

Preferably, the step of reserving the K child nodes with the smallest partial Euclidean distance for the search of the next layer in step S23 is replaced by: retaining the K child nodes with the smallest partial Euclidean distance, and deleting the Euclidean distance from the K child nodes The child nodes whose value is greater than the distance threshold, and the remaining child nodes are used for the search of the next layer.

Preferably, the number of K decreases as the number of search layers increases.

The present invention also provides a signal detection device for a MIMO system, including:

a signal receiving unit, configured to receive symbols;

The signal detection unit is configured to first process the symbols received by the signal receiving unit by using the known channel matrix and possible transmission symbols to form a search tree, and search the Nth layer of the search tree in the following manner:

Dividing the currently reserved nodes into M groups, each node corresponds to a partial transmission vector composed of a part of elements of a transmission symbol, and the transmission symbol number is expressed in the form of a transmission vector;

According to the principle that the number of expanded child nodes of nodes in each group is equal, and the number of expanded child nodes of nodes in different groups decreases sequentially from M, the child nodes of the Nth layer node are expanded. Add new elements behind the part of the sending vector until the entire sending vector is formed, the child nodes correspond to the new elements, and N is the total number of layers of the search tree;

Secondly, sort according to the partial Euclidean distance between the partial sending vector and the partial receiving vector corresponding to all expanded sub-nodes, keep the K sub-nodes with the smallest partial Euclidean distance for the search of the next layer, and the partial receiving vector is determined by Composed of a part of elements of a received symbol corresponding to the sent symbol, the received symbol is expressed in the form of a received vector, and N, K, and M are all positive integers;

Search the other layers of the search tree again, and find the path with the smallest Euclidean distance in the path composed of child nodes as the search result;

A signal output unit, configured to output the search result obtained by the signal detection unit.

(3) Beneficial effects

The above-mentioned technical solution has the following advantages: the number of nodes to be searched is reduced by limiting the number of expanded nodes in each layer of the search tree, and the complexity of signal detection is reduced without substantially reducing the detection performance, and the reception is improved. The detection efficiency of the data is improved, and its hardware implementation circuit is simpler, the circuit area is smaller, and the power consumption is lower. The method of the present invention can also be applied to multi-user signal detection and other signal detection.

Description of drawings

Fig. 1 is a flowchart of a signal detection method according to an embodiment of the present invention;

FIG. 2 is a schematic diagram of a transmitting and receiving structure of a MIMO system according to an embodiment of the present invention;

Fig. 3 is the schematic diagram of the search tree of maximum likelihood search;

4 is a schematic diagram of a search tree formed after processing received signals according to an embodiment of the present invention;

Fig. 5 is the first result figure of the simulation experiment according to an embodiment of the present invention;

Fig. 6 is a second result diagram of a simulation experiment according to an embodiment of the present invention.

Detailed ways

The specific implementation manners of the present invention will be further described in detail below in conjunction with the accompanying drawings and embodiments. The following examples are used to illustrate the present invention, but are not intended to limit the scope of the present invention.

FIG. 1 is a flowchart of a signal detection method for a MIMO system according to an embodiment of the present invention. As shown in FIG. 1, the method includes the following steps:

S1. Process the received signal and use the known channel matrix and possible transmission symbols to form a search tree;

Specifically: each row of the channel matrix is sorted, and then the sorted channel matrix is triangulated using the possible transmission symbols to form a search tree, and the sorting method is from small to large (if all possible node numbers in this layer not greater than K)/gain from large to small (if the number of all possible nodes in this layer is greater than K), sorting can speed up the search and further increase the speed of signal detection. The possible transmission symbols are selected from a set formed by all permutations and combinations of possible transmission signals of each antenna.

S2. Search the Nth layer of the search tree according to the following steps, where N is the total number of layers of the search tree;

S21. Divide the K currently reserved nodes into M groups, each node corresponds to a partial transmission vector composed of a part of elements of a transmission symbol, the transmission symbol is expressed in the form of a transmission vector, and the initial value of M can be set as a modulation constellation The number of points in the graph, that is, the number of all sent signals; the number of groups M in each layer is not equal, and the number of M decreases as the number of search layers increases, and the reduction can be linear (such as the number of M in each layer decreases fixed number), it can also be non-linear (for example, M in the first few layers is unchanged, and M in the next few layers is directly reduced to 1); the number of groups in each layer is different, and the number of child nodes of the group with the most child nodes in each layer The number can also be different, and a trade-off can be made according to the requirements of performance and complexity.

S22. According to the principle that the number of expanded child nodes of nodes in each group is equal, and the number of expanded child nodes of nodes in different groups decreases sequentially from M, expand the child nodes of the Nth layer node. The way to expand the child nodes is to Adding a new element behind the corresponding part of the sending vector until the entire sending vector is formed, and the child node corresponds to the new element;

S23. Sorting according to the partial Euclidean distances between the partial sending vectors and the partial receiving vectors (corresponding to the partial receiving vectors) corresponding to all expanded sub-nodes, and retaining K sub-nodes whose partial Euclidean distances are smaller than the threshold for the next step One layer of search, the part of the receiving vector is composed of a part of the elements of the receiving symbol corresponding to the sending symbol, the receiving symbol is expressed as a receiving vector, and N, K, and M are all positive integers; the number of K varies with Decrease as the number of search layers increases; the threshold can be obtained according to the noise distribution; the step of reserving the K subnodes with the smallest Euclidean distance in step S23 for the search of the next layer can be replaced by: retaining the K subnodes with the smallest Euclidean distance K child nodes, and delete the child nodes whose Euclidean distance value is greater than the distance threshold from the K child nodes, and the remaining child nodes are used for the search of the next layer.

S3. Search all the layers according to steps S21-S23, and find the path with the smallest Euclidean distance among the paths composed of child nodes as the search result.

The present invention will be described in more detail below.

1. Mathematical representation of received signal

For the MIMO system, under the condition of abundant scattering paths, it is assumed that the number of transmitting antennas is Nt, and the number of receiving antennas is Nr, satisfying Nt≥Nr. The system model is shown in Figure 3. For the convenience of expression, take Nt=Nr=N. s=(s ₁ s ₂ …s _N ) ^T is a complex sending vector, y=(y ₁ y ₂ …y _N ) ^T is a complex receiving vector, n=(n ₁ n ₂ …n _N ) ^T is a complex The noise vector of which satisfies the variance of the real part and the imaginary part is σ, and H is a complex channel matrix of order N×N, in which each element is a complex Gaussian random variable with independent and identical distribution, satisfying E(|h _{i, j} | ² ) =1. Then the received signal can be expressed as:

y=Hs+n (1)

That is, one transmitted symbol (vector s) in MIMO contains multiple transmitted signals (scalars).

If the maximum likelihood detection (ML) algorithm is used for signal detection, the maximum likelihood solution is to select a set of satisfy:

$\stackrel{<span\; class="notranslate">\; ^</span>}{\mathrm{<span\; class="notranslate">\; the\; s</span>}}<span\; class="notranslate">\; =</span>\underset{<span\; class="notranslate">\; \&aleph;</span>}{\mathrm{<span\; class="notranslate">\; arg</span>}\mathrm{<span\; class="notranslate">\; min</span>}}{<span\; class="notranslate">\; |</span><span\; class="notranslate">\; |</span>\mathrm{<span\; class="notranslate">\; the\; y</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; Hs</span>}<span\; class="notranslate">\; |</span><span\; class="notranslate">\; |</span>}^{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 2</span>}<span\; class="notranslate">\; )</span>$

in, Set for all possible send signals. When the modulation order of the signal is m, find 2 ^m×N possible permutations need to be exhausted. When the number of antennas is large and the modulation order is high, the complexity of the ML algorithm is so high that it cannot be applied, so in fact, a suboptimal algorithm is generally used, for example, the Sphere Decoding (SphereDecoding, which is based on the present invention) SD) breadth-first algorithm (BFSD algorithm), often referred to as the K-best SD algorithm in the industry.

2. Triangulation and tree search of matrix

Perform QR decomposition on the matrix H, H=QR, where Q is a unitary matrix of N×N order, Q ^H Q=E, E is an identity matrix, and R is an upper triangular matrix of N×N order, expressed as:

Then formula (1) can be further expressed as:

ρ=Rs+η (3)

Among them: ρ=Q ^H y, η=Q ^H n.

Alternatively, formula (3) can also be expressed as:

Further expand formula (3.a), we can get:

${\mathrm{<span\; class="notranslate">\; \&rho;</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; =</span>{\mathrm{<span\; class="notranslate">\; r</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; i</span>}}{\mathrm{<span\; class="notranslate">\; the\; s</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; +</span>\underset{\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\overset{\mathrm{<span\; class="notranslate">\; N</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; r</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; j</span>}}{\mathrm{<span\; class="notranslate">\; the\; s</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; +</span>{\mathrm{<span\; class="notranslate">\; \&eta;</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1,2</span>}<span\; class="notranslate">\; ,</span><span\; class="notranslate">\; .</span><span\; class="notranslate">\; .</span><span\; class="notranslate">\; .</span>\mathrm{<span\; class="notranslate">\; N</span>}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 4</span>}<span\; class="notranslate">\; )</span>$

Each calculated ρ _i can be regarded as a layer of data. It can be seen from formula (4) that the received data ρ _N of the Nth layer is only determined by the data _s i sent by this layer and is not interfered by other layers; while the received data ρ _N-1 of the Nth layer is not related to the Nth layer In addition to the data s _N-1 sent by layer 1, it is also affected by the data s _N sent by layer N. Generally speaking, the received data ρ _i of the i-th layer is not only related to the sent data s _i of the i-th layer, but also affected by the data sent from the i+1 layer to the N-th layer (s _i+1 …s _N ), so When the transmission data of the i+1th layer to the Nth layer are known, the transmission data of the i-th layer can be obtained.

Let the estimated value of the transmit vector be:

$\stackrel{<span\; class="notranslate">\; ^</span>}{\mathrm{<span\; class="notranslate">\; the\; s</span>}}<span\; class="notranslate">\; =</span>{\left(\begin{array}{cccc}{\stackrel{<span\; class="notranslate">\; ^</span>}{\mathrm{<span\; class="notranslate">\; the\; s</span>}}}_{\mathrm{<span\; class="notranslate">\; 1</span>}}& {\stackrel{<span\; class="notranslate">\; ^</span>}{\mathrm{<span\; class="notranslate">\; the\; s</span>}}}_{\mathrm{<span\; class="notranslate">\; 2</span>}}& <span\; class="notranslate">\; \&Center\; Dot;</span><span\; class="notranslate">\; \&Center\; Dot;</span><span\; class="notranslate">\; \&Center\; Dot;</span>& {\stackrel{<span\; class="notranslate">\; ^</span>}{\mathrm{<span\; class="notranslate">\; the\; s</span>}}}_{\mathrm{<span\; class="notranslate">\; N</span>}}\end{array}\right)}^{\mathrm{<span\; class="notranslate">\; T</span>}}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 5</span>}<span\; class="notranslate">\; )</span>$

in Send the estimate of the vector for the i-th layer, T for the transpose. Then up to the i-th layer, the Euclidean distance between the estimated value of the transmitted vector and the received value of the detected part (which can be understood as the distance between the transmitted signal and the received signal through the channel) is:

${\mathrm{<span\; class="notranslate">\; d</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; =</span>\underset{\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; i</span>}}{\overset{\mathrm{<span\; class="notranslate">\; N</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}}{<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; \&rho;</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}}<span\; class="notranslate">\; -</span>\underset{\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; j</span>}}{\overset{\mathrm{<span\; class="notranslate">\; N</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}}{\mathrm{<span\; class="notranslate">\; r</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; k</span>}}{\stackrel{<span\; class="notranslate">\; ^</span>}{\mathrm{<span\; class="notranslate">\; the\; s</span>}}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; )</span>}^{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 6</span>}<span\; class="notranslate">\; )</span>$

Because the received data ρ _N of the Nth layer is only related to the sent data s _N of the Nth layer, the signal detection can start from the Nth layer first and proceed to the first layer in sequence. If all the values of the modulation mode of the transmitted signal are considered, there are 2 ^m choices in the first Nth layer of detection, and each value choice is regarded as a node, and each node is a possible solution. In this way, there are 2 ^m nodes in the Nth layer; the second detected layer N-1 is equivalent to expanding each node in the Nth layer to calculate 2 ^m nodes on the basis of the Nth layer, so there are 2 ^m × 2 ^m nodes; after each layer, the number of nodes in the previous layer is multiplied by 2 ^m , and the last layer has (2 ^m ) ^N =2 ^N×m nodes. Connect these nodes according to the order they are visited to form 2 ^N×m paths, these paths start from the same node, and each node divides 2 ^m nodes in the next layer, similar to an inverted tree, such as Figure 4 shows. The search process of the SD series algorithm can be depicted in Figure 4, and each node is a possible solution.

Assuming that the estimated value of the sending vector is formula (5), the partial Euclidean distance when the tree search reaches the i-th layer is (6), and then the partial Euclidean distance of all nodes searched in this layer is calculated according to the order from small to large Arranged in order, only keep the smallest K nodes (or can be understood as a certain number of nodes) for new searches.

The calculation of the new layer is to expand the new value selection on the basis of the reserved nodes of the previous layer, and use these values to calculate the partial Euclidean distance between the partial sending vector and the partial receiving vector. These partial Euclidean distances are sorted from small to large or from large to small, and a fixed number of nodes with partial Euclidean distances smaller than the threshold are reserved, and the new layer of detection ends.

In an example of the present invention, FIG. 3 shows a schematic diagram of a four-level maximum likelihood search.

Figure 4 shows a schematic diagram of a tree search of the present invention, in which it is assumed that each layer retains up to 8 nodes, and the maximum number of possible child nodes for each node is 4. The 8 nodes are arranged from left to right in order from small to large , they are divided into 4 groups, each group has two nodes, the number of child nodes of the two nodes in the first group is 4, and the number of child nodes of the second, third, and fourth groups are 3, 2, and 1, respectively. In FIG. 4 , the circles corresponding to the solid straight lines represent expanded nodes, and the circles corresponding to the dashed straight lines represent possible but not expanded nodes. Circles with solid lines indicate nodes that are kept, and circles with dashed lines indicate nodes that are not kept.

Fig. 5 and Fig. 6 show the results of simulation experiments according to an embodiment of the present invention. The experimental environment is 8 transmitting antennas, 8 receiving antennas, and 16-/64-/256-QAM modulation mode (represented by m=4, m=8, and m=16 respectively). The performance comparison of three kinds of algorithms is arranged among Fig. 5 and Fig. 6, traditional K-best SD algorithm (represented by KSD among the figure), the algorithm of the present invention (represented by SRKSD among the figure), and the spherical decoding algorithm of fixed complexity (represented by KSD among the figure) It is represented by FSD in the figure). Figure 5 shows the performance curve without sorting, and Figure 6 shows the performance curve in the sorting case, and all algorithms are preceded by "O" to show sorting. The horizontal axis of the graph represents the signal-to-noise ratio SNR, and the vertical axis represents the bit error rate BER performance. Wherein the value of K of the traditional K-best SD algorithm is the same as the algorithm of the present invention, so the computational complexity is greater than the present algorithm, and the computational complexity of the fixed-complexity sphere decoding is the same as the present algorithm. It can be seen from Fig. 5 and Fig. 6 that the performance of the method of the present invention is basically the same as that of the K-best SD algorithm, and the performance difference cannot be distinguished from the figures. It can be seen from Fig. 5 that without sorting, the performance of the method of the present invention is much better than that of the FSD algorithm, and in the case of BER=10 ^-3 , the performance of the method of the present invention is better than that of the three modulation modes FSD algorithm is greater than 5dB. It can be seen from Fig. 6 that in the case of sorting, the performance of the method of the present invention is better than that of the FSD algorithm. When BER=10 ^-3 and the modulation mode is 16-QAM, the method of the present invention is about 0.5dB better than the FSD algorithm with the same complexity.

Thus it can be seen that the embodiment of the present invention achieves the same performance as the breadth-first sphere decoding algorithm in the case of the same width, and in the case of the same computational complexity, the performance is better than that of the fixed-complexity sphere decoding algorithm. algorithm.

The present invention also provides a signal detection device for a MIMO system, including:

a signal receiving unit, configured to receive symbols;

The signal detection unit is configured to first process the symbols received by the signal receiving unit by using the known channel matrix and possible transmission symbols to form a search tree, and search the Nth layer of the search tree in the following manner:

Dividing the currently reserved nodes into M groups, each node corresponds to a partial transmission vector composed of a part of elements of a transmission symbol, and the transmission symbol number is expressed in the form of a transmission vector;

According to the principle that the number of expanded child nodes of nodes in each group is equal, and the number of expanded child nodes of nodes in different groups decreases sequentially from M, the child nodes of the Nth layer node are expanded. Add new elements behind the part of the sending vector until the entire sending vector is formed, the child nodes correspond to the new elements, and N is the total number of layers of the search tree;

Secondly, sort according to the partial Euclidean distance between the partial sending vector and the partial receiving vector corresponding to all expanded sub-nodes, keep the K sub-nodes with the smallest partial Euclidean distance for the search of the next layer, and the partial receiving vector is determined by Composed of a part of elements of a received symbol corresponding to the sent symbol, the received symbol is expressed in the form of a received vector, and N, K, and M are all positive integers;

Search the other layers of the search tree again, and find the path with the smallest Euclidean distance in the path composed of child nodes as the search result;

A signal output unit, configured to output the search result obtained by the signal detection unit.

It can be seen from the above embodiments that the present invention reduces the number of nodes to be searched by limiting the number of nodes to be expanded when searching for each layer of the search tree, and reduces the complexity of signal detection without substantially reducing the detection performance. The detection efficiency of the received data is improved, and its hardware implementation circuit is simpler, the circuit area is smaller, and the power consumption is lower. The method of the present invention can also be applied to multi-user signal detection and other signal detection.

The above is only a preferred embodiment of the present invention, it should be pointed out that for those of ordinary skill in the art, without departing from the technical principle of the present invention, some improvements and replacements can also be made, these improvements and replacements It should also be regarded as the protection scope of the present invention.

Claims (10)

1. a signal detecting method for mimo system, is characterized in that, comprises the following steps:

S1, utilize known channel matrix and possible transmission symbol construction search tree;

S2, according to step S21～S23, the N layer of described search tree is searched for total number of plies that N is described search tree:

S21, current reservation node is divided into M group, the part that the corresponding a part of element that sends symbol of each node forms sends vector, and described transmission symbol table is shown the form that sends vector;

S22, according to the expansion son node number of every group of interior nodes equates, the expansion son node number of interior nodes does not start to successively decrease successively from M on the same group principle, launch the child node of N node layer, launch the mode of child node for add new element after the corresponding part transmission of described node vector, until form whole transmission vector, the corresponding described new element of described child node;

S23, the part of Euclidean distance receiving between vector according to part transmission vector corresponding to the child node of all expansion and part sort, K child node of reserve part Euclidean distance minimum is for the search of lower one deck, described part reception vector is comprised of a part of element of the receiving symbol corresponding with described transmission symbol, described receiving symbol is expressed as the form that receives vector, and N, K, M are positive integer;

S3, according to the mode of step S21～S23 search for described search tree other layer, in the path being formed by child node, find the path of Euclidean distance minimum as Search Results.

2. the method for claim 1, is characterized in that, step S1 is specially: by each line ordering of advancing of described channel matrix, then utilize described possible transmission symbol trigonometric ratio to form described search tree the channel matrix obtaining after sequence.

3. the method for claim 1, is characterized in that, the described possible transmission symbol choosing set that freely all permutation and combination of each antenna possibility transmitted signal form.

4. method as claimed in claim 2, is characterized in that, described by the mode of channel matrix trigonometric ratio for channel matrix is carried out to QR decomposition.

5. method as claimed in claim 2, it is characterized in that, in step S1, by the mode of each line ordering of advancing of described channel matrix, be: if all nodes of this row are less than or equal to K, according to gain order from small to large, sort, otherwise sort according to gain order from big to small.

6. the method for claim 1, is characterized in that, the initial value of M is for sending the number of symbol, and the value of M reduces along with increasing of the number of plies of search.

7. the method for claim 1, is characterized in that, the mode that forms path by child node in step S3 is: from every one deck, select a child node to be connected in turn and form 2 ^{n * m}paths, and the child node of selected j layer is all to launch from the selected child node of last layer, j=2 ..., N, m is for sending the order of modulation of symbol.

8. the method for claim 1, it is characterized in that, step by K child node of reserve part Euclidean distance minimum in step S23 for the search of lower one deck replaces with: K child node of reserve part Euclidean distance minimum, and delete the child node that Euclidean distance value is greater than distance threshold from a described K child node, remaining child node is for the search of lower one deck.

9. method as claimed in claim 7, is characterized in that, the number of K reduces along with increasing of the number of plies of search.

10. a signal supervisory instrument for mimo system, is characterized in that, comprising:

Signal receiving unit, for receiving symbol;

Detecting signal unit, for first utilizing known channel matrix and possible transmission symbol to process symbol that described signal receiving unit receives to form search tree, and in such a way the N layer of described search tree is searched for:

Current reservation node is divided into M group, and the corresponding part that sends a part of element composition of symbol of each node sends vector, and described transmission symbol table is shown the form that sends vector;

According to the principle that the expansion son node number of every group of interior nodes equates, the expansion son node number of interior nodes does not start to successively decrease successively from M on the same group, launch the child node of N node layer, launch the mode of child node for add new element after the corresponding part transmission of described node vector, until form whole transmission vector, the corresponding described new element of described child node, total number of plies that N is described search tree;

Secondly the part of Euclidean distance that sends vector according to part corresponding to the child node of all expansion and partly receive between vector sorts, K child node of reserve part Euclidean distance minimum is for the search of lower one deck, described part reception vector is comprised of a part of element of the receiving symbol corresponding with described transmission symbol, described receiving symbol is expressed as the form that receives vector, and N, K, M are positive integer;

Again search for other layer of described search tree, in the path being formed by child node, find the path of Euclidean distance minimum as Search Results;

Signal output unit, for the Search Results output that described detecting signal unit is obtained.