CN101582750B - Sphere decoding detection method based on breadth-first search - Google Patents
Sphere decoding detection method based on breadth-first search Download PDFInfo
- Publication number
- CN101582750B CN101582750B CN200910086118XA CN200910086118A CN101582750B CN 101582750 B CN101582750 B CN 101582750B CN 200910086118X A CN200910086118X A CN 200910086118XA CN 200910086118 A CN200910086118 A CN 200910086118A CN 101582750 B CN101582750 B CN 101582750B
- Authority
- CN
- China
- Prior art keywords
- node
- nodes
- matrix
- weight
- layer
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Expired - Fee Related
Links
- 238000001514 detection method Methods 0.000 title claims abstract description 31
- 239000011159 matrix material Substances 0.000 claims abstract description 46
- 238000000354 decomposition reaction Methods 0.000 claims abstract description 16
- 238000012163 sequencing technique Methods 0.000 claims description 9
- 238000000034 method Methods 0.000 abstract description 9
- 239000010410 layer Substances 0.000 description 38
- 238000007476 Maximum Likelihood Methods 0.000 description 9
- 238000010586 diagram Methods 0.000 description 4
- 238000004891 communication Methods 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 3
- 239000011229 interlayer Substances 0.000 description 2
- 238000013507 mapping Methods 0.000 description 2
- URWAJWIAIPFPJE-YFMIWBNJSA-N sisomycin Chemical compound O1C[C@@](O)(C)[C@H](NC)[C@@H](O)[C@H]1O[C@@H]1[C@@H](O)[C@H](O[C@@H]2[C@@H](CC=C(CN)O2)N)[C@@H](N)C[C@H]1N URWAJWIAIPFPJE-YFMIWBNJSA-N 0.000 description 2
- 206010033799 Paralysis Diseases 0.000 description 1
- 230000003044 adaptive effect Effects 0.000 description 1
- 230000005587 bubbling Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 238000005562 fading Methods 0.000 description 1
- 230000007774 longterm Effects 0.000 description 1
Images
Landscapes
- Radio Transmission System (AREA)
Abstract
The invention provides a sphere decoding detection method based on breadth-first search, comprising the following steps: step A, carrying out QR decomposition on channel matrix H to obtain Q matrix and R matrix; step B, multiplying conjugate transpose of Q matrix with received signal y to obtain equalizing signal rho of the received signal; step C, setting search nodes of Ki, i=1, 2, ..., N<T>, and N<T> is the number of sending antennae; and step D, carrying out breadth-first search on R matrix and rho, wherein preserving Ki nodes with the lowest weight when carrying out the i layer search. The method of the invention can effectively reduce operating complexity of the sphere decoding and is easy to realize by hardware.
Description
Technical Field
The invention belongs to the field of wireless communication, in particular to a sphere decoding detection method based on breadth-first search for a multiple-input multiple-output (MIMO) system.
Background
In the current wireless communication standard and its evolution, mimo antenna technology has been widely adopted. Both in 3GPP Long Term Evolution (LTE) and 802.16 series of evolution versions, orthogonal frequency division multiplexing and MIMO techniques are widely used as key technologies. Compared with the traditional single input and output (SISO) system, the MIMO system carries out MIMO signal detection under the condition that the time and the frequency are overlapped with each other, so the MIMO signal detection complexity is greatly higher than that of the traditional SISO signal detection.
In theory, MIMO signals can be detected by Maximum Likelihood (ML) detection methods. However, the number of constellation points for which maximum likelihood detection needs to be traversed and searched increases exponentially with the number of transmit antennas and the degree of freedom of a modulation mode, and the operation complexity is hard to bear in an actual system under the conditions of a large number of transmit antennas and high-order modulation. Therefore, finding a detection method with performance close to ML and greatly reduced complexity becomes a key factor for whether the MIMO detection technology can be implemented in an actual system.
Thus, viterbi et al proposed a Sphere Decoding (SD) algorithm for signals with a grid-like constellation based on the study of Pohst et al. While Damen generalizes this algorithm to MIMO signal detection and achieves better performance than vertical bell labs layered space-time (V-BLAST) detection.
However, the conventional sphere decoding has a disadvantage that the algorithm has different computational complexity due to different channel conditions, signal quality and initial radius. Especially for singular matrixes, the whole algorithm may not be converged, and the system is paralyzed. Therefore, how to effectively control the complexity of the sphere decoding algorithm, the stability and the robustness of the system is particularly important for a real-time wireless communication system. For the above reasons, a K-Best algorithm using a breadth-first algorithm instead of a depth-first algorithm has emerged, and a core idea thereof is to keep only K nodes with the smallest weight value when searching for an optimal path at each layer, and then continue to search from the reserved K nodes to the lower layer until the bottommost layer.
However, for the K-Best algorithm, there is also how to derive from KxMc(McModulation point number) nodes, and how to reduce the complexity of the algorithm as much as possible under the condition that the performance is basically consistent with the Maximum Likelihood (ML) (maximum likelihood) performance.
Disclosure of Invention
The invention aims to solve the technical problem of providing a sphere decoding detection method based on breadth-first search so as to reduce the operation complexity of sphere decoding and be easy to realize through hardware.
In order to solve the technical problems, the invention provides the following technical scheme:
a sphere decoding detection method based on breadth-first search comprises the following steps:
A. carrying out QR decomposition on the channel matrix H to obtain a Q matrix and an R matrix;
B. multiplying the conjugate transpose of the Q matrix by the received signal y to obtain an equalized signal rho of the received signal;
C. setting the number K of search nodes of the ith layeri，i＝1，2，...，NT，NTThe number of transmitting antennas;
D. performing breadth-first search according to the R matrix and the rho, wherein K with the minimum weight is reserved when the ith layer search is performediAnd (4) each node.
In the above sphere decoding detection method, in step a, the QR decomposition is a sorted QR decomposition, so that a modulus of an ith element on a diagonal of the R matrix is not greater than a modulus of an (i + 1) th element.
In the spherical decoding detection method, in step C, the number K of search nodes of the i-th layer is set according to the modulation mode, the target bit error rate and the channel state informationi。
In the sphere decoding detection method, in step C, the set i-th layerNumber of search nodes KiNumber of search nodes K not less than i +1 th layeri+1。
In the foregoing sphere decoding detection method, step D specifically includes:
d1, performing table lookup and sorting on the nodes of the layer 1 according to the spherical decoding expression determined by the R matrix and the rho, and reserving K with the minimum Euclidean distance1A node and calculates the reserved K1The weight of each node;
d2, K reserved for the i-1 th layer when the i-th layer search is executedi-1Path expansion is carried out on each node, sub-nodes obtained by path expansion are subjected to division, combination and sequencing, and K with the smallest weight is reservediA node;
d3, after the last layer is searched, outputting a decoding result.
In the spherical decoding detection method, in step D2, the dividing, combining and sorting includes:
d21 for Ki-1The subnodes of each node perform table lookup and sorting according to the spherical decoding expression determined by the R matrix and the rho, and calculate the Ki-1The weight of the child node of each node;
d22, comparing the weight of the child node of the 1 st node of the i-1 st layer with the weight of the child node of the 2 nd node according to the table lookup sorting result of the step D21, and selecting the K with the minimum weightiThe child node then selects K with the smallest weightiThe weight of the child node is compared with the weight of the child node of the 3 rd node, and K with the smallest weight is selectediThe child nodes, and so on, select the K with the smallest weightiThe child nodes act as reserved nodes of the ith layer.
In the sphere decoding detection method, in step D21, for the Ki-1And calculating the weight values of different numbers of child nodes by each node in the nodes.
The invention improves the traditional K-best algorithm, comprising QR decomposition of the channel matrix and equalization processing of the received signal, and adaptively sets the number of nodes searched in each layer, thereby reducing the operation complexity of the spherical decoding. The invention also further carries out QR decomposition of sequencing of the channel matrix, and carries out in-node sequencing by adopting a table look-up mode, and carries out in-layer sequencing based on a divide-and-conquer merging algorithm on the basis of the in-node sequencing, thereby further reducing the operation complexity of spherical decoding and being easy to realize through VLSI hardware.
Drawings
FIG. 1 is a diagram of a basic MIMO system model;
fig. 2 is a flowchart of a sphere decoding detection method based on breadth-first search according to an embodiment of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention will be described in detail with reference to the accompanying drawings and specific embodiments.
For the basic MIMO system model, as shown in fig. 1, assume that the number of transmit antennas is NtThe number of receiving antennas is NRAnd the channel is a flat fading channel, the system can be represented by the following formula:
y＝H·s+n (1)
wherein y is NRX 1 received signal vector, s is NtX 1 of the transmitted signal vector, N being NRA noise vector of x 1 with mean 0 and variance N0H is NR×NtDimensional channel model vector (channel matrix).
Adopting a maximum likelihood estimation algorithm, the expression is as follows:
where Ω is the effective modulation signal point. In order to reduce the complexity of the maximum likelihood algorithm, the core idea of the traditional K-Best algorithm is based on the breadth-first idea, namely, a certain number of nodes are selected in each layer, and then the processes of path expansion and node search are carried out.
The sphere decoding detection method based on the breadth-first improves the traditional K-Best algorithm. Firstly, QR decomposition is performed on H to obtain a unitary matrix Q and an upper triangular matrix R, and the following expression is obtained according to formula (2):
since R is an upper triangular matrix, the following expression (4) can be obtained by using an iterative method:
wherein, <math>
<mrow>
<mi>ρ</mi>
<mo>=</mo>
<msup>
<mi>Q</mi>
<mi>H</mi>
</msup>
<mi>y</mi>
<mo>=</mo>
<msup>
<mrow>
<mo>(</mo>
<msub>
<mi>ρ</mi>
<mn>1</mn>
</msub>
<mo>,</mo>
<msub>
<mi>ρ</mi>
<mn>2</mn>
</msub>
<mo>,</mo>
<msub>
<mi>ρ</mi>
<mn>3</mn>
</msub>
<mo>,</mo>
<mo>.</mo>
<mo>.</mo>
<mo>.</mo>
<mo>,</mo>
<msub>
<mi>ρ</mi>
<msub>
<mi>N</mi>
<mi>R</mi>
</msub>
</msub>
<mo>)</mo>
</mrow>
<mi>T</mi>
</msup>
<mo>,</mo>
</mrow>
</math> ri，jthe elements representing the ith row and jth column of the matrix R, (-)HRepresents the conjugate transpose of (-) and,
in order to demodulate the vector(s),is composed of
The jth element of the vector.
In the present invention, for the convenience of description, e in the above formula isiCalled the euclidean distance of the node and Ei the weight of the node.
Referring to fig. 2, the sphere decoding detection method based on breadth-first search according to the embodiment of the present invention mainly includes the following steps:
step 201: carrying out QR decomposition on the channel matrix H to obtain a Q matrix and an R matrix;
preferably, the QR decomposition is a sorted QR decomposition, and results in a Q matrix, an R matrix, and a P matrix (sorted switching matrix). Through sorted QR decomposition, the modulus value of the ith element on the diagonal of the R matrix is not larger than the modulus value of the (i + 1) th element, namely, the diagonal elements of the R matrix are arranged from the upper left corner to the lower right corner according to the order of the modulus values from small to large. The QR decomposition of the ordering is equivalent to the interlayer ordering of the search tree. By the interlayer sequencing, the Euclidean distance of the high-level nodes is relatively large, and the probability that the correct constellation point is not searched can be reduced by combining the self-adaptive setting of the number of nodes searched in each layer in the subsequent step, so that the searching speed is increased, and the operation complexity of the spherical decoding is reduced.
Step 202: the conjugate transpose of the Q matrix is multiplied by the received signal y to obtain the equalized signal ρ of the received signal, i.e., ρ is QHy；
In the invention, rho is used as an equivalent received signal, an R matrix is used as an equivalent channel matrix, and a search tree is constructed according to the R matrix and the rho to carry out spherical decoding.
Step 203: setting the number K of search nodes of the ith layeri，i＝1，2，...，NT，NTThe number of transmitting antennas;
in the traditional K-Best algorithm, the number of nodes searched in each layer is fixed; the invention improves the method and flexibly sets the K value of each layer according to different system parameters and channel parameters.
Specifically, the number K of search nodes of the i-th layer may be set according to a modulation scheme, a target bit error rate, and Channel State Information (CSI)i(this makes the number of nodes searched per layer adaptive). Preferably, the number K of search nodes of the ith layer can be setiNumber of search nodes K not less than i +1 th layeri+1I.e. the number of nodes searched at a higher levelRelatively much. The number of nodes searched by the high level is relatively large, so that the probability that the correct constellation point is not searched can be reduced.
Step 204: performing breadth-first search according to the R matrix and the rho, wherein K with the minimum weight is reserved when the ith layer search is performediAnd (4) each node.
Step 204 specifically includes:
d1, performing table lookup and sorting on the nodes of the layer 1 according to the spherical decoding expression determined by the R matrix and the rho, and reserving K with the minimum Euclidean distance1A node and calculates the reserved K1The weight of each node;
d2, K reserved for the i-1 th layer when the i-th layer search is executedi-1Path expansion is carried out on each node, sub-nodes obtained by path expansion are subjected to division, combination and sequencing, and K with the smallest weight is reservediA node;
d3, when the last layer is searched, outputting the decoding result. In this step, the constellation point mapping values of the path corresponding to the lowest-layer node with the minimum final weight are combined into a spherical decoding result. It should be noted that, when the sorted QR decomposition is performed in step 201, the constellation point mapping values are further sorted according to the P matrix, and the sequence obtained by the sorting is used as a final decoding result.
The dividing and treating and merging and sorting comprises the following steps:
d21 for Ki-1The subnodes of each node perform table lookup and sorting according to the spherical decoding expression determined by the R matrix and the rho, and calculate the Ki-1The weight of the child node of each node;
in this step, preferably, for said Ki-1Each node in the nodes can calculate the weights of different numbers of child nodes according to the actual situation (i.e. the weights of only part of the child nodes of the node can be calculated). In the following step D22, only the child nodes for which the weights are calculated are compared.
D22, comparing the weight of the child node of the 1 st node of the i-1 st layer with the weight of the child node of the 2 nd node according to the table lookup sorting result of the step D21, and selecting the K with the minimum weightiThe child node then selects K with the smallest weightiThe weight of the child node is compared with the weight of the child node of the 3 rd node, and K with the smallest weight is selectediThe child nodes, and so on, select the K with the smallest weightiThe child nodes act as reserved nodes of the ith layer.
For the sequencing of the child nodes of a node, the traditional method is to calculate the Euclidean distance of each child node, and then to sequence according to the Euclidean distance, for example, to sequence by using methods such as bubbling and inserting, and the algorithm complexity is high. In the invention, the Euclidean distance of each child node is not required to be calculated firstly, but table lookup and sorting are directly carried out according to the spherical decoding expression (3) determined by the R matrix and the rho, thereby achieving the purposes of reducing the algorithm complexity and improving the system performance.
The table lookup ordering means: carrying out iterative solution according to the spherical decoding expression (3) to obtain a demodulation vector
Determining
The component of the vector in the current layer and the position of the component in the source signal constellation diagram of the current layer; the nodes are ordered according to the relative size of the distance (Euclidean distance) between each constellation point (node) of the current layer source signal constellation diagram and the position of the component. Since the constellation points in the constellation diagram are regularly distributed, after the position is determined, the relative size of the Euclidean distances between all the constellation points and the position can be directly determined according to the distribution rule without calculating the Euclidean distances.
After the intra-node sorting is carried out in a table look-up mode, the intra-layer sorting can be carried out based on a divide-and-conquer combination algorithm on the basis of the intra-node sorting, so that the complexity of the sorting algorithm can be reduced on one hand, and the realization of VLSI hardware is facilitated through the table look-up method on the other hand.
The complexity of the conventional K-Best algorithm and the algorithm of the present invention is compared as follows (assuming that K of the current layer is 8, the modulation mode is 64QAM, and the conventional algorithm adopts bubble sorting):
| complexity of calculation | Traditional K-Best algorithm | Algorithm of the invention |
| In-node ordering | 63+62+...+57+56＝504 | 0 |
| In-layer ordering | 511+510+...+504＝4060 | 20×7＝140 |
As can be seen from the above table, the method of the present invention can greatly reduce the computational complexity.
Finally, it should be noted that the above embodiments are only for illustrating the technical solutions of the present invention and not for limiting, and those skilled in the art should understand that the technical solutions of the present invention can be modified or substituted with equivalents without departing from the spirit scope of the technical solutions of the present invention, which should be covered by the scope of the claims of the present invention.
Claims (7)
1. A sphere decoding detection method based on breadth-first search is characterized by comprising the following steps:
A. carrying out QR decomposition on the channel matrix H to obtain a Q matrix and an R matrix;
B. multiplying the conjugate transpose of the Q matrix by the received signal y to obtain an equalized signal rho of the received signal;
C. setting the number K of search nodes of the ith layeri，i＝1，2，...，NT，NTThe number of transmitting antennas;
D. performing a width from the R matrix and rhoDegree-first search, wherein K with the smallest weight is reserved when the ith layer search is executediAnd (4) each node.
2. The sphere decoding detection method of claim 1, wherein:
in step a, the QR decomposition is a sorted QR decomposition such that a modulus value of an ith element on a diagonal of the R matrix is not greater than a modulus value of an i +1 th element.
3. The sphere decoding detection method of claim 1, wherein:
in step C, the number K of searching nodes of the i-th layer is set according to the modulation mode, the target bit error rate and the channel state informationi。
4. The sphere decoding detection method of claim 1, wherein:
in step C, the set number K of search nodes of the ith layeriNumber of search nodes K not less than i +1 th layeri+1。
5. The sphere decoding detection method of claim 1, wherein said step D specifically comprises:
d1, performing table lookup and sorting on the nodes of the layer 1 according to the spherical decoding expression determined by the R matrix and the rho, and reserving K with the minimum Euclidean distance1A node and calculates the reserved K1The weight of each node;
d2, K reserved for the i-1 th layer when the i-th layer search is executedi-1Path expansion is carried out on each node, sub-nodes obtained by path expansion are subjected to division, combination and sequencing, and K with the smallest weight is reservediA node;
d3, after the last layer is searched, outputting a decoding result.
6. The sphere decoding detection method of claim 5, wherein in step D2, said dividing, merging and sorting includes:
d21 for Ki-1The subnodes of each node perform table lookup and sorting according to the spherical decoding expression determined by the R matrix and the rho, and calculate the Ki-1The weight of the child node of each node;
d22, comparing the weight of the child node of the 1 st node of the i-1 st layer with the weight of the child node of the 2 nd node according to the table lookup sorting result of the step D21, and selecting the K with the minimum weightiThe child node then selects K with the smallest weightiThe weight of the child node is compared with the weight of the child node of the 3 rd node, and K with the smallest weight is selectediThe child nodes, and so on, select the K with the smallest weightiThe child nodes act as reserved nodes of the ith layer.
7. The sphere decoding detection method of claim 6, wherein:
in step D21, for the Ki-1And each node in the nodes calculates the weight of partial child nodes of the node.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN200910086118XA CN101582750B (en) | 2009-06-02 | 2009-06-02 | Sphere decoding detection method based on breadth-first search |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN200910086118XA CN101582750B (en) | 2009-06-02 | 2009-06-02 | Sphere decoding detection method based on breadth-first search |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN101582750A CN101582750A (en) | 2009-11-18 |
| CN101582750B true CN101582750B (en) | 2012-11-21 |
Family
ID=41364736
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN200910086118XA Expired - Fee Related CN101582750B (en) | 2009-06-02 | 2009-06-02 | Sphere decoding detection method based on breadth-first search |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN101582750B (en) |
Families Citing this family (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101997657B (en) * | 2010-11-03 | 2013-01-09 | 北京邮电大学 | Detection method for breadth-first sphere decoding in MIMO (multiple input multiple output) system |
| CN102006148B (en) * | 2010-12-07 | 2013-04-17 | 西安电子科技大学 | Multiple-input multiple-output (MIMO) signal detection method based on breadth-first tree search |
| CN103856254B (en) * | 2012-11-29 | 2017-09-12 | 中兴通讯股份有限公司 | A kind of fixed complexity globular decoding detection method of soft output and device |
| CN103973602B (en) * | 2013-01-28 | 2017-07-21 | 电信科学技术研究院 | Signal detecting method and device |
Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101064579A (en) * | 2006-04-25 | 2007-10-31 | 上海无线通信研究中心 | Method for detecting low-complexity globular decoding |
-
2009
- 2009-06-02 CN CN200910086118XA patent/CN101582750B/en not_active Expired - Fee Related
Patent Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101064579A (en) * | 2006-04-25 | 2007-10-31 | 上海无线通信研究中心 | Method for detecting low-complexity globular decoding |
Also Published As
| Publication number | Publication date |
|---|---|
| CN101582750A (en) | 2009-11-18 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| TWI424699B (en) | Multilevel cluster-based mimo detection method and mimo detector thereof | |
| CN101582748B (en) | Method and device for detecting low-complexity signal of MIMO system | |
| CN101268647B (en) | Apparatus and method for detecting communications from multiple sources | |
| US11711155B2 (en) | Self-adaptive MIMO detection method and system | |
| JP4460412B2 (en) | Reception device and partial bit determination method | |
| US7817740B2 (en) | Method and system for minimum mean squared error soft interference cancellation (MMSE-SIC) based suboptimal maximum likelihood (ML) detection for multiple input multiple output (MIMO) wireless system | |
| CN101499840B (en) | Iteration detection method for MIMO system | |
| US9008240B1 (en) | Near maximum likelihood spatial multiplexing receiver | |
| JP2009060616A (en) | Method and device for detecting signal with low complexity used for mimo system | |
| CN100589469C (en) | Iterative decoding algorithm for space hour bit interlaced modulating system and receiving system | |
| CN101345592B (en) | Self-adapting signal detector and detection method used for MIMO | |
| CN101562464B (en) | Method for detecting spherical decode based on depth-first search | |
| JP4381901B2 (en) | Channel estimation and data detection method | |
| CN101453444A (en) | Constellation map simplifying apparatus, method and receiver | |
| CN112215335B (en) | System detection method based on deep learning | |
| CN101582750B (en) | Sphere decoding detection method based on breadth-first search | |
| US10374772B2 (en) | Method for slicing K-best detection in multiple-input multiple-output wireless communications system | |
| CN101557269B (en) | Sphere decoding detection method based on ultra large scale integrated circuit | |
| CN101227254A (en) | Method for detecting V-BLAST in MIMO system | |
| CN101964667B (en) | High-efficiency multi-antenna detection method for long term evolution scheme | |
| CN109286587A (en) | A kind of how active generalized space method for modulation detection | |
| US9066247B2 (en) | Communication devices and methods for signal detection | |
| CN104038457A (en) | Soft output sphere decoding method in coding MIMO system based on initial sphere radius | |
| KR101348557B1 (en) | Method for detecting signal using mimo-ofdm system and apparatus thereof | |
| CN107359911B (en) | Maximum likelihood detector |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20121121 Termination date: 20180602 |
|
| CF01 | Termination of patent right due to non-payment of annual fee |