CN103873120B - A kind of multiaerial system parallel detecting method based on breadth-first search - Google Patents
A kind of multiaerial system parallel detecting method based on breadth-first search Download PDFInfo
- Publication number
- CN103873120B CN103873120B CN201410139513.0A CN201410139513A CN103873120B CN 103873120 B CN103873120 B CN 103873120B CN 201410139513 A CN201410139513 A CN 201410139513A CN 103873120 B CN103873120 B CN 103873120B
- Authority
- CN
- China
- Prior art keywords
- input
- output
- unit
- cost value
- module
- 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
- 238000000034 method Methods 0.000 title claims abstract description 33
- 239000011159 matrix material Substances 0.000 claims abstract description 38
- 238000001514 detection method Methods 0.000 claims abstract description 23
- 238000000354 decomposition reaction Methods 0.000 claims abstract description 16
- 238000012163 sequencing technique Methods 0.000 claims description 10
- 238000006243 chemical reaction Methods 0.000 abstract description 3
- 230000007423 decrease Effects 0.000 abstract description 2
- 238000004891 communication Methods 0.000 description 5
- 238000010586 diagram Methods 0.000 description 3
- 230000004083 survival effect Effects 0.000 description 3
- 238000004364 calculation method Methods 0.000 description 2
- FFBHFFJDDLITSX-UHFFFAOYSA-N benzyl N-[2-hydroxy-4-(3-oxomorpholin-4-yl)phenyl]carbamate Chemical compound OC1=C(NC(=O)OCC2=CC=CC=C2)C=CC(=C1)N1CCOCC1=O FFBHFFJDDLITSX-UHFFFAOYSA-N 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000010295 mobile communication Methods 0.000 description 1
- 238000007781 pre-processing Methods 0.000 description 1
- 230000000717 retained effect Effects 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
- 230000017105 transposition Effects 0.000 description 1
Landscapes
- Radio Transmission System (AREA)
Abstract
The invention discloses a kind of multiaerial system parallel detecting method based on breadth-first search, feature is processing system by inputting access pretreatment module, the outfan of pretreatment module is connected with the input of tree search module, the outfan of tree search module is connected with the input of output module, and the outfan of output module exports for system；Introduce a kind of new plural detection model to the conversion of real number detection model, using the special nature through just handing in upper triangular matrix after triangle decomposition for the real channel matrix under new real number detection model, employ the method Accelerating running that two-layer calculates simultaneously, to improve the degree of parallelism of algorithm, the method simultaneously adopting subtree multiplexing reduces operand, thus reducing instruction cycles, accelerating the algorithm speed of service, improving the throughput of decoding.Optimum with existing list（Kbest）Detection method is compared, can be so that execution cycle number needed for multiaerial system detection process averagely reduces 35% while ensureing that performance does not decline using the inventive method.
Description
Technical field
The invention belongs to the parallel detecting method technical field in digital communication is and in particular to be based on width in multiaerial system
The parallel detecting method of degree first search.
Background technology
With multiaerial system（MIMO）For principal character forth generation mobile communication system improve system data rate
While, the growth with number of antennas and the increase of order of modulation, the complexity of its detection algorithm is also exponentially increased.Width
Preferential Sphere Decoding Algorithm（Breadth-First Sphere Decoding）Use for reference the search in classical Sphere Decoding Algorithm
Tree-model, simultaneously by a kind of unidirectional in the way of successively search for forward, and in this process according to certain criterion remove partial branch,
Because its structure is simple, it is suitable to hardware and realizes, thus be used widely.
《International Electrical IEEE-communications field offprint》（IEEE Journal on Selected Areas
In Communications, VOL.24, No.3, pp.491-503,2006）Describe a kind of breadth-first ball being easily achieved
Shape decoding algorithm list is optimum（Kbest）Algorithm, this algorithm selects the part Euclidean distance minimum of current calculating every time
Front K paths, remove the branch road under other child nodes simultaneously, and this method characteristic is that data flowing is regular, there is not backtracking behaviour
Make, be suitable for hardware concurrentization and realize.
But it is optimum in list（Kbest）In the search tree of algorithm, due to relying in front and back between layers, that is, after
The calculating of layer depends on the result of front layer, otherwise cannot calculate the cost value of this layer, therefore algorithm is between the layers
Strict sequential.In order to improve the execution efficiency of algorithm further, need to manage to eliminate dependency between layers, improve meter
The degree of parallelism calculated, reduces calculating cycle, improves data throughput rate.
Content of the invention
The present invention proposes a kind of multiaerial system based on breadth-first search（MIMO）Parallel detecting method, to improve calculation
The degree of parallelism of method, thus reducing instruction cycles, accelerating the algorithm speed of service, improving the throughput of decoding.
The multiaerial system parallel detecting method based on breadth-first search for the present invention is it is characterised in that processing system is by pre-
Processing module A1, tree search module A2 and output module A3 press following configuration and connect composition：The input of system accesses pretreatment
Modules A 1, the outfan of pretreatment module A1 is connected with the input of tree search module A2, the outfan of tree search module A2
It is connected with the input of output module A3, the outfan of output module A3 exports for system；
The structure of wherein pretreatment module A1 is：One complex received signal vector y of systemcInput is linked into multiple real-turn
Change the input of unit B 1, another complex channel matrix H of systemcInput is linked into the input that unit B 2 is changed in multiple real-turn；
The real channel matrix H output that unit B 2 is changed in multiple real-turn is linked into the input just handing in triangle decomposition unit B 4；Multiple real-turn is changed
The real number receipt signal vector of unit B 1Output exports as square with an orthogonal matrix Q just handing in triangle decomposition unit B 4
The input of battle array multiplying unit B3；The real number vector y of matrix multiple unit B 3 exports and is just handing in the another of triangle decomposition unit B 4
Individual upper triangular matrix R is output as two outputs of pretreatment module A1；
The structure of described tree search module A2 is：One real number vector y output of pretreatment module A1 is linked into even level
The input of cost value computing unit C1 and the input of odd-level cost value computing unit C2, pretreatment module A2 another
Upper triangular matrix R output is linked into the input of even level cost value computing unit C1 and odd-level cost value computing unit C2's
Input；The outfan of the outfan of even level cost value computing unit C1 and odd-level cost value computing unit C2 is linked into be asked
Input with unit C3；The outfan of sum unit C3 is linked into the input of sequencing unit C4；Two of sequencing unit C4
Outfan is linked into the input of even level cost value computing unit C1 and the input of odd-level cost value computing unit C2, separately
One outfan is as the output setting search module A2；
The structure of described output module A3 is：The output of tree search module A2 is as the real input answering converting unit D1, real
The output of multiple converting unit D1 is as the output of output module A4；
Parallel detection operation is carried out as follows：First multiaerial system plural number detection model is converted into real number detection
Model, that is,：The real part of each element of complex received signal vector plural number sending signal vector and imaginary part are replaced and arranges, that is,
Order arrangement according to " real part-imaginary part-real part-imaginary part "；Real channel matrix after complex channel matrixing is carried out
Orthogonal matrix and upper triangular matrix is respectively obtained, the calculating using odd-level cost value does not rely on after just handing in triangle decomposition
This property of the result of its preceding layer, is calculated to adjacent odd-level and even level simultaneously：The search tree mould that parity price goes out
Every two-layer of type calculates its cost value respectively, is then added the cost value of corresponding two nodes on same paths and sues for peace
Cost value to current level of child nodes；And when calculating even level cost value, its fortune is reduced using the method for subtree multiplexing
Calculation amount, that is, utilize the even level of the search tree property separate when calculating with odd-level, is calculating odd-level in idol
During the cost value of the individual child node of several layers of log2 expanding (M) * log2 (M), before only calculating, log2 (M) is individual, then replicates log2
(M) secondary, M therein is order of modulation.
Because the present invention is based on introducing a kind of new answering in the multiaerial system parallel detecting method of breadth-first search
Number detection model, to the conversion of real number detection model, is just handing in triangle decomposition using under new model（QR decomposes）Go up three angular moments afterwards
The special nature of battle array, employs the method Accelerating running that two-layer calculates simultaneously, to improve the degree of parallelism of algorithm, adopts subtree simultaneously
The method of multiplexing reduces operand, thus reducing instruction cycles, accelerating the algorithm speed of service, improving the throughput of decoding.With
Existing list is optimum（Kbest）Detection method is compared, and can be made using the inventive method while ensureing that performance does not decline
Obtain execution cycle number needed for multiaerial system detection process and averagely reduce 35%.
Brief description
Fig. 1 is the processing system based on the multi-antenna space division multiplexing system parallel detecting method of breadth-first search for the present invention
Schematic diagram.
Fig. 2 is for the present invention based on two-layer in the multi-antenna space division multiplexing system parallel detecting method of breadth-first search simultaneously
Calculating and the schematic diagram understanding from tree search angle of corresponding subtree multiplexing operation.
Specific embodiment
By specific embodiment, the present invention is described in further detail below in conjunction with the accompanying drawings.
Embodiment 1：
The processing system that Fig. 1 gives the multi-antenna space division multiplexing system parallel detecting method based on breadth-first search is shown
It is intended to.Pressed as follows by pretreatment module A1, tree search module A2 and output module A3 using the processing system of the inventive method
Configuration connects composition：The input of system accesses pretreatment module A1, and the outfan of pretreatment module A1 is with tree search module A2's
Input is connected, and the outfan of tree search module A2 is connected with the input of output module A3, the output of output module A3
Hold as system output.
The structure of wherein pretreatment module A1 is：One complex received signal vector y of systemcInput is linked into multiple real-turn
Change the input of unit B 1, another complex channel matrix H of systemcInput is linked into the input that unit B 2 is changed in multiple real-turn；
The real channel matrix H output that unit B 2 is changed in multiple real-turn is linked into the input just handing in triangle decomposition unit B 4；Multiple real-turn is changed
The real number receipt signal vector of unit B 1Output exports as square with an orthogonal matrix Q just handing in triangle decomposition unit B 4
The input of battle array multiplying unit B3；The real number vector y of matrix multiple unit B 3 exports and is just handing in the another of triangle decomposition unit B 4
Individual upper triangular matrix R is output as two outputs of pretreatment module A1；
The structure of described tree search module A2 is：One real number vector y output of pretreatment module A1 is linked into even level
The input of cost value computing unit C1 and the input of odd-level cost value computing unit C2, pretreatment module A2 another
Upper triangular matrix R output is linked into the input of even level cost value computing unit C1 and odd-level cost value computing unit C2's
Input；The outfan of the outfan of even level cost value computing unit C1 and odd-level cost value computing unit C2 is linked into be asked
Input with unit C3；The outfan of sum unit C3 is linked into the input of sequencing unit C4；Two of sequencing unit C4
Outfan is linked into the input of even level cost value computing unit C1 and the input of odd-level cost value computing unit C2, separately
One outfan is as the output setting search module A2；Subtree is adopted to be multiplexed to reduce calculating when wherein calculating odd-level cost value
Amount.
The structure of described output module A3 is：The output of tree search module A2 is as the real input answering converting unit D1, real
The output of multiple converting unit D1 is as the output of output module A4.
Set in the present embodiment the parameter of communication system as：4 transmitting antennas, 4 reception antenna 16QAM constellation modulation, choosing
Select list length parameter K=4.This communication system is y in the plural detection model of receiving terminalc=Hcsc+nc, wherein ycIt is that plural number connects
Receive signal phasor, scIt is sending signal vector to be detected, HcIt is complex channel matrix, ncIt is white Gaussian noise vector.Above-mentioned multiple
The concrete expanded form that element expansion pressed by number detection model is as follows：
（1）Pretreatment module A1 operates
First, by complex received signal vector ycWith complex channel matrix HcIt is separately input to the square of pretreatment module in Fig. 1
The corresponding multiple real-turn of battle array is changed in unit B 1 and B2.
Change in unit B 1 in multiple real-turn, by above-mentioned plural number detection model according to " real part-imaginary part-real part-imaginary part " arrangement
It is sequentially converted into real number detection model as follows, whereinIt is the real number receipt signal vector after converting and conduct replicates conversion
The output of unit B 1：
Change in unit B 2 in multiple real-turn, as follows complex channel matrix H c is converted to real channel matrix H, and
Change the output of unit B 2 as multiple real-turn：
In just handing in triangle decomposition unit B 4, multiple real-turn is changed unit B 2 output real channel matrix H carry out orthogonal
Upper triangle decomposition（QR decomposes）H=QR, wherein Q are orthogonal matrix, are connected to the input of matrix multiple unit B 3；R is upper triangle
Matrix, as the output of pretreatment module A1, and upper triangular matrix R has special nature：The diagonal line element of all even columns
The next element of element is 0, i.e. R2k,2k+1=0, k=1,2,3,4.
In matrix multiple module B3, by real number receipt signal vectorWith just hand in triangle decomposition unit output upper three
The transposition of angular moment battle array Q is multiplied, and obtains real number vector y, that is,Detailed process is as follows：
（2）Tree search module A2 operation
Output real number vector y and upper triangular matrix R from pretreatment module A1 is respectively connected to even level cost value
In computing unit C1 and odd-level cost value computing unit C2, initial build cost value PED9(s9)=0.
First calculate the 8th layer and the 7th layer, directly calculate the 8th layer of cost value（The i.e. part Euclidean distance of this layer）1 ＜ p ＜ 4,1 ＜ q ＜ 4, the value of wherein constellation point sets vector S is
{-3,-1,1,3}.Here calculating the 7th layer of cost valueOnly need to calculate 4, rather than 16, using subtree multiplexing
Property.Then the output of even level cost value computing unit C1Output with odd-level cost value computing unit C2It is input in sum unit C3, calculate accumulated costs value now1 ＜
L ＜ 16 exporting and being input in sequencing unit C4 as sum unit C3, to this 16 value sequences in sequencing unit C4, protects
Stay minimum 4, corresponding cost value is designated as1 ＜ k ＜ 4, survivor path nodal information is input to even level
In cost value computing unit Cl and odd-level cost value computing unit C2, corresponding survivor path cost value information input is to summation
In unit C4, sue for peace for the cost value that next round calculates.
Fig. 2 is for the present invention based on two-layer in multi-antenna space division multiplexing system parallel detection ten thousand method of breadth-first search simultaneously
Calculating and the schematic diagram understanding from tree search angle of corresponding subtree multiplexing operation.The partial enlargement arrow of Fig. 2 middle and lower part
Symbol represents and carries out the refinement content further after partial enlargement by calculating the 6th layer and the 5th layer of procedure division.
The specific operation process of described subtree multiplexing is as follows:
Next calculate the 6th layer and the 5th layer, even level cost value computing unit Cl calculates residual vector1 ＜ k ＜ 4,1 ＜ p ＜ 4, in odd-level cost value computing unit
Middle calculating1 ＜ k ＜ 4,1 ＜ q ＜ 4, notes here with son
The property of tree multiplexing, as shown in Fig. 2 local magnification region, that is, 16 expansions only calculating the 5th layer in Fig. 2 local magnification region
In exhibition child node first 4, i.e. 1～No. 4 extension child node, as shown in dotted line frame little in Fig. 2 local magnification region, then by this
The cost value of 4 nodes is directly assigned to the cost value of 5～No. 8,9～No. 12 and 13～No. 16 extension child nodes successively, here it is
Subtree multiplex process.Then the output of even level cost value computing unit C1With odd-level cost value computing unit C2's
OutputIt is input in sum unit C3, and and the last survivor path cost value letter inputting from sequencing unit C4
BreathSue for peace together, the accumulated costs value calculating now is
1 ＜ l ＜ 16,1 ＜ k ＜ 4, the cost value being so calculated the 5th layer of 64 child node to be selected is input in sequencing unit C4, warp
4 minimum survival child nodes of cost value are retained, corresponding cost value is designated as after crossing sequence1 ＜ k ＜ 4, will good fortune
Deposit path node information input in even level cost value computing unit C1 and odd-level cost value computing unit C2, corresponding good fortune
Deposit path cost value information input in sum unit C4, the cost value summation calculating for next round.
Repeat then to calculate the 3rd layer of accumulated costs value according to said process1 ＜ k ＜ 4, then calculates
64 last layers（I.e. the 1st layer）Accumulated costs value PED1(s1) value when, directly from this 64 values, select accumulated costs value
The corresponding survival node of a little value is as the output of tree search module A2.
（3）Output module A3 operates
By last layer survival node from tree search module A2 in multiple converting unit D1 of the reality in output module A3
The reverse transformation that puts in order according to " real part-imaginary part-real part-imaginary part " is complex vector located, and the final output as system.
Table 1 below is employing during different antennae number using 16QAM modulation and in the case of list length parameter K=4
The inventive method is optimum with using traditional list（Kbest）The required execution cycle number on parallel computing platform of detection method
Comparison.
Table 1
As it can be seen from table 1 it is significantly few using the required execution cycle number on parallel computing platform of the inventive method
Optimum in traditional list（Kbest）Detection method.On average, can be so that execution cycle number reduces 35% using the present invention
Left and right, thus greatly accelerating run time, improves the data throughput of decoding.
Claims (1)
1. a kind of multiaerial system parallel detecting method based on breadth-first search is it is characterised in that processing system is by pretreatment
Module（A1）, tree search module（A2）, and output module（A3）Connect composition by following configuration：The input of system accesses pre- locating
Reason module（A1）, pretreatment module（A1）Outfan with tree search module（A2）Input be connected, set search module
（A2）Outfan and output module（A3）Input be connected, output module（A3）Outfan be system output；
Wherein pretreatment module（A1）Structure be：One complex received signal vector of system（yc）Input is linked into multiple real-turn
Change unit（B1）Input, another complex channel matrix of system（Hc）Input is linked into multiple real converting unit（B2）Defeated
Enter end；Multiple reality converting unit（B2）Real channel matrix（H）Output is linked into just hands in triangle decomposition unit（B4）Input
End；Multiple reality converting unit（B1）Real number receipt signal vectorExport and just handing in triangle decomposition unit（B4）One just
Hand over matrix（Q）Output is as matrix multiple unit（B3）Input；Matrix multiple unit（B3）Real number vector y output and orthogonal
Upper triangle decomposition unit（B4）Another upper triangular matrix（R）It is output as pretreatment module（A1）Two output；
Described tree search module（A2）Structure be：Pretreatment module（A1）A real number vector（y）Output is linked into even number
Layer cost value computing unit（C1）Input and odd-level cost value computing unit（C2）Input, pretreatment module（A2）
Another upper triangular matrix（R）Output is linked into even level cost value computing unit（C1）Input and odd-level cost value
Computing unit（C2）Input；Even level cost value computing unit（C1）Outfan and odd-level cost value computing unit
（C2）Outfan be linked into sum unit（C3）Input；Sum unit（C3）Outfan be linked into sequencing unit（C4）
Input；Sequencing unit（C4）Two outfans be linked into even level cost value computing unit（C1）Input and odd number
Layer cost value computing unit（C2）Input, another outfan as tree search module（A2）Output；
Described output module（A3）Structure be：Tree search module（A2）Output answer converting unit as real（D1）Input,
Real converting unit again（D1）Output as output module（A4）Output；
Parallel detection operation is carried out as follows：First multiaerial system plural number detection model is converted into real number detection mould
Type, that is,：The real part of each element of complex received signal vector plural number sending signal vector and imaginary part are replaced and arranges, that is, press
Order arrangement according to " real part-imaginary part-real part-imaginary part "；Real channel matrix after complex channel matrixing is just carried out
Orthogonal matrix and upper triangular matrix is respectively obtained, the calculating using odd-level cost value does not rely on it after handing in triangle decomposition
This property of the result of preceding layer, is calculated to adjacent odd-level and even level simultaneously：The search tree-model that parity price goes out
Every two-layer calculate its cost value respectively, then the cost value of corresponding two nodes on same paths is added summation and obtains
The cost value of current level of child nodes；And when calculating even level cost value, its computing is reduced using the method for subtree multiplexing
Amount, that is, utilize the even level of the search tree property separate when calculating with odd-level, is calculating odd-level in even number
During the cost value of the individual child node of log2 (M) * log2 (M) that layer expands, before only calculating, log2 (M) is individual, then replicates log2 (M)
Secondary, M therein is order of modulation.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410139513.0A CN103873120B (en) | 2014-04-08 | 2014-04-08 | A kind of multiaerial system parallel detecting method based on breadth-first search |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410139513.0A CN103873120B (en) | 2014-04-08 | 2014-04-08 | A kind of multiaerial system parallel detecting method based on breadth-first search |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN103873120A CN103873120A (en) | 2014-06-18 |
| CN103873120B true CN103873120B (en) | 2017-03-01 |
Family
ID=50911289
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201410139513.0A Expired - Fee Related CN103873120B (en) | 2014-04-08 | 2014-04-08 | A kind of multiaerial system parallel detecting method based on breadth-first search |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN103873120B (en) |
Families Citing this family (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104618061B (en) * | 2015-01-29 | 2017-11-28 | 清华大学 | The detection method of multiple user signals in a kind of large-scale multi-antenna system |
| GB201509910D0 (en) * | 2015-06-08 | 2015-07-22 | Univ Surrey | Parallel processing of sphere decoders and other vector finding approaches using tree search |
| CN115001546B (en) * | 2022-05-26 | 2023-09-08 | 西安电子科技大学 | MIMO system signal detection method based on parallel K-Best search |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101674160A (en) * | 2009-10-22 | 2010-03-17 | 复旦大学 | Signal detection method and device for multiple-input-multiple-output wireless communication system |
| CN102281091A (en) * | 2011-07-22 | 2011-12-14 | 北京工业大学 | Reception method for multi-antenna communication system |
| CN103532609A (en) * | 2013-10-08 | 2014-01-22 | 上海交通大学 | MIMO (Multiple Input Multiple Output) signal detection method and detector on basis of K-Best algorithm |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US8391378B2 (en) * | 2008-03-11 | 2013-03-05 | Ikanos Communications, Inc. | Metric computation for lowering complexity of MIMO detection algorithms |
-
2014
- 2014-04-08 CN CN201410139513.0A patent/CN103873120B/en not_active Expired - Fee Related
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101674160A (en) * | 2009-10-22 | 2010-03-17 | 复旦大学 | Signal detection method and device for multiple-input-multiple-output wireless communication system |
| CN102281091A (en) * | 2011-07-22 | 2011-12-14 | 北京工业大学 | Reception method for multi-antenna communication system |
| CN103532609A (en) * | 2013-10-08 | 2014-01-22 | 上海交通大学 | MIMO (Multiple Input Multiple Output) signal detection method and detector on basis of K-Best algorithm |
Non-Patent Citations (1)
| Title |
|---|
| 低复杂度K-Best MIMO检测器设计;刘飞，黎海涛;《电路与系统学报》;20111231;第16卷(第6期);第1节第9-10行、第2节第1-4行、第3.1节、第3.3节 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN103873120A (en) | 2014-06-18 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| He et al. | Model-driven deep learning for physical layer communications | |
| CN101674160B (en) | Signal detection method and device for multiple-input-multiple-output wireless communication system | |
| JP2009100478A (en) | Wireless communication apparatus | |
| CN105359510B (en) | The device and method for transmitting broadcast singal, the device and method for receiving broadcast singal | |
| CN106130690A (en) | Mimo system joint-detection interpretation method in conjunction with polarization code | |
| JP2009135906A (en) | Radio communication equipment | |
| CN102857235B (en) | LDPC (low-density parity-check) encoder and encoding method in DTMB (digital terrestrial multimedia broadcasting) system based on shared register | |
| CN106253913B (en) | The block encoder of polarization code and its coding method | |
| WO2010078772A1 (en) | Space time coding method in orthogonal network and relay transmission system | |
| CN100373840C (en) | Method and apparatus for detecting normalized iterative soft interference cancelling signal | |
| CN103873120B (en) | A kind of multiaerial system parallel detecting method based on breadth-first search | |
| CN104580039A (en) | Receiver detection method assisted by lattice reduction algorithm and applied to wireless MIMO system | |
| Amiri et al. | A high throughput configurable SDR detector for multi-user MIMO wireless systems | |
| CN110429999A (en) | Extensive MIMO detection method based on lp-Box ADMM algorithm | |
| CN105978655A (en) | Packet data transmission method and system in wireless network | |
| CN105515627B (en) | A kind of extensive MIMO detection method and detection device | |
| CN102710393A (en) | Interference alignment precoding method based on Stiefel manifold | |
| CN105940625A (en) | Base station device, wireless communication system, and communication method | |
| CN107919895A (en) | A kind of Distributed Detection method of large-scale and multiple users mimo system | |
| Wang et al. | Cca-net: A lightweight network using criss-cross attention for csi feedback | |
| CN101964667B (en) | High-efficiency multi-antenna detection method for long term evolution scheme | |
| CN102882654B (en) | Encoding constraint and probability calculation based encoding and decoding synchronization method | |
| CN107919937B (en) | Decoding method and device based on overlapping multiplexing and modulation and demodulation method and system | |
| CN101102295A (en) | Method for space collection multiplexing and multi-input and output communication system | |
| CN105553614B (en) | Integrating device based on signal detection algorithm |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20170301 |