CN103532609B - MIMO signal detection method based on K Best algorithm and detector - Google Patents
MIMO signal detection method based on K Best algorithm and detector Download PDFInfo
- Publication number
- CN103532609B CN103532609B CN201310465151.XA CN201310465151A CN103532609B CN 103532609 B CN103532609 B CN 103532609B CN 201310465151 A CN201310465151 A CN 201310465151A CN 103532609 B CN103532609 B CN 103532609B
- Authority
- CN
- China
- Prior art keywords
- path
- node
- detection
- bar
- 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.)
- Active
Links
Abstract
The invention discloses a kind of MIMO signal detection method based on K Best algorithm and detector, the method comprises the steps：The antenna number specified according to user and the modulation system configuration detection number of plies；Obtain the product of the associate matrix y vectorial with reception of upper triangular matrix R after QR decomposition for the channel transfer matrix H and unitary matrice；The FC parallel path of all constellation point of pth=M shell and 1 layer of p=M is launched, produces two groups of path candidate parallel processings with interleaving mode；Path candidate is launched on demand find out the K bar survivor path of p layer and 2 discarding paths of K using parallel mode；The survivor path of each detection layers is processed and obtains survivor path output, and process to abandoning path and being reused；Path is abandoned to K bar survivor path and K 2 and carries out log-likelihood process, the present invention can solve conventional detection devices throughput lowly, very flexible, detection poor performance and hardware configuration are difficult the problems such as realize.
Description
Technical field
The present invention relates to MIMO communication technical field, more particularly to a kind of MIMO signal detection based on K-Best algorithm
Method and detector.
Background technology
Multiple-input and multiple-output (Multiple Input Multiple Output, MIMO) technology can not increase work(
Capacity and the reliability of wireless communication system under the cost of rate, time and spectral bandwidth, are significantly increased.Wherein, MIMO skill
The transfer rate that the spatial multiplexing modes of art enable to system assumes linear increase [1] with number of antennas.MIMO skill
This huge theoretical breakthrough of art causes academia and the huge emerging of industrial quarters takes, and also creates many achievements in research.MIMO
Technology is also adopted by increasing new generation of wireless communication standard.
Although MIMO technology has above huge advantage, also so it is widely used in each standard.But realize MIMO
The computational complexity of detection technique is also considerably beyond the development of hardware circuit.How to select a kind of excellent performance, calculate complexity
Degree is moderate, and the MIMO detection algorithm that can be easily and efficiently realized by super large-scale integration becomes asking of urgent need solution
Topic.
MIMO detection algorithm is generally divided into maximum likelihood algorithm, tree searching and detecting algorithm, successive interference cancellation to examine
Method of determining and calculating and linear detection algorithm.Although the detection of MIMO receiving end signal can be obtained by maximum likelihood algorithm, its
Computational complexity growth exponentially with the number of antenna and the growth of number of modulation levels, at present can not possibly be real with integrated circuit
Existing.Order according to traversal tree and the method for beta pruning, tree searching and detecting algorithm is divided three classes：Depth-first search globular decoding
（Sphere Decoding,SD）Algorithm；Width first traversal；Tolerance is preferential（Metric-first）Algorithm.K-Best detection is calculated
Method is the representative of width first traversal.
K-Best detection algorithm principle is the full candidate constellation points path launching every layer, and is sorted.By every depth amount
The minimum K paths of value elect the K-Best path of current layer as.The detection performance of algorithm and detection complexity can also be by K values
Adjustment reach balance.If system index is more strict for detection performance, can be realized by expanding K value.In the same manner, if uncommon
Hope the throughput rate obtaining lower computational complexity and Geng Gao it is necessary to less K value is for detecting.K-Best detection algorithm
Although not reaching optimal detection as Sphere Decoding Algorithm.But because not setting the backtracking of search in its algorithm characteristic, can
With the pipeline processes being advantageously used in VLSI.
According to the investigation to existing document, deliver on IEEE Journal one of Z.Guo and P.Nilsson2006
“Algorithm and Implementation of the K-Best Sphere Decoding for MIMO
Propose one kind in Detection " and support 4 × 4 antennas, the soft output detector structure that 16QAM modulates, although detector performance
To significantly improving, but complexity is still very high, and cannot expanded application in 64QAM modulating system.
" the Relaxed K-Best MIMO that S.Z.Chen and T.Zhang delivered on IEEE transaction in 2007
It is based on traditional K-Best soft decision detection in a Signal Detector Design and VLSI Implementation " literary composition
Algorithm, devises and is applied to 4 × 4 antennas, the soft output detector of 64QAM modulation.Because traditional detection algorithm complex is very high,
Be not suitable for realizing on hardware although reducing complexity to a certain extent using distributed sort method, but detector
Throughput and performance also suffer from limiting.
" the VLSI that D.Patel, V.Smolyakov et al. delivered on IEEE ISCAS in 2010
Implementation of a WiMAX/LTE Compliant Low-Complexity High-Throughput Soft-
4 × 4 antennas, the soft output detector of 64QAM modulation is devised in output K-Best MIMO Detector ".It adopts one
Series is approximate and method for simplifying greatly reduces detector complexity, but its configuration is fixing, does not possess present invention spirit
Work carries out antenna and the feature of modulation system configuration, and its detection performance is not also high.
Content of the invention
For overcoming the shortcomings of that above-mentioned prior art exists, the purpose of the present invention be to provide a kind of based on K-Best algorithm
MIMO signal detection method and detector, it can solve conventional detection devices throughput lowly, very flexible, detection poor performance
And hardware configuration is difficult the problems such as realize.
For reaching above and other purpose, the present invention proposes a kind of MIMO signal detection method based on K-Best algorithm, bag
Include following steps：
Step one, the antenna number specified according to user and the modulation system configuration detection number of plies；
Step 2, from MIMO detect pretreatment unit obtain channel transfer matrix H through QR decomposition after upper triangular matrix R and
The associate matrix of unitary matrice Q and the product receiving vectorial y
Step 3, after data enters K-Best detection, by the FC path of all constellation point of pth=M shell and p=M-1 layer simultaneously
Row launches, and produces two groups of path candidate parallel processings with interleaving mode；
Step 4, launches on demand to path candidate to find out the K bar survivor path of p layer using parallel mode and K-2 bar abandons
Path, every layer completed within K/2 clock cycle；
Step 5, is processed to the K bar survivor path of each detection layers, obtains final K bar survivor path output；
Step 6, processes to the K-2 bar discarding path of each detection layers, is reused to abandoning path；
Step 7, abandons path to K bar survivor path and K-2 bar and carries out log-likelihood process.
Further, in step 4, path candidate launches to carry out in such a way on demand：
Each father node is only launched with its best child node, K father node has altogether and produce K preferably child node；
This K preferably child node is ranked up, chooses the minimum best child node of accumulation PED, take out and replaced
Secondary good child node for its father node is put queue into and is repeated above-mentioned sequencer procedure, to the last obtains K optimum node.
Further, best child node by upper triangular matrix R matrix and receives vectorExtrapolate.
Further, the production method of secondary good child node adopts "the" shape search method.
Further, in step 5, the process for detection layers survivor path is as follows：If current pth layer is not leaf
The last layer of sublayer, that is, p be not equal to 2, then each in 2 groups of path candidates using finding out parallel by the way of accessing child node on demand
K/2 bar survivor path, if p=2 layer is detected, directly using the FC path of path candidate as the final survivor path detecting
For exporting.
Further, abandon path to be produced by procedure below：
Have accessed altogether K+K-1=2K-1 node during K bar survivor path is selected in the sequence of each detection layers, its
In finally pick out K node, remaining K-1 node be abandon node.
Further, in step 6, ZF is directly carried out to each layer of K-2 bar discarding path and extends to leaf section
Point.Then the PED utilizing this K-2 bar fullpath updates soft information value.
Further, this signal detecting method be applied to 1 × 1,1 × 2,1 × 3,1 × 4,2 × 2,2 × 3,2 × 4,3 × 3,
3 × 4,4 × 4 antenna configurations, the configurable multimode detector of BPSK, QPSK, 16-QAM and 64-QAM modulation.
For reaching above-mentioned purpose, the present invention also provides a kind of MIMO signal detector based on K-Best algorithm, at least wraps
Include：
KPE module, is responsible for selecting each detection layers to obtain optimum K node, its antenna number specified according to user and tune
Mode processed, to configure the detection number of plies, input data is divided into 2 batches of parallel processings, and often a collection of detection path replaces original K with K/2
Value；
DPE module, is responsible for this KPE module is reused in the discarding path that each detection layers obtain；
LLR module, configures according to different antennae, and the detection output of this KPE module and this DPE module is strobed into LLR mould
Carry out Soft Inform ation calculating in block；
Control module, controls different submodules by reception antenna configuration control signal and modulation system control signal
Work or stopping.
Further, this control module also produces gating signal, is configured according to different antennae, by different KPEn_n+1 modules
Detection output with DPE_Ln module is strobed in this LLR module and carries out Soft Inform ation calculating.
Compared with prior art, a kind of MIMO signal detection method based on K-Best algorithm of the present invention and detector are in inspection
The mode that during survey, path candidate launches on demand can greatly improve expansion efficiency, and intertexture parallel processing can not affect detection property
Under the premise of energy, improve data throughput 60%, the mode abandoning path reuse can expand the path candidate during Soft Inform ation computing
Number, thus increasing the accuracy of LLR, leaf layer does not need accurately to sort and just obtains the mode of survivor path and can not affect yet
Reduce time delay on the premise of BER performance.
Brief description
Fig. 1 is a kind of flow chart of steps of the MIMO signal detection method based on K-Best algorithm of the present invention；
Fig. 2 is a kind of thin portion flow process of the preferred embodiment of the MIMO signal detection method based on K-Best algorithm of the present invention
Figure；
Fig. 3 is a kind of system architecture diagram of the MIMO signal detector based on K-Best algorithm of the present invention；
Fig. 4 is a kind of structure of the preferred embodiment of the configurable MIMO signal detector based on K-Best algorithm of the present invention
Schematic diagram；
Fig. 5 is top-level module port information schematic diagram in present pre-ferred embodiments；
Fig. 6 is that in present pre-ferred embodiments, KPE1_2 internal module divides figure；
Fig. 7 is the structure chart of KPE2_3 internal module in present pre-ferred embodiments；
Fig. 8 is DPE_L2 internal module partition structure schematic diagram in present pre-ferred embodiments；
Fig. 9 is the hardware architecture diagram of Update_BMT module in present pre-ferred embodiments；
Figure 10 is that a kind of MIMO signal detector based on K-Best algorithm proposed by the present invention is representational with other
The Performance Simulation Results comparison diagram of K-Best detection algorithm.
Specific embodiment
Below by way of specific instantiation and embodiments of the present invention are described with reference to the drawings, those skilled in the art can
Understand further advantage and effect of the present invention by content disclosed in the present specification easily.The present invention also can be by other different
Instantiation implemented or applied, the every details in this specification also can be based on different viewpoints and application, without departing substantially from
Carry out various modification and change under the spirit of the present invention.
Fig. 1 is a kind of flow chart of steps of the MIMO signal detection method based on K-Best algorithm of the present invention.As Fig. 1 institute
Show that a kind of MIMO signal detection method based on K-Best algorithm of the present invention comprises the steps：
Step 101, the antenna number specified according to user and the modulation system configuration detection number of plies；
From MIMO, step 102, detects that pretreatment unit obtains upper triangular matrix R after QR decomposition for the channel transfer matrix H
And the associate matrix of unitary matrice and the product receiving vectorial y
Step 103, after data enters K-Best detection, by pth=M shell（Initiation layer）All constellation point and p=M-1 layer
FC（First Child, best child node）Parallel path launches, and produces two groups of path candidate parallel processings with interleaving mode；
Step 104, launches on demand to path candidate to find out the K bar survivor path of p layer using parallel mode and K-2 bar abandons
Path, every layer completed within K/2 clock cycle.
Step 105, is processed to the K bar survivor path of each detection layers, obtains final K bar survivor path output.
Step 106, processes to the K-2 bar discarding path of each detection layers, is reused to abandoning path, to each layer
K-2 bar abandon path and directly carry out ZF (Zero Forcing, ZF) and extend to leaf node, then utilize this K-2 bar complete
The PED in whole path updates soft information value, to increase the number of path candidate.Due to being divided into 2 pipeline processes, every streamline
Select K/2 bar survivor path, produce K/2-1 bar and abandon path, therefore 2 streamlines abandon path and are added together is exactly 2* (K/
2-1)=K-2 bar abandons path.
Step 107, abandons path to K bar survivor path and K-2 bar and carries out LLR（Log-likelihood ratio, logarithm
Likelihood）Calculate, thus updating Soft Inform ation.
Fig. 2 is a kind of thin portion flow process of the preferred embodiment of the MIMO signal detection method based on K-Best algorithm of the present invention
Figure.Below in conjunction with Fig. 2, the present invention is further described based on the MIMO signal detection method of K-Best algorithm.
In present pre-ferred embodiments, in step 104, path candidate launches to carry out in such a way on demand：
Each father node is only launched with its best child node（First Child, abbreviation FC).FC point can be by upper three angular moments
Battle array R matrix and reception vectorExtrapolate.This extended mode is referred to as ZF（ZF Zero Forcing）Extension.K
Father node has altogether and produces K FC node.Subsequently this K FC point is ranked up, chooses accumulation PED（Partial Euclidean
Distance, part Euclidean distance）（I.e.：Each constellation point Euclidean distance add up and）Minimum FC point, takes out and is replaced with
The secondary good child node of its father node（Next Child, abbreviation NC）Put queue into and repeat above-mentioned sequencer procedure, to the last
To K optimum node.
The production method of secondary good child node NC can adopt "the" shape search method, is exactly by best child node FC in brief
Threshold values between the position of point and constellation point（For example 64QAM is modulated, after real number decomposes, constellation point coordinate figure is -7, -
5, -3...7, then can arrange threshold values is -6, -4, -2 etc.) make comparisons so that it is determined that the direction of search of NC.Search for real for convenience
Existing, constellation point coordinates can be mapped to 0,1,2,3...7.After mapping, way of search just becomes relatively easy.Assume FC
Point falls within 1.2 positions after mapping, then its nearest integral point 1 of selected distance is as its mapping symbols, and now determines
Its next step direction of search is just, step-size in search initial value is 1, and the position of therefore NC point is 2.First when subsequently searching for each time
First to judge whether to have reached the boundary point (0 or 7 of search）.Assume that Searching point has reached 0, then the direction of search afterwards
Just it is all just, and step-size in search remains 1.Otherwise step-size in search Jia 1 every time.
In step 105, the process for detection layers survivor path is as follows：If current pth layer is not the upper of leaf layer
One layer, that is, p is not equal to 2, then using finding out the good fortune of each K/2 bar in 2 groups of path candidates by the way of accessing child node on demand parallel
Depositing path, if p=2 layer is detected, directly the FC path of path candidate being used for as the survivor path of final detection defeated
Go out.
Discarding path described in step 104 is produced by procedure below：
Have accessed altogether K+K-1=2K-1 node during K bar survivor path is selected in the sequence of each detection layers, its
In finally pick out K node, remaining K-1 node be abandon node.These abandon node is to remove K optimum section in current layer
Best, if they are given up obviously wasted very much outside point.The degree of reliability that Soft Inform ation calculates is heavily dependent on time
The number in routing footpath, therefore, in order to allow soft information value become more reliable, can be in addition sharp to K-1 node of each layer of discarding
With.ZF is directly carried out to each layer of K-1 bar discarding path and extends to leaf node.Then utilize this K-1 bar fullpath
PED updates soft information value, can greatly increase the number of path candidate.
Except ground floor will not produce discarding path in detection process, as long as remaining detection layers has survival paths ordering
The process selecting will produce discarding path.In fact for which floor detection last, because error propagation scope has compared
Little, the raising to performance does not have much affect for the expansion in discarding path.Therefore detection layers last which floor take simplification
Process, no longer extension abandons path.
In step 106, as follows for the process abandoning path：Due to detecting the configurability of antenna, therefore different
Antenna configurations will reuse the discarding path of the different numbers of plies.end(M)<=p<In=start (M), end (M) and start (M) is real
Number decomposes the function that aft antenna configures number of plies M.For 4 × 4 antenna configurations, start (M)=7, end (M)=4；3 × 3 antennas are joined
Put, start (M)=5, end (M)=3；2 × 2 antenna configurations, start (M)=3=end (M).
Fig. 3 is a kind of system architecture diagram of the MIMO signal detector based on K-Best algorithm of the present invention.As shown in figure 3,
A kind of MIMO signal detector based on K-Best algorithm of the present invention, including：KPE module 10, DPE module 20, LLR module 30 with
And control module 40.
Wherein, the responsible selection of KPE (K-Best Processing Element, K-Best processing unit) module 10 is each
Detection layers obtain optimum K node, and its antenna number specified according to user and modulation system to configure the detection number of plies, will input number
According to being divided into 2 batches of parallel processings, often a collection of detection path replaces original K value with K/2；DPE(Depleted paths
Processing Element, abandon path processing unit) module 20 be responsible for KPE module 10 in the discarding road of each detection layers
Footpath reuses；LLR（Log-likelihood ratio, log-likelihood）Module, configures according to different antennae, by KPE module and DPE
The detection output of module is strobed in LLR module and carries out Soft Inform ation calculating, thus updating Soft Inform ation；Control module 40 is passed through to receive
Antenna configurations control signal mod_mode and modulation system control signal ant_mode are controlling the work of different submodules or to stop
Only.
Fig. 4 is a kind of structural representation of the preferred embodiment of the MIMO signal detector based on K-Best algorithm of the present invention
Figure, Fig. 5 is top-level module port information schematic diagram in present pre-ferred embodiments.In present pre-ferred embodiments,
Scal_qy_in signal is QHIt is defeated that y signal obtains after different coefficient factor under different modulating mode quantifies
Enter data.The purpose being multiplied by coefficient factor is that in K-Best detection afterwards, the constellation point symbol that real number decomposes can represent
For integer, facilitate detection calculations.R_mat is that H-matrix carries out the upper triangular matrix after MMSE-SQRD decomposition.Order signal is
The result of preprocessing ranking.The last output LLR value of K-Best detection needs for clooating sequence to revert to input sequence.mod_
Mode and ant_mode is antenna configurations control signal and the modulation system control signal of detector.Clip_val and n0 is last
LLR value calculates required clipping value and N0 value.
In present pre-ferred embodiments, because the MIMO signal detector of the present invention takes K value to take 10 sorting in parallel
Detection mode, therefore the maximum data throughput are 5 clock cycle to process 1 batch data.In_valid signal puts high 1 clock week
It is effective that phase represents a collection of input data.Can be with 5 clock cycle built-in high 1 clock in the case of therefore in_valid signal is the fastest
Cycle.Kbest_req is handshake of fetching data, and represents to channel pretreatment module new data can be had to pass to this when putting high
The MIMO signal detector of invention.
In present pre-ferred embodiments, KPE module has multiple KPEn_n+1 modules, for selecting each detection layers to obtain
Obtain optimum K node, DPE module also has multiple DPE_Ln modules.Control module in Fig. 4（core_CTRL）It is responsible for inspection
Survey device to carry out configuring control.Its major function is to control letter by reception antenna configuration control signal mod_mode and modulation system
Number ant_mode is controlling work or the stopping of different submodules.Also produce gating signal, configured according to different antennae, by difference
The detection output of KPEn_n+1 module and DPE_Ln module is strobed into LLR（Log-Likelihood Ratio log-likelihood ratio）Mould
Block（LLR_Unit）In carry out Soft Inform ation calculating.For example, under 3 × 3 antenna configurations, core_CTRL be not then KPE7_8 and
DPE_L5 produces and enables signal, and the output of KPE6_7 and DPE_L4 is strobed into LLR module（LLR_Unit）In carry out soft
Information calculates.
From fig. 4, it can be seen that KPE1_2 modular concurrent launches the 1st layer of all constellation point and self-corresponding 2nd layer of Qi Ge
FC path.After the sequence of PED value is pressed in 2nd layer of FC path, it is interleaved and is divided into two groups.Follow-up testing circuit enters to this two groups of paths
Row concurrently launches detection on demand.According to different antenna configurations situations, the discarding path of the 2nd, 3,4 and 5 layers of generation（DP）Will
It is reused, carry out ZF and extend to the calculating for Soft Inform ation for the total length.
By launching the survivor path of leaf layer that obtains on demand and to abandon, through total length, the BMT that routing update crosses last all
Enter the calculating that LLR_Unit module carries out Soft Inform ation.
Because 1 × 1,2 × 2,3 × 3 and 4 × 4 antenna configurations supported by detector.Therefore under different antennae configuring condition,
The output of different KPE modules will be input to LLR_Unit module by gating signal.For example, in the case of 2 × 2 antenna configurations,
The output of KPE4_5 will be directed into LLR_Unit module.
By continuing cooperation Fig. 4 and Fig. 5, module each in present pre-ferred embodiments is described further below.
（1）KPE1_2 module
In present pre-ferred embodiments, KPE1_2 module completes following operation：
Complete the signal detection of the 1st layer of antenna, (W-QAM modulates, and real number divides to launch the 1st layer of all possible constellation point
Solution), and calculate its PED.Computing formula is as follows：
Carry out ZF parallel to the 1st layer of individual constellation point launched to launch to obtain respective First Child (FC).And calculate
Corresponding PED_L2.Launch FC formula be：
Wherein [x] expression takes the integer the most close with numerical value x in bracket, represents the FC of the 2nd layer of certain paths.
Each PED_L2 is carried out sorting in parallel according to size, obtains tactic PED_L2 sequence.Each PED is in storage
When after have 3 extra orders, in order to store respective ID of trace route path, because sorting in parallel can upset original storage order,
ID of trace route path can record the storage location corresponding to the PED in certain FC path, is easy to subsequent detection.
KPE1_2 internal module divides figure such as Fig. 6
The concrete function of each module is as shown in table 1
Table 1KPE1_2 submodule function describes
（2）KPEn_n+1 module
KPE2_3 realizes following operation：
Receive the 2nd layer to have sortedBar FC path, the PED value according to them is launched to select the 2nd layer of K/2 on demand
Bar K-Best survivor path.One clock cycle selects a K-Best survivor path, and the Next Child node generation with it
Replace and proceed to sort.
For K/2 article of survivor path of the 2nd layer of generation, flowing water launches their the 3rd layer FC paths successively, calculates its correspondence
PED_L3, and the parameter of NC.And according to PED value, the 3rd layer of FC path is ranked up, Sequential output is to next stage.
The structure chart of KPE2_3 internal module such as Fig. 7
The concrete function of each module is as shown in table 2
Table 2KPE1_2 submodule function describes
（3）DPE_Ln module
The hardware configuration thought of DPE_L2～DPE_L5 is identical, therefore here, illustrates the function of DPE module taking DPE_L2 as a example
And hardware implementation mode.
DPE_L2 module receives two KPE2_3 modules and is detecting 6 articles (the 64-QAM modulation) or 2 articles producing when the 2nd layer
(16-QAM modulation) abandon path, ZF is carried out to every paths and is expanded to leaf node, obtain whole piece path constellation point and
Its corresponding path metric；Then generate BMT by this 6 or 2 fullpaths, pass to next stage DPE unit and be updated.
Path after every ZF extension can update BMT（Bit Metric Table, for storing every minimum Europe
The table of formula distance, when BMT is used for recording every bit for 1 or 0, the PED value in each self-corresponding minimum total length path) once,
The path of only current examination is in a certain position xiCorresponding path metric is less than the path metric of this storage at present in BMT
When, then it is replaced.
ZF estimation be during drops down every layer all calculate FC point, this module will successively by abandon path from the 3rd layer start count
Calculate FC up to leaf node.The process of realization can be shown in table 3 below：
Table 3DPE_L2 algorithm flow
Fig. 8 is DPE_L2 internal module partition structure schematic diagram in present pre-ferred embodiments.Under being modulated due to 64-QAM,
2nd layer of detection can produce 6 articles of discarding paths, and data highest throughput be 5 clock cycle process batch of data it is therefore desirable to
6 discarding paths of 2 DPE_L2_ZF processing module parallel processings could meet data throughput.
The function of the internal submodule of DPE_L2 described in detail by table 4.
Table 4DPE_L2 submodule function describes
The discarding path extending to total length in DPE_L2_ZF enters Update_BMT module generation BMT.Update_BMT
The hardware configuration of module such as Fig. 9.
Wherein, S1～S8 represents the constellation point in the discarding path opened up to total length.If antenna configurations pattern is M × M, S1
The symbol of～S2M is that effective total length abandons path constellation point.
De-Map module（Modular converter）The constellation symbol index of input is changed into corresponding bit position.
BMC module（Comparison module）Will expand to the PED value in discarding path and its path corresponding bit position in BMT of total length
The PED value that numerical value is stored is compared, if the PED value abandoning path is less, the PED value of corresponding bits position in BMT is carried out
Update, the value otherwise retaining in BMT is constant.
Figure 10 is that a kind of MIMO signal detector based on K-Best algorithm proposed by the present invention is representational with other
The Performance Simulation Results comparison diagram of K-Best detection algorithm.The software platform based on Matlab and C language for the emulation of the present invention.Letter
Road adopts the Rayleigh channel that element is Gauss distribution, and it is 1/3 that encoding and decoding adopt code check, and code length is 1024 Turbo code.
From fig. 10 it can be seen that the MIMO signal detector based on K-Best algorithm of the present invention adopts grouping parallel to sort
K=10 has almost consistent detection performance with serial sort K=8.And hardware realize on, take grouping parallel sequence based on
The detection of K-Best improves 60% than the throughput of improved serial sort.
In sum, a kind of MIMO signal detection method based on K-Best algorithm of the present invention and detector are in detection
The mode that routing footpath is launched on demand can greatly improve expansion efficiency, and intertexture parallel processing can not affect to detect performance premise
Under, improve data throughput 60%, the mode that discarding path reuses can expand path candidate number during Soft Inform ation computing, from
And increasing the accuracy of LLR, leaf layer does not need accurately to sort and just obtains the mode of survivor path and can not affect BER performance yet
On the premise of reduce time delay, compared with prior art, data throughput of the present invention and detection performance on have significant advantage.
Above-described embodiment only principle of the illustrative present invention and its effect, not for the restriction present invention.Any
Skilled person all can be modified to above-described embodiment and changed without prejudice under the spirit and the scope of the present invention.Therefore,
The scope of the present invention, should be as listed by claims.
Claims (7)
1. a kind of MIMO signal detection method based on K-Best algorithm, comprises the steps：
Step one, the antenna number specified according to user and the modulation system configuration detection number of plies；
From MIMO, step 2, detects that pretreatment unit obtains upper triangular matrix R after QR decomposition for the channel transfer matrix H and tenth of the twelve Earthly Branches square
Battle array Q1Associate matrix and the product receiving vectorial y
Step 3, after data enters K-Best detection, by the best child node of all constellation point of pth=M shell and p=M-1 layer
Parallel path launches, and produces two groups of path candidate parallel processings with interleaving mode；
Step 4, launches on demand to path candidate to find out the K bar survivor path of p layer using parallel mode and K-2 bar abandons path,
Every layer completed within K/2 clock cycle；In this step 4, path candidate launches to carry out in such a way on demand：
Each father node is only launched with its best child node, K father node has altogether and produce K preferably child node；
This K preferably child node is ranked up, chooses the minimum best child node of accumulation PED, take out and be replaced with it
The secondary good child node of father node is put queue into and is repeated above-mentioned sequencer procedure, to the last obtains K optimum node；
Step 5, is processed to the K bar survivor path of each detection layers, obtains final K bar survivor path output；This step 5
In, if current pth layer is not the last layer of leaf layer, that is, p is not equal to 2, then by the way of accessing child node on demand simultaneously
Row finds out each K/2 bar survivor path in 2 groups of path candidates, if p=2 layer is detected, directly that path candidate is best
Child node path is used for exporting as the survivor path of final detection；
Step 6, processes to the K-2 bar discarding path of each detection layers, is reused to abandoning path；And, to each layer
K-2 bar abandon path and directly carry out ZF and extend to leaf node, then utilize the PED of this K-2 bar fullpath to update soft
The value of information；
Step 7, abandons path to K bar survivor path and K-2 bar and carries out log-likelihood process.
2. the MIMO signal detection method based on K-Best algorithm as claimed in claim 1 it is characterised in that：Preferably child node
By upper triangular matrix R matrix and reception vectorExtrapolate.
3. the MIMO signal detection method based on K-Best algorithm as claimed in claim 1 it is characterised in that：Secondary good child node
Production method adopt "the" shape search method.
4. the MIMO signal detection method based on K-Best algorithm as claimed in claim 1 it is characterised in that abandon path by
Procedure below produces：
Have accessed altogether K+K-1=2K-1 node during K bar survivor path is selected in the sequence of each detection layers, wherein
After pick out K node, remaining K-1 node be abandon node.
5. the MIMO signal detection method based on K-Best algorithm as claimed in claim 4 is it is characterised in that this signal detection
Method is applied to 1 × 1,1 × 2,1 × 3,1 × 4,2 × 2,2 × 3,2 × 4,3 × 3,3 × 4,4 × 4 antenna configurations, BPSK,
The configurable multimode detector of QPSK, 16-QAM and 64-QAM modulation.
6. a kind of MIMO signal detector based on K-Best algorithm, at least includes：
Processing unit module, is responsible for selecting each detection layers to obtain optimum K node, its antenna number specified according to user and tune
Mode processed, to configure the detection number of plies, input data is divided into 2 batches of parallel processings, and often a collection of detection path replaces original K with K/2
Value；This processing unit module includes each father node is only launched its best child node, and K father node has altogether and produce K
Good child node, this K preferably child node is ranked up, and chooses the minimum best child node of accumulation PED, takes out and replaced
Secondary good child node for its father node is put queue into and is repeated above-mentioned sequencer procedure, to the last obtains K optimum node；
Abandon path processing unit module, be responsible for this processing unit module is reused in the discarding path that each detection layers obtain；
This discarding path processing unit module includes abandoning path to each layer of K-2 bar and directly carries out ZF and extend to leaf node,
Then the PED utilizing this K-2 bar fullpath updates soft information value；
Log-likelihood module, configures according to different antennae, by this processing unit module and this discarding path processing unit module
Detection output is strobed in log-likelihood module and carries out Soft Inform ation calculating；
Control module, controls the work of different submodules by reception antenna configuration control signal and modulation system control signal
Or stop.
7. as claimed in claim 6 a kind of MIMO signal detector based on K-Best algorithm it is characterised in that：This process list
Element module and abandon the quantity of path processing unit module and be multiple, this control module also produces gating signal, according to not
Same antenna configurations, the detection output of different processing units module and discarding path processing unit module is strobed into this log-likelihood
Carry out Soft Inform ation calculating in module.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201310465151.XA CN103532609B (en) | 2013-10-08 | 2013-10-08 | MIMO signal detection method based on K Best algorithm and detector |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201310465151.XA CN103532609B (en) | 2013-10-08 | 2013-10-08 | MIMO signal detection method based on K Best algorithm and detector |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN103532609A CN103532609A (en) | 2014-01-22 |
| CN103532609B true CN103532609B (en) | 2017-03-01 |
Family
ID=49934336
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201310465151.XA Active CN103532609B (en) | 2013-10-08 | 2013-10-08 | MIMO signal detection method based on K Best algorithm and detector |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN103532609B (en) |
Families Citing this family (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103873120B (en) * | 2014-04-08 | 2017-03-01 | 中国科学技术大学 | A kind of multiaerial system parallel detecting method based on breadth-first search |
| EP3337112A1 (en) * | 2016-12-19 | 2018-06-20 | Institut Mines-Telecom | Methods and devices for sub-block decoding data signals |
| CN109981503B (en) * | 2019-03-31 | 2021-09-21 | 华南理工大学 | Layered k-best spherical decoding method and device |
| CN110932762B (en) * | 2019-10-29 | 2022-05-17 | 上海交通大学 | MIMO detection-oriented lattice reduction-assisted channel preprocessing method and device |
| WO2023108436A1 (en) * | 2021-12-14 | 2023-06-22 | 北京小米移动软件有限公司 | Signal detection method and apparatus |
Citations (2)
| 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 |
| CN102246452A (en) * | 2011-06-28 | 2011-11-16 | 华为技术有限公司 | Detection method and system for MIMO system |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US8488721B2 (en) * | 2009-08-20 | 2013-07-16 | Electronics And Telecommunications Research Institute | Adaptive QRD-M algorithm based signal detecting method by using constellation set grouping in spatial multiplexing multiple-input multiple-output system |
| WO2011150428A1 (en) * | 2010-05-28 | 2011-12-01 | Maxlinear, Inc. | Method and system for a low-complexity soft-output mimo detection |
-
2013
- 2013-10-08 CN CN201310465151.XA patent/CN103532609B/en active Active
Patent Citations (2)
| 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 |
| CN102246452A (en) * | 2011-06-28 | 2011-11-16 | 华为技术有限公司 | Detection method and system for MIMO system |
Non-Patent Citations (1)
| Title |
|---|
| MIMO-OFDM系统自适应排序K-best检测器研究;朱勇旭,吴斌,周玉梅,石殊;《重庆邮电大学学报(自然科学版)》;20110228;第23卷(第2期);全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN103532609A (en) | 2014-01-22 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN103532609B (en) | MIMO signal detection method based on K Best algorithm and detector | |
| Liu et al. | A 1.1-Gb/s 115-pJ/bit Configurable MIMO Detector Using 0.13-$\mu\hbox {m} $ CMOS Technology | |
| CN107005504A (en) | Method and device for the data in the tree searching and detecting cordless communication network by reducing complexity | |
| Liao et al. | A 74.8 mW Soft-Output Detector IC for 8$\,\times\, $8 Spatial-Multiplexing MIMO Communications | |
| Wu et al. | Implementation of a high throughput soft MIMO detector on GPU | |
| CN103888217B (en) | A kind of globular decoding detection method and device | |
| CN101432972A (en) | Radix-4 Viterbi decoding | |
| Yan et al. | Design and implementation of flexible dual-mode soft-output MIMO detector with channel preprocessing | |
| CN102006148A (en) | Multiple-input multiple-output (MIMO) signal detection method based on breadth-first tree search | |
| Amiri et al. | A high throughput configurable SDR detector for multi-user MIMO wireless systems | |
| Bello et al. | VLSI implementation of a fully-pipelined K-best MIMO detector with successive interference cancellation | |
| CN101541023B (en) | Joint iterative detection decoding method and device thereof | |
| WO2015131840A1 (en) | Detection method and apparatus of mimo system | |
| CN101964667B (en) | High-efficiency multi-antenna detection method for long term evolution scheme | |
| CN105553614B (en) | Integrating device based on signal detection algorithm | |
| CN102307080A (en) | Method and device for detecting serial block signal in MIMO (multiple-input multiple-output) system | |
| CN109660473B (en) | Spherical decoding detection method and device and computer readable storage medium | |
| Heidmann et al. | Architecture and FPGA-implementation of a high throughput K+-Best detector | |
| Fasthuber et al. | Novel energy-efficient scalable soft-output SSFE MIMO detector architectures | |
| CN102724160A (en) | Signal detecting method in high-order modulated multi-input multi-output system | |
| CN107342802A (en) | Random antenna system of selection, device and extensive mimo system | |
| CN106850475B (en) | Low-complexity sequencing A detection algorithm applied to spatial modulation | |
| Studer et al. | VLSI implementation of hard-and soft-output sphere decoding for wide-band MIMO systems | |
| CN109981503A (en) | A kind of layering k-best spherical decoding method and device | |
| CN107979389A (en) | For General Packet Radio Service（GPRS）The Sequences of Low Complexity estimator of system |
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 |