CN106230493B - A kind of selection of multiuser MIMO uplink antenna and user scheduling method - Google Patents
A kind of selection of multiuser MIMO uplink antenna and user scheduling method Download PDFInfo
- Publication number
- CN106230493B CN106230493B CN201610879164.5A CN201610879164A CN106230493B CN 106230493 B CN106230493 B CN 106230493B CN 201610879164 A CN201610879164 A CN 201610879164A CN 106230493 B CN106230493 B CN 106230493B
- Authority
- CN
- China
- Prior art keywords
- user
- antenna
- algorithm
- selection
- branch
- 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
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04B—TRANSMISSION
- H04B7/00—Radio transmission systems, i.e. using radiation field
- H04B7/02—Diversity systems; Multi-antenna system, i.e. transmission or reception using multiple antennas
- H04B7/04—Diversity systems; Multi-antenna system, i.e. transmission or reception using multiple antennas using two or more spaced independent antennas
- H04B7/0413—MIMO systems
- H04B7/0452—Multi-user MIMO systems
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04B—TRANSMISSION
- H04B7/00—Radio transmission systems, i.e. using radiation field
- H04B7/02—Diversity systems; Multi-antenna system, i.e. transmission or reception using multiple antennas
- H04B7/04—Diversity systems; Multi-antenna system, i.e. transmission or reception using multiple antennas using two or more spaced independent antennas
- H04B7/06—Diversity systems; Multi-antenna system, i.e. transmission or reception using multiple antennas using two or more spaced independent antennas at the transmitting station
- H04B7/0602—Diversity systems; Multi-antenna system, i.e. transmission or reception using multiple antennas using two or more spaced independent antennas at the transmitting station using antenna switching
- H04B7/0608—Antenna selection according to transmission parameters
Landscapes
- Engineering & Computer Science (AREA)
- Computer Networks & Wireless Communication (AREA)
- Signal Processing (AREA)
- Radio Transmission System (AREA)
- Mobile Radio Communication Systems (AREA)
Abstract
The present invention relates to a kind of selection of multiuser MIMO uplink antenna and user scheduling methods, downward branch-and-bound (Downwards Branch and Bound based on the selection of user's semi-orthogonal, DBAB), local iteration's optimizing algorithm search channel matrix maximum MSV realizes joint antenna selection and user's scheduling；For the algorithm using channel capacity as evaluation index, the rate capability of local optimum is more preferable, approaches exhaust algorithm (Brute-Force Search, BFS)；Meanwhile the computation complexity of the algorithm is far below the method for exhaustion, has larger application value, is a kind of promising day line options and user scheduling method in multi-user MIMO system.
Description
Technical field
The present invention relates to forth generations and the 5th third-generation mobile communication field.To solve the above problems, the invention proposes one kind
The selection of multiuser MIMO uplink antenna and user scheduling method (Joint Antenna Selection and User
Scheduling, JASUS), the rate capability of the local optimum of the algorithm is more preferable, approaches exhaust algorithm (Brute-Force
Search,BFS)；Meanwhile the computation complexity of the algorithm is far below the method for exhaustion, has larger application value.Therefore the present invention is more
It is a kind of promising day line options and user scheduling method in user's mimo system.
Background technique
In modern and future mobile communication system, base station services multiple users simultaneously by deployment aerial array, from
And constitute multi-user's multiple-input and multiple-output (Multiple-Input Multiple-Output, MIMO) system.
In the past decade, the MIMO technology in wireless communication is with its higher power system capacity and spectrum efficiency and by generation
Researcher's extensive concern within the scope of boundary [is shown in document 1:G.Xu, A.Liu, W.Jiang, H.Xiang, and W.Luo, " Joint
user scheduling and antenna selection in distributed massive MIMO systems
with limited backhaul capacity,”China Communications,,vol.11,no. 5,pp.17-30,
2014].In commercial LTE system, the communication system based on MIMO technology has been the online experience band of mobile subscriber
Tremendous increase is carried out.
In point-to-point MIMO communication, sender unit and receiver all include more antennas, it can be achieved that the space communicated
Diversity and multiplexing.Theoretically, in the case where not increasing extra bandwidth and sending power, capacity acquired by mimo system with
The minimum value of system dual-mode antenna quantity linearly increase [see document 2:F.Rusek, D.Persson, L.Buon Kiong,
E.G.Larsson,T.L.Marzetta,O.Edfors,and F.Tufvesson, “Scaling up MIMO:
opportunities and challenges with very large arrays,”IEEE Signal Processing
Mag.,vol.30,pp.40-60,2013].In mobile cellular communication, multiuser MIMO (Multi-User MIMO, MU-
MIMO) communication system has been widely used.When mobile base station obtains the channel state information of itself and different user
When (Channel State Information, CSI), base station can use suitable precoding (wave beam forming of broad sense) skill
Art gives multiple users to provide information services simultaneously, significantly improve traffic rate [see document 3-5:H.Q.Ngo, E.G.Larsson,
and T.L.Marzetta, “Energy and spectral efficiency of very large multiuser
MIMO systems,"IEEE Trans. Commun.,vol.61,no.4,pp.1436–1449,Apr.2012；L.Dai,
Z.Wang,and Z.Yang, “Spectrally efficient time-frequency training OFDM for
mobile large-scale MIMO systems,”IEEE J.Sel.Areas Commun.,vol.31,no.2,pp.251–
263,Feb.2013；Z.Lu, J.Ning,Y.Zhang,T.Xie,and W.Shen,"Richardson method based
linear precoding with low complexity for massive MIMO systems,”in Proc.of
IEEE 81st VTC Spring, 2015,pp.1-4].Currently, in the LTE-Advanced protocol specification of one of 4G standard
It at most only include small scale MIMO (small-scale MIMO) [document 5:Z.Lu, J.Ning, Y.Zhang, T. of 8 antennas
Xie,and W.Shen,“Richardson method based linear precoding with low complexity
For massive MIMO systems, " in Proc.of IEEE 81st VTC Spring, 2015, pp.1-4], and
Massive MIMO (also referred to as large-scale MIMO) technical requirements base station end includes 64 even up to a hundred antennas, is considered
Be most be hopeful in following 5G communication with the technology of prospect [see document 6-7:E.Larsson, O.Edfors, F.Tufvesson,
and T.Marzetta,“Massive MIMO for next generation wireless systems,”IEEE
Commun.Mag.,vol.52,no.2,pp.186-195,Feb.2014；Y. Wang and Y.Dong,"A genetic
antenna selection algorithm for massive MIMO systems with channel estimation
Error,”in Proc.of Advances in Wireless and Optical Commun.(RTUWO),2015,pp.1–
4].In multiuser MIMO especially Massive mimo system, when antenna for base station quantity is enough, different user to base station
Channel approximation mutually orthogonal [see document 6:Y.Wang and Y.Dong, " A genetic antenna selection
algorithm for massive MIMO systems with channel estimation Error,”in Proc.of
Advances in Wireless and Optical Commun. (RTUWO), 2015, pp.1-4], small scale channel fading and
Uncorrelated noise can also be eliminated [see document 8:M.Benmimoune, E.Driouch, W.Ajib, and D.Massicotte,
“Joint transmit antenna selection and user scheduling for massive MIMO
systems,"in Proc.of IEEE WCNC,2015,pp.381-386]；If, can be with using suitable precoding technique
Completely eliminate the interference of information between different user.However, needing deployment and antenna in design for the antenna for making full use of base station
The equal AFE module of quantity, this is not applicable in actual design.Due to base station AFE(analog front end) (Analog Front-End,
AFE) module number is limited, uplink can simultaneous selection antenna amount and the user that is serviced all suffer restraints.
For diversity, the spatial multiplexing gain for making full use of big aerial array, radio-frequency front-end resource is saved, day line options are to close very much
A kind of suitable selection [is shown in document 8:M.Benmimoune, E.Driouch, W.Ajib, and D. Massicotte, " Joint
transmit antenna selection and user scheduling for massive MIMO systems,”in
Proc.of IEEE WCNC,2015,pp.381-386].In the limited situation of AFE resource, day line options can be selected " most
One group of antenna well " is communicated, such as selects channel conditions preferably or the highest one group of antenna of output signal-to-noise ratio.Study people
There are many research achievements in terms of antenna selection by member.In the uplink, [see document 9:Y.Gao, W.Jiang, and
T.Kaiser,“Bidirectional branch and bound based antenna selection in massive
MIMO systems, " in Proc.of IEEE 26th PIMRC, 2015, pp.563-568] it proposes based on finding matrix
The dicomponent of maximum minimum singular value (Minimum Singular Value, MSV) delimits (Bidirectional
Branch and Bound, BBAB) algorithm realization day line options；Since there are monotonicity, this day line options energy for singular values of a matrix
Globally optimal solution is enough searched for, and more much lower than the complexity of the method for exhaustion.In channel there are in the case where error, [document 7] is utilized
Gene hepatitis B vaccine can realize the day line options based on water-filling；Low signal-to-noise ratio (Signal-to-Noise Ratio,
When SNR), which is able to achieve better channel capacity than traditional algorithm.[document 10:S.E.El-Khamy, K.H.
Moussa,A.A.El-Sherif,“On the performance of massive multiuser MIMO with
different transmit beamforming techniques and antenna selection,”in Proc.of
2015 1st URSI Atlantic Radio Science Conference (URSI AT-RASC), 2015, pp.1-10] it mentions
Go out based on the maximum Antenna Selection Algorithem of two norm of subscriber channel vector, which can be realized in the case of wave beam forming
(Bit Error Rate, BER) performance boost.
On the other hand, due to the constraint of AFE module, most multipotency while the number of users of service when base station sends and receives signal
Amount is no more than AFE resource module quantity, and the orthogonality of user also will affect multiuser MIMO system performance.Therefore user's tune
Degree is also extremely important in system performance realization.It is assessed based on user distribution formula, [document 11:X.Xie and X.Zhang,
“Scalable user selection for MU-MIMO networks,”in Proc.of IEEE INFOCOM,2014,
Pp.808-816] user's dispatching algorithm based on competitive channel feedback is proposed, when which can effectively save CSI acquisition
Between.Using zero-forcing beamforming (Zero-Forcing Beamforming, ZFBF) precoding, [document 1] is based on winding capacity
Constraint proposes the day line options and user's dispatching algorithm of three kinds of searching locally optimal solutions；In existing local antenna set and user
On the basis of set, locally optimal solution is jumped out by exchange antenna and user element, close to globally optimal solution.It is complicated to reduce
Degree, [document 8] are proposed in the downlink based on the day line options of user's subspace orthogonality and user's scheduling, the algorithm energy
Enough optimal performances [seeing document 8] for realizing the approximate method of exhaustion.
In addition, [document 12:Y.Cao and V.Kariwala, " Bidirectional branch and bound for
controlled variable selection:Part I.Principles and minimum singular value
Criterion, " Computers&Chemical Engineering, vol.32, no.10, pp.2306-2319,2008] and
[document 13:T.Yoo and A.Goldsmith, " On the optimality of multiantenna broad-cast
scheduling using zero-forcing beamforming,”IEEE J.Select.Areas Commun.,
Vol.24, no.3, pp.528-541, Mar.2006.] in also proposed day line options and user's dispatching algorithm.
But various algorithms should realize the performances such as standard deviation of superior system velocity, rate, again reduce to calculate
Complexity is currently to be difficult to realize.
Summary of the invention
To solve the above problems, the invention proposes a kind of selection of multiuser MIMO uplink antenna and user's dispatching parties
Method (is named as Joint Antenna Selection and User Scheduling, JASUS), has preferable part most
Excellent rate capability, but the complexity of algorithm is lower.
2, for this purpose, a kind of multiuser MIMO uplink antenna selection of the invention and user scheduling method include following step
Rapid: S1, the downward branch-bound algorithm based on semi-orthogonal user selection: finding keeps the minimum singular value of channel matrix maximum
The locally optimal solution of antenna set A and user's set U；S2, the local iteration's optimizing exchanged based on antenna with user's set element
Algorithm: in the case where having obtained A and U, fixed set A, U, exchanges the user element chosen and antenna element respectively；If
It was found that set A, U after commutative element can make channel speed bigger, then current locally optimal solution is jumped out to another performance more
Excellent locally optimal solution；S3, step S2 is repeated until antenna set A and user's set U are no longer changed.
Downward branch-and-bound (the Downwards that the above method proposed by the present invention is selected based on user's semi-orthogonal
Branch and Bound, DBAB), local iteration optimizing algorithm search channel matrix maximum MSV, realize joint antenna selection and
User's scheduling；For the algorithm using channel capacity as evaluation index, the rate capability of local optimum is more preferable, approaches exhaust algorithm
(Brute-Force Search,BFS)；Meanwhile the computation complexity of the algorithm is far below the method for exhaustion, has larger using valence
Value.Therefore the present invention is a kind of promising day line options and user scheduling method in multi-user MIMO system.
Detailed description of the invention
Fig. 1 is the downward branch-bound algorithm schematic diagram of the embodiment of the present invention.
Fig. 2 is local iteration of embodiment of the present invention optimizing algorithm schematic diagram.
Fig. 3 is downward branch-bound algorithm (DBAB) flow chart that the embodiment of the present invention is selected based on semi-orthogonal user.
Fig. 4 is local iteration of embodiment of the present invention optimizing algorithm flow chart.
Fig. 5 be algorithms of different of the embodiment of the present invention uplink and rate with signal-to-noise ratio variation schematic diagram.
Fig. 6 is the variation schematic diagram of algorithms of different of embodiment of the present invention uplink and speed standard difference with signal-to-noise ratio.
Fig. 7 is algorithms of different of the embodiment of the present invention and rate with AFE(analog front end) quantity variation schematic diagram.
Fig. 8 is the mean iterative number of time of algorithms of different of the embodiment of the present invention with the variation schematic diagram of AFE(analog front end).
Specific embodiment
As previously mentioned, the antenna cost of base station is very low in multiuser MIMO especially Massive MIMO communication system,
Aerial array can be constituted with large scale deployment；But AFE(analog front end) resource is relatively expensive, and all disposing an AFE for every antenna can be big
Amount increases cost, and will cause the wasting of resources.Considering the limited practical application scene of AFE module, proposition of the embodiment of the present invention
A kind of day line selection for the search maximum MSV that the downward branch-and-bound based on user's semi-orthogonal, local iteration's optimizing combine
It selects and user scheduling method, does not obviously increase complexity in the case where improving performance.Below from algorithmic descriptions, method and step
It is illustrated with the several aspects of performance evaluation.
One, algorithmic descriptions
1.1 system models and problem rate
1.1.1 system model
Assuming that in multiuser MIMO cellular cell, base station deployment M root antenna and N number of AFE module, and M > > N.Meanwhile
Assuming that there are k single-antenna subscriber etc. is to be serviced in the cell, k > N.Because the quantity of base station AFE module limits, base station is each
The N root antenna that at most channel quality can only be selected best from M root antenna sends and receives signals, and can only at most service N simultaneously
A user.If considering the up channel of multiuser MIMO, antenna for base station received signal meets following formula:
Wherein, y is received aerial signal vector, andUser to base station it is quasi-static solely
Vertical same distribution Rayleigh channel matrix；It is the emission signal vector that k user is uploaded to base station,It is M day
Line receives the additive white Gaussian noise (Additive White Gaussian Noise, AWGN) being superimposed on signal, and each point
Amount all obey mean value be 0, the multiple Gauss random distribution that variance is 1, i.e. zi~CN (0,1).
In base station receiving end, optimum reception can be achieved with using linear receiver such as squeeze theorem etc..Wherein,It is the receiving matrix of multi-user MIMO system, and W=(HHH)-1HH/||(HHH)-1HH||2, wherein | | | |2For
Two norms, ()HIt is the conjugate transposition of matrix.
It is as follows that the user that then receiver detects sends signal:
Wherein, τ=| | (HHH)-1HH||2.Then the SNR of user k meets [9]:
Wherein, λmin(H) minimum singular value of matrix H is represented.
1.1.2 problem describes
If the uplink transmission power P of single user is certain, the quantity N of the AFE module of base station deployment is certain, then base station is each
N root antenna can only be at most utilized, constitutes antenna set A, and system at most can provide communication service simultaneously for N user,
Constitute user's set U；The channel matrix then finally actually used are as follows:
HA,U=H (A, U) (4)
Because communication process is random process, therefore system is maximum ergodic and rate are as follows:
In formula, E { } is mathematic expectaion, and antenna set and user's set need to meet constraint condition:
|A|≤N
|U|≤N (6)
In formula, | | it is cardinal of the set.By formula (4), (5), (6) it is found that when communication system rate maximum, user gathers
The element number of U and antenna set A is N.According to formula (3) and (5) it is found that make the rate of system maximum, then need to meet
Following condition:
λmin(HA,U)≥λmin(H′)
H∈{Hs|Hs∈CN×N,Hs=H(N,N)} (7)
The channel matrix H finally selected based on antenna set A and user's set UA,UIn all N of H × N-dimensional submatrix
H(N,N)In have maximum MSV [9].
For such NP-hard problem, the method for exhaustion can search out globally optimal solution, but computation complexity is high.It is full
Sufficient practical application, the present invention propose day line options and user scheduling method (the Largest MSV based-based on maximum MSV
JASUS, LMSV-JASUS) find locally optimal solution.The algorithm and the selection of existing joint antenna and user scheduling method
(JASUS) algorithm complexity is suitable, by by (Throughput and Complexity Balanced JASUS, TCB-
JASUS) algorithm commutative element can skip the thinking of current local optimum, and performance is made to have further promotion, therefore be a kind of close
The algorithm of globally optimal solution.
The JASUS algorithm of 1.2 search maximum MSV
It is analyzed based on the problem of 1.1 section, this section will be described in detail based on the downward branch-and-bound of user's semi-orthogonal, part
The LMSV-JASUS algorithm that iteration optimizing combines.The algorithm is largely divided into two parts:
1) the downward branch-bound algorithm based on semi-orthogonal user selection
By the algorithm can find as far as possible system performance close to the method for exhaustion antenna set A and user's set U；
2) the local iteration's optimizing algorithm exchanged based on antenna with user's set element
In the case where having obtained A and U, fixed set A, U, exchanges the user element chosen and antenna element respectively.
If it find that set A, U after commutative element can make channel speed bigger, then current locally optimal solution is jumped out to another individual character
It can more preferably locally optimal solution.Because globally optimal solution is the maximum value in locally optimal solution, therefore this method has more maximum probability
Obtain globally optimal solution.
1.2.1 the downward branch-bound algorithm based on semi-orthogonal
1) downward branch-bound algorithm
Downward branch-bound algorithm is a kind of unsupervised classification algorithm, may search for the overall situation when meeting downward monotonicity most
It is excellent.Its category theory is as follows:
Target: assuming that there is set X={ 1,2,3,4,5 } includes 5 elements, it is now desired to select 2 from this collection
The set X that a element is constituted2, meet objective function J (X):
Downward monotonicity: set X if it existsn, whenWhen, always have:
J(Xn)≥J(Xm)(9)
Objective function J (X) is then claimed to meet downward monotonicity.If set XnMeet:
Then set XSPlace branch is the optimal direction of search.
Global optimum: it is based on the downward monotonicity of objective function J (X), by successive ignition, which can be obtained entirely
Office optimal solution X2。
As shown in Figure 1, branch-bound algorithm is divided into search branch and boundary determines two processes.Steps are as follows:
A) root node is original collection X={ 1,2,3,4,5 }, includes 5 elements.
B) branch is searched for: set X size is n, can be divided into n search branch, and each branch, which represents, removes it from set X
In an element i, constitute set X-i, i=1 ..., n；In the first round, set X has 5 elements, therefore is segmented into 5
Branch；Number in branch is represented removes the element from set X, therefore has 5 branches.
C) boundary determines: according to b) principle, the set X that each branch is generated-iIt brings objective function into, calculates J (X-i)
Value；Find out J (X-i) in make the maximum set X of function valueS(meeting formula (10)).In the first round, after removing element 3
Set Xs={ 1,2,4,5 } are the optimal directions of search.
It d) will set XSIt is replaced by X, whether the element number for calculating the set meets the requirements.If it is greater than 2, go to
Step b) is executed；If being equal to 2, just terminate search process.
Based on a)~d), it is searched for by three-wheel, each wheel successively removes element 3,2,5；Finally obtain globally optimal solution X2=
{1,4}.The algorithm needs to search for altogether (M-N) wheel, and total computation complexity is not high.
2) the downward branch-bound algorithm based on semi-orthogonal user selection
If J (Xn) meet dullness downwards, globally optimal solution can be searched out according to branch-bound algorithm principle.In [9]
In, when fixed by service user, only carry out day line options；The maximum MSV of channel matrix H is J (A)=λ at this timemin(H(A,:))
Meet downwards dull [9,12], globally optimal solution can be obtained.
In the present invention, when number of users is greater than AFE quantity (k > N), it is necessary to while antenna and user's selection are carried out,
J (A, U)=λ at this timemin(H(A,U)) just not exclusively meet dullness downwards, therefore globally optimal solution cannot be obtained.
Based on the above analysis, LMSV-JASUS algorithm of the invention is just proposed using the principle for finding local maxima MSV
Realize that the downward branch-bound algorithm of JASUS, the algorithm are true using antenna subset branch, semi-orthogonal user selection, the direction of search
Fixed three steps carry out day line options.To select most suitable N number of user from k user, the present invention is utilized in [13] and is proposed
The maximum user's semi-orthogonal of the best and orthogonal amplitude based on user's semi-orthogonal select (Semi-orthogonal User
Selection, SUS) algorithm.Therefore each node of Fig. 1 can first with the most suitable user's set U of SUS algorithms selection, then
It solves based on currently by each antenna branch of service user's set UDetermine optimal search
DirectionRemoving influences maximum single antenna m to system performance.It is secondary by (M-N)
Iteration can obtain local optimum set A, U under this condition.
1.2.2 local iteration's optimizing algorithm
Under normal circumstances, 1.2.1 algorithm solve antenna and user's set A, U can all fall into local optimum rather than the overall situation most
It is excellent, and the fluctuation of locally optimal solution is very big.For the fluctuation for reducing locally optimal solution, the present invention is based on the parts of [1] design
Iteration optimizing algorithm can jump out current locally optimal solution and find the better locally optimal solution of performance, increase and obtain globally optimal solution
Possibility.Its concrete thought is as shown in Figure 2:
Assuming that each point in plane gathers (A by an antenna, user there are two-dimensional surface spacei,Uj) indicateTarget is exactly that find optimal point allows system and rate maximum.
A) antenna and user's set A, U that local optimum is obtained according to the downward branch-bound algorithm of 1.2.1, are set to
Initial point P (A1, U1).
B) fixed antenna set A1 finds user's set U2 that antenna set determines using SUS algorithm；Gather (A1, U2)
Constitute point q.If q point calculate acquisition system and rate be greater than p point and rate, update optimal solution (A, U)=
(A1,U2)。
C) fixed user's set U2, optimal antenna set A2 is searched out using DBAB algorithm, constitutes point R.If in R
The system and rate of point are greater than system and rate in q point, then updating optimal solution is (A, U)=(A2, U2).
D) step c), d) is repeated, until system and rate are not further added by.Solution set (A, U) at this time is exactly final output
Antenna set and user set.
Two, the specific implementation step of each algorithm
In 1.2 trifles, structure composition and realization principle of the detailed analysis of the present invention based on LMSV-JASUS algorithm.Its
Mainly it is made of downward branch-bound algorithm, semi-orthogonal user selection algorithm and local iteration's optimizing algorithm.The tool of each algorithm
Shown in body realizes that steps are as follows.
2.1 downward branch-bound algorithms (DBAB)
Downward branch-bound algorithm of the invention determines three by antenna subset branch, semi-orthogonal user selection, the direction of search
A step is constituted, according to the analysis of 1.2.1 trifle it is found that the algorithm can obtain the antenna set A of local optimum and user gathers
U.Specific steps process is as shown in Figure 3:
A) base station obtains channel matrix H, obtains AFE module number N, antenna amount M and the number of users k of system；Initially
Change the wheel number t of antenna set A to be selected and search circulation:
A=1 ..., M } (11)
T=1 (12)
B) for any antenna element i ∈ A, set A is generated after removing antenna i in antenna set A-i, and being based on should
Antenna set obtains current channel matrix H-i；
C) it calls semi-orthogonal user to select (SUS) algorithm (2.2 detailed analysis), transmits respective channels matrix H-iWith single energy
Number of users that may be served N generates corresponding user's set U-i；
D) according to A-i、U-iCalculate the minimum singular value of corresponding channel matrixIt finds out and makes minimum unusual
It is worth maximum antenna set and user's set A-i、U-i, instruction satisfaction:
A←{i∈A,i≠m} (14)
U=U-m (15)
If e) | A | > N, t ← t+1 repeat b)~d)；If | A |=N stops iterative process, exports local optimum
Antenna set and user's set A, U.
2.2 semi-orthogonal users select (SUS) [8]
Semi-orthogonal user's selection algorithm be the user U that selects N number of orthogonality best in the case where antenna set A is fixed into
Row service, the algorithm are taken turns search by N and are completed.Every wheel selects one from user to be selected set and has selected family collection orthogonality
Best user is added by service user's set, until including N number of element by service user's set.Input parameter is CSI matrix H
=[h1,…,hk] and the number of users N that can be serviced simultaneously.Implementation step is as follows:
A) parameter initialization:
According to CSI information, total number of users is k in cell.Initialize user's set T to be selected1With the i-th selected user:
T1=1 ..., k } (16)
I=1 (17)
Being selected user's set U of service is empty set:
B) for each user k ∈ Ti, it is based on channel vector hkCalculate itself and extending space { g(1),…,g(i-1)It is orthogonal
Component gk:
As i=1, gk=hk.Wherein, orthogonal vectors base g(j)To selected user j=1 ..., the efficient orthogonal of (i-1)
Component.
C) finding i-th, most preferably by service user π (i), (symbol refers to that the i-th wheel collects T from user to be selectediThe condition of middle selection
The best number by service user):
U←U∪π(i) (21)
g(i)=gπ(i) (22)
If d) | U | < N, i ← i+1 update antenna set T to be selectedi+1Are as follows:
Ti+1={ k ∈ Ti,k≠π(i)} (23)
Repeat step b)~d)；Otherwise, it goes to e)；
E) returning to space size is N by selection user's set U.
The optimizing of 2.3 local iterations solves
By 2.1 and 2.2 constitute the JASUS based on user's semi-orthogonal can obtain one group of locally optimal solution (A,
U).The process of the key step of the local iteration's optimizing algorithm carried out on this basis is as shown in Figure 4:
F) antenna and user's set A, U that local optimum is obtained according to 2.1 downward branch-bound algorithms, are set to just
Initial point (A, U).
G) current optimal antenna and user are saved and gathers (A_t, U_t)=(A, U).
H) (i=1,2 ... be iteration wheel number) the fixed antenna set A in the i-th wheel, optimal use is solved using SUS algorithm
Family set Ui.If point (A, U in plane at this timei) make system velocity R (A, Ui) > R (A, U) meets, then updates user's set U
=Ui, system and rate R (A, U)=R (A, Ui)；Otherwise, U and R (A, U) are not updated.
I) fixed user's set U can solve current optimal antenna set using downward branch-bound algorithm (not including SUS)
Close Ai.If point (the A in plane at this timei, U) and make system velocity R (Ai, U) and > R (A, U), then update antenna set A=Ai, be
Unite and rate R (A, U)=R (Ai,U)；Otherwise, not more fresh target solution A and R (A, U).
J) whether identical compare the optimal solution (A, U) updated after optimal solution (A_t, U_t) and iteration.If it is different, again
Execute c)~e)；If identical, stop iterative process, executes method k).
K) optimal solution (A, U) for obtaining the algorithm, solve the system using formula (4), (5) and rate R.
Three, performance evaluation
The evaluation of 3.1 numerical results
In this trifle, by LMSV-JASUS algorithm proposed by the present invention and JASUS [8], TCB-SUS [1], the method for exhaustion
(BFS) system performance of scheduling algorithm is compared, and analyzes the superiority and inferiority of various methods.In emulation experiment, due to the meter of the method for exhaustion
Calculation complexity is too high, and when being related to compared with method of exhaustion performance, parameter setting is all smaller.
Fig. 5 be different joint antenna selection and user's dispatching algorithm Mean Speed with SNR comparison diagram.Because wireless
Channel is random process, therefore the rate and standard deviation of taking system in testing are as measurement index；And because the height of the method for exhaustion is multiple
Miscellaneous degree, therefore reduced parameter.Assuming that in nest bee system single subdistrict, 10 antennas of base station deployment and 5 AFE modules share 10
A user needs to be serviced.Due to AFE limited amount, can only at most there are 5 users to be serviced every time in uplink communication.Therefore base
Five antennas can at most be selected every time while service 5 users by standing.Assuming that system is ergodic, the equal of 50 subrates is solved in text
It is worth the expectation as system velocity.When the SNR of system progressively increases to 30dB by 0dB, LMSV-JASUS algorithm is than existing
JASUS, TCB-SUS have better rate it is expected, and LMSV-JASUS algorithm solves in patent system and rate approach BFS's
Optimal value.
What the standard deviation that Fig. 6 is different the average and rate of joint antenna selection and user's dispatching algorithm changed with SNR
Comparison diagram.Its major parameter is consistent with the setting of Fig. 5.With the increase of SNR, the LMSV- minimum with speed standard difference of the method for exhaustion
JASUS algorithm approaches the method for exhaustion with speed standard difference, has lower value than existing JASUS, TCB-SUS, system and fast
Rate is relatively more stable.
Fig. 7 is that different joint antennas select and the rate of user's dispatching algorithm is with AFE change curve.Assuming that in cellular system
In single subdistrict, base station deployment has 64 antennas, which shares user 25, and the SNR of user transmitting terminal is 10dB.Because it
Line number amount and number of users are more, and the rate of BFS does not just compare again.When the quantity of AFE progressively increases to 12 by 4,
The TCB-JASUS for only relying on element exchange loses the upper hand quickly, falls into locally optimal solution；JASUS and LMSV-JASUS algorithm
Rate can continue increasing, and algorithm ratio JASUS performance of the invention is more preferably.
Fig. 8 is that different antennae selects and the mean iterative number of time of user's dispatching algorithm is with the increased change curve of AFE.
JASUS does not have iterative process, therefore not drawn in the figure.It can be found that with the increase of AFE quantity, the iteration of TCB-JASUS
Number is about 2, and the number of iterations for local iteration's optimizing algorithm that article proposes is always slightly less than 1, and relative constant.This explanation exists
In most cases, the result of the downward branch-bound algorithm based on the selection of user's semi-orthogonal is all can be by iteration again
It optimizes, but being obviously improved for performance can be achieved with by an iteration.
Based on the analysis to Fig. 5-8 it is found that the system velocity of realization of LMSV-JASUS algorithm, the standard deviation of rate all compare
The performance of more excellent, the closer method of exhaustion of TCB-JASUS, JASUS.Existing downward branch-bound algorithm realizes the solution of JASUS all
Local optimum can be fallen into, the performance for only relying on the iteration optimizing algorithm of element exchange is easy to lose work when selected set increases
With.Method proposed by the present invention had not only given full play to the advantage of two kinds of algorithms, but also not will increase the number of iterations.Therefore it is proposed by the present invention
The LMSV-JASUS algorithm that downward branch-and-bound, local iteration's optimizing based on the selection of user's semi-orthogonal combine is that have very much
Meaning.
3.2 analysis of complexity
The computation complexity of LMSV-JASUS algorithm proposed by the present invention mainly from based on user's semi-orthogonal select to
The computation complexity of inferior division key-machine and local iteration's optimizing algorithm is constituted.
Based on 1.2.1 section analysis it is found that the computation complexity based on downward branch-bound algorithm is mainlyAnd the computation complexity of objective function J (x) include orthogonal users set selection algorithm SUS, really
The complexity of SVD after determining user's set, i.e.,Therefore downward branch-bound algorithm
Complexity be
Analysis is saved based on 1.2.2 it is found that wheel local iteration's optimizing includes individual SUS and downward branch-bound algorithm.
So when computation complexity be
Therefore the computation complexity of LMSV-JASUS algorithm is aboutT is iteration wheel number.
Since the mean iterative number of time of actual emulation is lower than once, total average computation complexity is less than
So its computation complexity does not obviously increase compared with the JASUS algorithm in [8].
Claims (9)
1. a kind of multiuser MIMO uplink antenna selection and user scheduling method, characterized by the following steps:
S1, the downward branch-bound algorithm based on semi-orthogonal user selection: finding keeps the minimum singular value of channel matrix maximum
Antenna set A and user's set U locally optimal solution；
S2, optimal solution is updated based on local iteration's optimizing algorithm that antenna is exchanged with user's set element, comprising:
The antenna and user's set A, U of S2a, the local optimum obtained according to the downward branch-bound algorithm of step S1, are arranged
For initial point P (A1, U1)；
S2b, fixed antenna set A1 find user's set U2 using SUS algorithm based on determining antenna set A1；Set (A1,
U2) constitute point q, if q point calculate acquisition system and rate greater than p point and rate, update optimal solution (A, U)=
(A1,U2)；
S2c, fixed user's set U2, optimal antenna set A2 is searched out using DBAB algorithm, constitutes point R；If in R point
System and rate are greater than system and rate in q point, then updating optimal solution is (A, U)=(A2, U2)；
S3, step S2 is repeated until antenna set A and user's set U are no longer changed.
2. multiuser MIMO uplink antenna selection as described in claim 1 and user scheduling method, it is characterised in that: step
In rapid S1, determines to inferior division, reselection semi-orthogonal user set, last border using elder generation and carry out day line options.
3. multiuser MIMO uplink antenna selection as claimed in claim 2 and user scheduling method, it is characterised in that institute
It is specific as follows to state step S1: first with the most suitable user's set U of each antenna branch of SUS algorithms selection, then solving each antenna point
J (the A of branch-i, U) and=λmin(H(A-i,U)), determine optimal direction of search A-i, by (M-N) secondary iteration, current letter can be obtained
Local optimum set A, U under road matrix.
4. multiuser MIMO uplink antenna selection as described in claim 1 and user scheduling method, it is characterised in that institute
It is more specifically as follows to state step S1:
S1a, base station obtain channel matrix H, obtain AFE module number N, antenna amount M and the number of users k of system；Initialization
The wheel number t of antenna set A to be selected and search circulation:
A=1 ..., M }
T=1
S1b, for any antenna element i ∈ A, set A is generated after removing antenna i in antenna set A-i, and it is based on the day
Line set obtains current channel matrix H-i；
S1c, it calls semi-orthogonal user to select (SUS) algorithm, transmits respective channels matrix H-iWith single energy number of users that may be served
N generates corresponding user's set U-i；
S1d, according to A-i、U-iCalculate the minimum singular value of corresponding channel matrixIt finds out and allows minimum singular value
Maximum antenna set and user's set A-i、U-i, instruction satisfaction:
A←{i∈A,i≠m}
U=U-m
If S1e, | A | > N, t ← t+1 repeat S1b~S1d；If | A |=N stops iterative process, exports local optimum
Antenna set and user's set A, U.
5. multiuser MIMO uplink antenna selection as described in claim 1 and user scheduling method, it is characterized in that described
Step S3 are as follows: S2d, repeat step S2b, S2c, until system and rate are not further added by, solution set at this time is exactly final output
Antenna set and user set.
6. multiuser MIMO uplink antenna selection as claimed in claim 2 and user scheduling method, it is characterized in that step
Downward branch-bound algorithm used in downward branching step in S1 includes antenna subset branch, semi-orthogonal user selection, searches
Suo Fangxiang determines three steps.
7. multiuser MIMO uplink antenna selection as claimed in claim 6 and user scheduling method, it is characterized in that: described
Antenna subset branch includes the following steps:
A) base station obtains channel matrix H, obtains AFE module number N, antenna amount M and the number of users k of system；Initialization to
Select the wheel number t of antenna set A and search circulation:
A=1 ..., M }
T=1
B) for any antenna element i ∈ A, set A is generated after removing antenna i in antenna set A-i, and it is based on the antenna
Set obtains current channel matrix H-i；
C) it calls semi-orthogonal user to select (SUS) algorithm, transmits respective channels matrix H-iWith single energy number of users that may be served N,
Generate corresponding user's set U-i；
D) according to A-i、U-iCalculate the minimum singular value of corresponding channel matrixFind out allow minimum singular value most
Big antenna set and user's set A-i、U-i, instruction satisfaction:
A←{i∈A,i≠m}
U=U-m
If | A | > N, t ← t+1 repeat b)~d)；If | A |=N stops iterative process, exports the antenna set of local optimum
It closes and user's set A, U.
8. multiuser MIMO uplink antenna selection as claimed in claim 6 and user scheduling method,
It is characterized in that the semi-orthogonal user selection includes:
Parameter initialization: according to CSI information, total number of users is K in cell, initializes user's set T to be selected1It is selected with i-th
User:
T1=1 ..., K }
I=1
Being selected user's set U of service is empty set:
For each user k ∈ Ti, it is based on channel vector hkCalculate itself and extending space { g(1),…,g(i-1)Quadrature component gk:
As i=1, gk=hk, wherein orthogonal vectors base g(j)To selected user j=1 ..., the efficient orthogonal component of (i-1)；
I-th is found most preferably by service user π (i), which refers to that the i-th wheel collects T from user to be selectediThe condition of middle selection is most
The good number by service user:
U←U∪π(i)
g(i)=gπ(i)
If | U | < N, i ← i+1 update antenna set T to be selectedi+1Are as follows:
Ti+1={ k ∈ Ti,k≠π(i)}
Returning to space size is N by selection user's set U.
9. a kind of multiuser MIMO uplink antenna selection and user scheduling method, it is characterized in that including:
A) according to downward branch-bound algorithm obtain local optimum antenna and user's set A, U, be set to initial point (A,
U)；
B) current optimal antenna and user are saved and gathers (A_t, U_t)=(A, U)；
C) the fixed antenna set A in the i-th wheel, optimal user's set U is solved using SUS algorithmi, i=1,2 ... be iteration wheel
Number；If point (A, U in plane at this timei) make system velocity R (A, Ui) > R (A, U) meets, then updates user's set U=Ui, be
Unite and rate R (A, U)=R (A, Ui)；Otherwise, U and R (A, U) are not updated；
D) the fixed user set U updated by step c), can solve current local optimum using downward branch-bound algorithm
Antenna set AiIf point (the A in plane at this timei, U) and make system velocity R (Ai, U) and > R (A, U), then update antenna set A=
Ai, system and rate R (A, U)=R (Ai,U)；Otherwise, not more fresh target solution A and R (A, U)；
E) whether identical compare the optimal solution (A, U) updated after optimal solution (A_t, U_t) and iteration；If it is different, executing again
C)~e)；If identical, stop iterative process, export last antenna set A and user's set U.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610879164.5A CN106230493B (en) | 2016-09-30 | 2016-09-30 | A kind of selection of multiuser MIMO uplink antenna and user scheduling method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610879164.5A CN106230493B (en) | 2016-09-30 | 2016-09-30 | A kind of selection of multiuser MIMO uplink antenna and user scheduling method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN106230493A CN106230493A (en) | 2016-12-14 |
| CN106230493B true CN106230493B (en) | 2019-08-06 |
Family
ID=58076296
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610879164.5A Active CN106230493B (en) | 2016-09-30 | 2016-09-30 | A kind of selection of multiuser MIMO uplink antenna and user scheduling method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106230493B (en) |
Families Citing this family (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106603134B (en) * | 2016-12-22 | 2020-10-27 | 东南大学 | Distributed antenna selection design method for bidirectional wireless communication system |
| CN107154815B (en) * | 2017-04-05 | 2020-07-03 | 西安邮电大学 | Multi-user system hybrid pre-coding method |
| CN115460616A (en) * | 2022-06-17 | 2022-12-09 | 东南大学 | Millimeter wave large-scale MIMO system user grouping method based on genetic algorithm |
Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103763782A (en) * | 2014-01-13 | 2014-04-30 | 西安电子科技大学 | Dispatching method for MU-MIMO down link based on fairness related to weighting users |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US9226092B2 (en) * | 2012-08-08 | 2015-12-29 | Golba Llc | Method and system for a distributed configurable transceiver architecture and implementation |
-
2016
- 2016-09-30 CN CN201610879164.5A patent/CN106230493B/en active Active
Patent Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103763782A (en) * | 2014-01-13 | 2014-04-30 | 西安电子科技大学 | Dispatching method for MU-MIMO down link based on fairness related to weighting users |
Non-Patent Citations (2)
| Title |
|---|
| 《Joint transmit antenna selection and user scheduling for Massive MIMO systems》;Mouncef Benmimoune etc.;《 2015 IEEE Wireless Communications and Networking Conference (WCNC)》;20150618;第563-568页 |
| 《Joint user scheduling and antenna selection in distributed massive MIMO systems with limited backhaul capacity》;Guozhen Xu etc.;《China Communications》;20140819;第11卷(第5期);第17-30页 |
Also Published As
| Publication number | Publication date |
|---|---|
| CN106230493A (en) | 2016-12-14 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Parfait et al. | Performance analysis and comparison of ZF and MRT based downlink massive MIMO systems | |
| JP5875344B2 (en) | Method for reducing interference in a wireless network | |
| CN103166685B (en) | A kind of interference alignment schemes based on joint Power distribution in LTE | |
| CN110299937B (en) | Beam forming method for uplink MIMO-NOMA wireless communication system | |
| CN107086886B (en) | Double-layer precoding design for large-scale MIMO system fusion zero forcing and Taylor series expansion | |
| CN110492915A (en) | A kind of power distribution method based on the short packet transmission of MIMO-NOMA | |
| WO2010026472A1 (en) | Beamforming systems and method | |
| CN106793108B (en) | Federated user selection and power distribution optimization method in mimo system downlink | |
| CN102882570B (en) | Optimum transceiving combined processing method for communication among equipment in mobile communication network | |
| CN105681009B (en) | Pilot tone optimization distribution joint method for precoding for multi-user's multiple-input and multiple-output | |
| CN106230493B (en) | A kind of selection of multiuser MIMO uplink antenna and user scheduling method | |
| Zhao et al. | Simultaneous wireless information and power transfer in interference alignment networks | |
| JP2009153139A (en) | Pre-coding processing method and apparatus for mimo downlink, and base station | |
| CN102624432A (en) | Multilayer beamforming method and terminal for realizing multilayer beamforming | |
| Dong et al. | Improved joint antenna selection and user scheduling for massive MIMO systems | |
| CN103384228B (en) | Continuous precoding and the user of a kind of multiuser MIMO broadcast channel select unified algorithm | |
| Hu et al. | Energy efficiency of massive MIMO wireless communication systems with antenna selection | |
| CN102684765B (en) | Method for forming distributed coordination beam on basis of maximum virtual signal-to-noise ratio | |
| CN104821840B (en) | A kind of anti-interference method of extensive multiple-input and multiple-output downlink system | |
| Ying et al. | Heterogeneous massive MIMO with small cells | |
| Song et al. | Downlink interference rejection in ultra dense network | |
| Li et al. | Optimal resource allocation for massive MIMO over spatially correlated fading channels | |
| Cheng et al. | User clustering and scheduling in UAV systems exploiting channel correlation | |
| CN103209496B (en) | Based on the user scheduling method of beam selection in multi-base station cooperative system | |
| Fodor et al. | Recent advances in acquiring channel state information in cellular mimo systems |
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 |