CN106712815A - Validity examination based CSI feedback overhead reduction method - Google Patents
Validity examination based CSI feedback overhead reduction method Download PDFInfo
- Publication number
- CN106712815A CN106712815A CN201710036013.8A CN201710036013A CN106712815A CN 106712815 A CN106712815 A CN 106712815A CN 201710036013 A CN201710036013 A CN 201710036013A CN 106712815 A CN106712815 A CN 106712815A
- Authority
- CN
- China
- Prior art keywords
- csi
- user
- time
- reduction method
- validity
- 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.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 41
- 230000009467 reduction Effects 0.000 title claims abstract description 16
- 230000005540 biological transmission Effects 0.000 claims abstract description 41
- 230000007246 mechanism Effects 0.000 claims description 42
- 238000012546 transfer Methods 0.000 claims description 25
- 239000011159 matrix material Substances 0.000 claims description 13
- 230000008569 process Effects 0.000 claims description 9
- YLKFDHTUAUWZPQ-UHFFFAOYSA-N N-Nitrosodi-n-propylamine Chemical compound CCCN(N=O)CCC YLKFDHTUAUWZPQ-UHFFFAOYSA-N 0.000 claims description 8
- 230000009471 action Effects 0.000 claims description 2
- 230000008713 feedback mechanism Effects 0.000 abstract description 5
- 238000012360 testing method Methods 0.000 description 7
- 238000012549 training Methods 0.000 description 7
- 230000008859 change Effects 0.000 description 5
- 238000009826 distribution Methods 0.000 description 5
- 108700026140 MAC combination Proteins 0.000 description 4
- 238000002474 experimental method Methods 0.000 description 4
- 230000000694 effects Effects 0.000 description 3
- 230000002776 aggregation Effects 0.000 description 2
- 238000004220 aggregation Methods 0.000 description 2
- 238000004891 communication Methods 0.000 description 2
- 238000001514 detection method Methods 0.000 description 2
- 230000001965 increasing effect Effects 0.000 description 2
- 238000007689 inspection Methods 0.000 description 2
- 238000006116 polymerization reaction Methods 0.000 description 2
- 230000018199 S phase Effects 0.000 description 1
- 238000004422 calculation algorithm Methods 0.000 description 1
- 238000012937 correction Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 230000000116 mitigating effect Effects 0.000 description 1
- 230000011664 signaling Effects 0.000 description 1
- 230000001360 synchronised effect Effects 0.000 description 1
- 238000010200 validation analysis Methods 0.000 description 1
- 238000012795 verification Methods 0.000 description 1
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
- H04B7/0613—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 simultaneous transmission
- H04B7/0615—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 simultaneous transmission of weighted versions of same signal
- H04B7/0619—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 simultaneous transmission of weighted versions of same signal using feedback from receiving side
- H04B7/0621—Feedback content
-
- 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
- H04B7/0613—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 simultaneous transmission
- H04B7/0615—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 simultaneous transmission of weighted versions of same signal
- H04B7/0619—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 simultaneous transmission of weighted versions of same signal using feedback from receiving side
- H04B7/0621—Feedback content
- H04B7/0623—Auxiliary parameters, e.g. power control [PCB] or not acknowledged commands [NACK], used as feedback information
-
- 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
- H04B7/0613—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 simultaneous transmission
- H04B7/0615—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 simultaneous transmission of weighted versions of same signal
- H04B7/0619—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 simultaneous transmission of weighted versions of same signal using feedback from receiving side
- H04B7/0621—Feedback content
- H04B7/0626—Channel coefficients, e.g. channel state information [CSI]
Landscapes
- Engineering & Computer Science (AREA)
- Computer Networks & Wireless Communication (AREA)
- Signal Processing (AREA)
- Mobile Radio Communication Systems (AREA)
Abstract
The invention discloses a validity examination based CSI feedback overhead reduction method. Aiming at the low-frequency problem of an existing CSI feedback mechanism, the validity examination based CSI feedback overhead reduction method increases the CSI utilization rate and reduces the CSI feedback overhead in an MU-MIMO network. The validity examination based CSI feedback overhead reduction method comprises the steps of introducing validity judgment of the CSI in uplink transmission and downlink transmission to verify the validity of the CSI reported last time; if the CSI reported last time is valid, distributing the CSI into a good group and feeding back an ACK frame to an AP; and if the CSI reported last time is invalid, recalculating a new CSI based on received NDP packets by a bad group user and feeding back the new CSI to the AP. In the uplink transmission, a group of the most orthogonal concurrent users are selected; and in the downlink transmission, the orthogonality of the concurrent users can be guaranteed by pre-coded signals.
Description
Technical field
The invention belongs to field of wireless communications networks, and in particular to a kind of CSI feedback expense drop checked based on the term of validity
Low method.
Background technology
In standard 802.11ac agreements, all user feedbacks of AP requirements its each CSI realizing selecting the use of " optimal "
Family group.In large scale network, the time loss of this process can easily exceed the real time for being used for data transfer very much.Cause
This, a mensurable user's selection mechanism for being called OPUS is arisen at the historic moment.OPUS at most only needs to M wheel CSI feedbacks, wherein M generations
The antenna number of Table A P.In each round, novelty can be used for user's evaluation and test and the detection mechanism quilt for having selected user's orthogonality
Using.It is one can select optimal user and carry out the competition dispersal mechanism of CSI feedback, and by taking OPUS as an example, Fig. 2 a are illustrated
The MAC operation related to CSI feedback of OPUS.
First, AP sends empty data assertion frame (NDPA), and the user of intentional data transmission is informed and to week with this
Enclose wlan network statement and retain channel.And then, AP sends the empty data frame (NDP) comprising training sequence.After NDP is received,
Each user estimates oneself and the CSI matrixes between transmitting terminal antenna.First user is immediately by the CSI feedback of oneself to AP.
Other intentional users participate in feedback contention mechanism after detection frame is received.The user for best suiting mechanism requirement obtains transmission route and incites somebody to action
The CSI feedback of oneself is to AP.User's order is transmitted by NDPA.
As shown in Figure 2 a, AP is total before every wheel data transfer needs 3 to take turns CSI feedback.Additionally, this expense can be with AP antennas
Quantity it is linearly increasing.The CSI feedback of these users is effective before every wheel data transferConsider some users in a wheel
Its CSI can't update after DTD, it means that AP spends the CSI that big cost is obtained not to be fully used.Therefore,
One direct solution is the expense for mitigating CSI feedback, and this by CSI utilization rates are at utmost improved to the greatest extent and eliminate unnecessary
CSI feedback operation.
The content of the invention
In order to solve the problems of the prior art, the present invention proposes a kind of CSI feedback expense drop checked based on the term of validity
Low method, for the low efficiency problem that existing CSI feedback mechanism is present, improves CSI utilization rates in MU-MIMO networks, reduces
CSI feedback expense.
In order to realize the above object the technical solution adopted in the present invention is：Comprise the following steps：
1) authenticating user identification and CML are initialized；
2) AP broadcast NDPA and NDP packets, state that this data transfer is transmitted or downlink transmission for up-link,
If up-link is transmitted, step 3 is gone to), if downlink transmission, go to step 4)；
3) up-link transmission：
3.1) intention of its uplink is stated in AP statements by NDPA and NDP bags；
3.2) user verifies that last time reports the validity of CSI after NDP bags are received, if last time reports CSI effectively, distributes
Enter good groups；If it is invalid that last time reports CSI, bad groups are designated as, bad groups user recalculates new according to the NDP bags for receiving
CSI；
3.3) in subsequent each round, the user for feeding back power is obtained by the CSI feedback of oneself to AP, if being located at good groups
It is interior, only to AP feeding back ACK frames；
3.4) AP updates CSI Matrix Lists, repeat step 3.2 according to CSI the or ACK frames of feedback) until selected use
Amount amount reaches AP antenna amounts M, and M user carries out transmitting uplink data；
4) downlink transmission：
4.1) user verifies that last time reports the validity of CSI after NDP bags are received, if last time reports CSI effectively, distributes
Enter good groups, if last time report CSI is invalid, be designated as bad groups；
4.2) good groups user to AP feeding back ACK frames, AP selects M concurrent user from good group users, if good groups
Number of users is less than M, then bad groups user recalculates new CSI and feed back to AP according to the NDP bags for receiving, and AP is received in succession
New CSI feedback reaches M until the quantity of user, and downlink transmission is carried out based on fair controlling mechanism.
The step 1) in authenticating user identification and CML initialization include：First user add Wi-Fi network in and complete
Authentication and association process, then AP New any and noted down in CSI Matrix Lists, New any includes four Column Properties：User
ID, CSI matrix, reusable and coherence time.
It is described reusable to include two values：0 and current time stamp, 0 represent current CSI matrixes lower whorl is transmitted it is invalid, one
Denier AP receives the CSI of new feedback, and CSI Matrix Lists record the timestamp that current CSI is reached, and indicate CSI from this moment on
Can use, it is reusable to be changed into 0 after associated user's statement CSI is invalid；The coherence time is channel coherency time of upper a moment, i.e.,
Last time reports the term of validity of CSI matrixes, by it is reusable be changed into 0 when timestamp subtract the timestamp for recording before and obtain.
The step 3) and step 4) in the Effective judgement of CSI be：Judge the CSI values H that last round of data transfer is usedl
It is whether equal with the CSI values Hn that epicycle is calculated, if Hl=Hn, then replace CSI as feedback with ACK；Otherwise count again
Calculate CSI and fed back to AP.
The step 4.2) in the process of setting up of fair controlling mechanism include determining user's continuous service time, set up user
Fair control in scheduling mechanism and large scale network.
The determination of user's continuous service time：When representing the continuous service to user i in scheduling every time using ni
Between, continuous service time is calculated by following equation：
Wherein, large time slice are the coherence time of all users' maximums in CSI Matrix Lists,
CoherenceTime i are the coherence time of user i；In formulaUnder round.
The upper limit of the large time slice is set to 30ms.
The user scheduling mechanism is set up：All users are ranked up according to coherence time, then according to order to
Family is scheduled.
Fair control in the large scale network：Multiple AP is set first in extensive MU-MIMO networks, and root is then
User is distributed to different AP according to coherence time in CSI Matrix Lists, being connected to the user of same AP has identical CSI phases
The dry time.
Compared with prior art, present invention validity of introducing CSI in up-link transmission and downlink transmission is sentenced
Disconnected, checking last time reports the validity of CSI, and last time reports CSI effectively, is then dispensed into good groups to AP feeding back ACK frames；Last time converges
Report CSI invalid, then need bad groups user to recalculate new CSI and feed back to AP according to the NDP bags for receiving, in uplink,
One group of concurrent user the most orthogonal can be chosen, and the orthogonality of concurrent user can be by the signal of precoding in downlink transfer
Ensure, always to the priority of the still effective user Geng Gao of CSI in CML, such preference can maximum journey for user scheduling mechanism
CSI feedback operation overhead in the reduction downlink transfer of degree, while in the case where fair controlling mechanism is introduced, it is ensured that user has justice
Data transfer, the present invention for existing CSI feedback mechanism exist low efficiency problem, in 802.11ac agreement MU-MIMO nets
CSI utilization rates are improved in network, CSI feedback expense is reduced, while ensure that the fairness of data transfer.
Further, in order to promote the justice of handling capacity, we introduce the general of " big timeslice " (large time slice)
Read, using the maximum coherence time of all users in CML as a single timeslice, each user can be in once dispatching
With the continuous time being serviced as close possible to big timeslice.
Further, the user with short coherence time can slow down the message transmission rate with long coherence time user, pin
To this problem, all users are ranked up according to coherence time, sequentially user is scheduled according to this then, it is adjacent
User there is similar channel coherency time, can simultaneously maximize CSI utilization rates and network throughput.
Further, many AP are utilized come in solving extensive MU-MIMO networks by extensive MU-MIMO networks
Fair control problem, distributes to different AP user according to the coherence time in CML, and user's meeting of same AP is connected to accordingly
With identical CSI coherence times, realize distributing to more closing it from current AP in the case where user's coherence time changes
Suitable AP, realizes the balance of load balance between network performance performance and AP.
Brief description of the drawings
Fig. 1 is flow chart of the method for the present invention；
Fig. 2 a are the CSI feedback mechanism under standard 802.11ac agreements, and Fig. 2 b are the CSI under Grandet methods of the present invention
Feedback mechanism；
Fig. 3 a, 3b and 3c are respectively the present invention in 2,3 and 4 antenna AP, and CSI is anti-in continuous 100 wheel transmission under 6/12 user
The total degree of feedback；
Fig. 4 a are 6 channel coherency time figures of user, under Fig. 4 b are standard 802.11ac agreements, Grandet of the present invention
Throughput gain figure of the method under 4 users and 8 users；
Fig. 5 a are in uplink Plays 802.11ac agreements, OPUS, Signpost and Grandet methods of the present invention
Handling capacity, Fig. 5 b are in downlink transfer Plays 802.11ac agreements, OPUS, Signpost and Grandet methods of the present invention
Handling capacity；
Fig. 6 a are scheduling mechanism effect figure of the invention, and Fig. 6 b are standard 802.11ac agreements, OPUS, Signpost and sheet
Assembly effect figure of the invention Grandet methods under different scheduling mechanisms；
Fig. 7 a are the total number of users of standard 802.11ac agreements, OPUS, Signpost and Grandet methods of the present invention to anti-
The influence of expense is presented, Fig. 7 b are the antenna number of standard 802.11ac agreements, OPUS, Signpost and Grandet methods of the present invention
The influence to feedback overhead is measured, 4 antenna AP are selected；
Fig. 8 a are standard 802.11ac agreements, the fair controlling mechanism (G w/o) of Grandet of the present invention introducings and no introducing
2 antenna AP under fair controlling mechanism (G w/), each user throughput of 4 users and the ratio of network throughput；Fig. 8 b are mark
Quasi- 802.11ac agreements, Grandet of the present invention introduce fair controlling mechanism (G w/o) and without the fair controlling mechanism (G of introducing
W/ 4 antenna AP under), each user throughput of 8 users；
Fig. 9 a are 2 antenna AP, each number of users of 8 users under standard 802.11ac agreements and Grandet methods of the present invention
According to conveying length；Fig. 9 b are 4 antenna AP, each user of 8 users under standard 802.11ac agreements and Grandet methods of the present invention
Data transfer length；
Figure 10 a are the effective planispheres of CSI, and Figure 10 b are the invalid planispheres of CSI；
Figure 11 a are the A-MPDU frame formats of standard 802.11ac, and Figure 11 b are ACK frame formats of the invention.
Specific embodiment
The present invention is further explained with reference to specific embodiment and Figure of description.
Referring to Fig. 1, the present invention specifically includes following steps：
1) authenticating user identification and initialization：
User add Wi-Fi network in and complete authentication and association process, AP can be at CSI Matrix Lists (CML)
For its newly-built one and it is noted down, this includes four Column Properties：ID, CSI matrixes, reusable and coherence time；
ID have recorded the IP address of each user, be used to distinguish different users；
CSI matrixes have recorded the last CSI matrixes reported between transmission antenna and user antenna；
It is reusable to include two values：' 0 ' and current time stamp, ' 0 ' represent current CSI matrixes lower whorl is transmitted it is invalid, one
Denier AP receives the CSI of new feedback, and CML records the timestamp that current CSI is reached, and indicates that CSI can use from this moment on, the value
Just it is changed into 0 after associated user's statement CSI is invalid；
Coherence time indicates a moment channel coherency time, is the term of validity that last time reports CSI matrixes, and it can pass through
Existing timestamp when being changed into 0 subtracts the timestamp for recording before and obtains, and reusable and timestamp can be used to ensure that downlink transfer
The fairness of middle handling capacity；
A complete MAC protocol generally comprises 3 parts in MU-MIMO networks：User is selected, and CSI is collected and data are passed
Defeated, because the user's selection mechanism in uplink and downlink transmission is not very identical, the present embodiment is separately discussed：
2) AP broadcast NDPA and NDP packets, state that this data transfer is up or descending, if up-link is passed
It is defeated, go to step 3)；If downlink transmission, step 4 is gone to)；
3) up-link transmission：
3.1) basic user selection：User's selection mechanism is carried out using OPUS in uplink, although it is set for descending
Meter, it remains to work well in uplink, and it at most only needs to M wheel user's selections, and M represents the antenna number of AP,
In each round, the user that maximum contribution can be made to downlink transmission capacity can be chosen；
Referring to Fig. 2 b, Grandet methods of the present invention inherit this increment user selection, but it is opened in reduction CSI feedback
There is noticeable effect on pin, the topmost differences of Grandet and OPUS are the method that CSI is collected, except selected
User needed in epicycle by CSI feedback to AP outside, Grandet again may be by CML obtain user CSI, by each
User assesses oneself last time and reports the validity of CSI and realize, and the latter is that Grandet collects CSI topmost methods；
3.2) MAC protocol is designed：Grandet remains the basic operation of 802.11ac agreements, removes its additional
Renewal on CML operations, selected user can report one of following two bags：CSI matrixes or ACK frames, higher level
On, Grandet works as follows：
I the intention of its uplink is stated in () AP statements by NDPA and NDP bags；
(ii) each user assesses the validity of oneself last time report CSI after NDP bags are received, and is verified by implementing CSI
Mechanism realizes, if the CSI that reports of its last time or effective, it can be dispensed into " good " group, otherwise, be designated as
" bad " group, the user of " bad " group should calculate a new CSI according to the NDP bags for receiving；
(iii) in subsequent each round, the user for feeding back power is won by the CSI feedback of oneself to AP, if it is located at
In " good " group, it only need to feed back an ACK frame to AP；
(iv) AP updates CML according to CSI/ACK feedbacks, and repeat step (iii) reaches M until selected number of users；
3.3) validity standard is designed：It is to judge the last round of CSI for using to the relatively straightforward method of CSI validation verifications
(Hl) CSI (H that are calculated with the wheeln) whether equal, if Hl=Hn, need to only replace CSI to be carried with this as feedback with ACK
High-throughput and general performance, otherwise, recalculate CSI and are fed back to AP, however, due to that can not directly obtain HlWith Hn
So can not be compared, one of them most important obstacle is noise can not to be picked from the signal for receiving for a full due
Remove；
Basic communication model as shown in Equation 1, Y, H, X and N~С N (0, σ2) it is respectively to receive signal, channel collaboration
CSI, transmitting end signal and complex-valued Gaussian noise：
Y=HX+N (1)
Because the change of H can cause the change of Y, the difference attempted between analysis Yl and Yn is used to react channel variation, its
Middle Yl、YnIt is that upper wheel takes turns the training symbol for receiving with this；
Consider a stochastic variable
η=Yn-Yl
=(Hn-Hl)X+(Nn-Nl) (2)
Situation 1：Hn=HlThe variance that η can be obtained is：
D(ηeq)=D (Nn-Nl)
=D (Nn)+D(Nl)
=2 σ2
Notice that Gaussian noise obeys stationary random process and be independent same distribution, therefore D (N can be taken apartn-Nl) and not
Need covariance；
Situation 2：Hn≠HlBecause noise is independently of signal, it is possible to split η as follows：
D(ηne)=D ((Hn-Hl)X+Nn-Nl)
=D ((Hn-Hl)X)+D(Nn-Nl)
The σ of=D (л)+22
Apparent D (л) reflects HnWith HlDifference because do not know represent complex-valued Gaussian noise distribution parameter σ2, nothing
Method Direct Analysis D (л)；
η is obtained in new round transmissiontest, it is necessary to judge HnWith HlIt is whether equal, check σ1 2=D (ηtest) and σ2 2=D
(ηeq) ratio, ifMay infer that HnAs close as Hl；
The physical meaning of above-mentioned ratio is described using the planisphere in Figure 10 a and 10b, positioned at the small square in circle center
The signaling point that is sent of representative, receives signal and is first mapped to closest constellation point in planisphere, due to channel-changing
With noise, receive signal and often be offset from transmission signal constellation point, it is closer compared with dark colour point due to the influence of ambient noise
In transmission signal point, after channel-changing, the signal calculated by history CSI generally can further off in transmission signal point, such as compared with
Shown in the point of light colour, and, only the offset signal in outermost layer circle just can be parsed correctly, use variance
The departure degree of signal is received to characterize, so σ2 2The departure degree under only environmental noise is represented, c σ are used2 2Represent letter
Number analysable maximum deviation degree, it is clear that this value is related to the variance of noise；
Based on above-mentioned and planisphere inherent characteristic, hypothesis below is set up：
C is set and is slightly larger than 1, ifMeet, it is meant that this wheel of the CSI of upper wheel report is still effective, it is assumed that inspection is used
F- distribution inspections；
Hypothesis testing：SelectionAs test data, wherein n1With n2Correlated samples number, S are represented respectively1 2n1、
S2 2n2Correlated samples variance is represented respectively, ifMeet, there are F~F (n1-1,n2- 1), wherein F represents F distributions, and correlation is refused
Distant and inaccessible land is：
Wherein α is significance, represents confidence level, and next the value of c is discussed；Tested under average SNR
To choose most suitable c=1.18；
η is obtained in every wheel data transfer, (12 short to there are 16 training symbols in the VHT preamble of NDP bags
Training symbol and 4 training symbols long), based on this, our CSI authentication mechanisms are described；
CSI authentication mechanisms：After every wheel receives NDP bags, user calculates following sample point and is based on these 16 training symbols：
yi=Yni-Yli, i=1,2 ... 16,16 training symbols entrained in a NDP bag have identical channel to estimate, following meter
Calculate yeq：
It is 0.01, F to set α0.01(15,15)=3.52, check S1 2And S2 2Ratio, wherein S1 2And S2 2Y is represented respectively
With yeqVariance, if ratio is less than c=1.18, it is meant that this wheel of the CSI of upper wheel is still effective；
3.4) CSI feedback Overhead Analysis：OPUS needs to obtain CSI from all selected users, and the present invention allows user
Replace CSI feedback to AP with ACK, showed to compare both, the propagation delay of following analysis CSI feedback and ACK feedbacks：
802.11n describes frame aggregation and reduces feedback overhead to relative, however, 802.11ac takes new polymerization side
Formula：All data frames being transmitted use polymerization MPDU (A-MPDU) form, even if contains only a frame in A-MPDU, because
This, simple CSI feedback bag needs substantial amounts of assistance position to transmit, and conveying length is defined with the time；Figure 11 a illustrate A-
The form of MPDU, CSI is a complex matrix for N × M × 52, and N represents user antenna number, and M represents AP antenna numbers, and 52 represent son
Carrier number, in standard 802.11ac agreements, a plural number needs 64bit to represent, for a 4 antenna AP, 1 antenna user
Scene, the size of CSI matrixes is：1 × 4 × 52 × (64/8)=1664bytes, this needs 32 OFDM symbols, can be every
Individual symbol increases by 4 μ s, and finally, this value is embedded into data field, additionally, VHT-LTF is mainly used in channel estimation and correction, its length
Degree depends on the number of transport stream, therefore is equal to 4 herein, therefore the total length of CSI feedback bag is about 180 μ s；
Figure 11 b illustrate the form of ACK frames, and ACK only has 14bytes altogether, can be by an OFDM symbol transmission close to 4
μ s, so it can be data transfer 176 μ s of saving to be fed back using ACK, and, at its best, the time of saving can surpass
700 μ s are crossed, from 4 ACK feedbacks, this is one 44 times of time tranfer, the data transfer of 3ms average for, this
Invention is capable of achieving 23% throughput gain, and this numeral can reach 47% under the scene of 8 antennas.
4) downlink transmission：
In MAC protocol it is up with descending only difference is that user scheduling mechanism, in uplink, one group the most orthogonal
Concurrent user can be chosen, but in downlink transfer concurrent user orthogonality can by precoding signal ensure, use
Always to the priority of the still effective user Geng Gao of CSI in CML, such preference can be reduced farthest family scheduling mechanism
CSI feedback operation overhead in downlink transfer, however, on the other hand, the user with longer CSI life cycles can obtain longer
Transmission time, then must be using a correct justice controlling mechanism；
4.1) MAC protocol summary：After NDP bags are received, each user k verifies that its last time reports the validity of CSI, if
Effectively then it can be divided to " good " group its history CSI, and feedback contention-based mechanism is divided into two parts, slightly has not with uplink
Together, in the first stage, own " good " group user and feed back an ACK frame to AP, but M user is only needed in uplink to AP
ACK frames are returned to, M represents the number that transmitting terminal transmits antenna, and M concurrent user is selected in the user that AP is organized from " good ", however,
If the number of users of " good " group is less than M, AP enters the second stage of operation, CSI feedback that what AP can be successive receive until with
The quantity at family reaches M；
The reason for needing all " good " group user rather than M user's transmission ACK frame is as follows：If each user is
The channel of downlink transfer and compete, and M wheel user selection in selected M user, a worse result is AP
It is not in need to user's transmission data of winning, simultaneously as the short many of the length of ACK, the cost that this process spends is
It is inappreciable；
Although the handling capacity realized depends on the orthogonality of user's channel status at once, the present invention pursues CSI multiplexings simultaneously
And it is not inefficiency to ignore orthogonality in user's selection course, the performance of network will not be tied down, the present invention is considered in CML
All CSI matrixes fed back after feedback contention-based, wherein have selected the user best with selected user orthogonality, it is true
Protect and selected the user with more preferable orthogonality in the user of " good " group；
4.2) fair controlling mechanism：Consider following scene：In the lobby on airport, the security walked about and the passenger being seated,
In at the restaurant, the waiter for walking about provides service for the guest being seated before dining table, it is apparent that the CSI of waiter is very
Can change in short time interval, they can obtain worse Consumer's Experience, because Grandet is more biased towards in service with more long
The user of coherence time.
For fair control problem, a direct solution is dispatch of taking turns, and this mechanism ensure that each user
There is the scheduled chance of identical, however, the timeslice distributed to scheduling every time is not identical, once user's quilt
Scheduling, it may always obtain the channel right to use and possess the right of transmission always before CSI failures, it is assumed that A and B it is relevant when
Between be respectively 30ms and 7ms, it means that, in its respective scheduling, A and B can obtain the consecutive hours of 30ms and 7ms respectively
Between, this mechanism based on coherence time unfair problem serious for handling capacity brings；
In order to promote the justice of handling capacity, a concept for being called " big timeslice " (large time slice) is introduced,
Using the maximum coherence time (CoherenceTime) of all users in CML as a single big timeslice (large time
Slice), then, each user can examine in once dispatching with the continuous time being serviced as close possible to " big timeslice "
Consider scene as described below, " big timeslice " is equal to 30ms, in addition, this 30ms means the data of the 10 a length of 3ms of wheel mean time
Transmission.User B should do 5 CSI feedbacks to obtain the service of 30ms timeslices, but the CSI of its 5th time feedback is only used
In the data transfer of 2ms, this efficiency is very low and has tied down whole network handling capacity, but which ensure that network throughput is used with each
Family handling capacity it is compromise；
The continuous service time to user i in scheduling every time is represented using ni, continuous service time can be by following public affairs
Formula is calculated：
In that case, in formulaUnder round, nB=28ms, it means that once user B
Dispatched by AP, it only being serviced 28ms；
In addition, in order to ensure user can frequently be switched, the upper limit for setting timeslice is 30ms, and this is truly to move
An average channel coherency time in dynamic MU-MIMO networks, additionally, 20 networks of user are connected with for one, often
Individual user it is per second it is interior being serviced 3 times；
4.3) user scheduling operation：
Consider there is an AP with two antennas, tri- user's coherence times of A, B, C in a simple MU-MIMO network
It is 30ms, 6ms, 27ms, it is assumed that user A and B concurrent being serviced, according to fair controlling mechanism, " big timeslice " is 30ms,
User B should do 5 CSI feedbacks to obtain a service time of 30ms, because the CSI of user A is always obtainable,
A can only be waited in the interval time of the CSI feedback for obtaining user B AP5 times, and in a word, the user with short coherence time can slow down tool
There is the message transmission rate of long coherence time user, this is not efficient for A；
In order to solve this problem, all users are ranked up according to coherence time, then according to this sequentially to
Family is scheduled, and adjacent user has similar channel coherency time, and it can simultaneously maximize CSI utilization rates and be gulped down with network
The amount of telling；
4.4) justice in large scale network：Solved by setting many AP in extensive MU-MIMO networks such
Fair control problem, distributes to different AP from the crowd for sitting quietly what is walked about according to the coherence time in CML, is connected to accordingly
The user of same AP can have identical CSI coherence times, and the vital part of this algorithm how is realized in user's phase
It is distributed to more suitably AP from current AP in the case of dry time change, additionally, this relate to network performance performance with
The balance of load balance between AP.
Embodiment is tested on the basis of Grandet prototypes are constructed and evaluates its performance.
A. experiment is set up：
A MU-MIMO OFDM program based on 802.11ac agreements is set up on USRP, Grandet is built according to this former
Type, this program realizes OFDM modulation, and detective is surveyed, and channel estimation and signal are demodulated, and in order to realize Grandet, are estimated to channel
CSI authentication mechanisms are added in meter model, CML and the user scheduling mechanism based on CML is furthermore achieved that, due to USRP interfaces
Delay, it is impossible to be directly realized by it is feedback contention-based in real time, however, due to all of USRP wireless devices connection on experimental bench
To a PC controller, user scheduling process is emulated on computers, the user that competition is won in its order starts CSI/ACK feedbacks.
For Performance comparision, 3 MU-MIMO systems are equally realized：The 802.11ac all of users of requirement of (i) standard
Its CSI is fed back in transmission every time；(ii) OPUS operation increments user selection, it at most only needs to M wheel CSI feedbacks, and M is equal to AP
Antenna number；(iii) Signpost realizes zero CSI expense competition mechanisms, but in order to data decoding needs orderly anti-of user
Feedback CSI.
Grandet prototypes are made up of USRP-N210 to USRP-X310 radio platforms and related UHD software kits, one
The AP of 2- antennas is set up on the basis of USRP-X310 and two SBX version, using 2 USRP-N210 as two simultaneously
The client of hair, each USRP-N210 is equipped with a SBX version, there is provided the bandwidth of 40MHZ, in order to allow many clients simultaneously
Hair transmission, two USRP-N210 is connected on notebook computer and controls it to transmit by command script.One similar finger
Script is made to be also contained in two SBX versions of USRP-X310.In order to realize accurate time synchronized, using an external clock
Model USRP CLOCK DISTRIBUTION 782979-01 connect the AP and client of 2 antennas as an ordinary clock source.
Environment implements experiment indoors.The unemployed and surrounding that all of MU-MIMO transmission operates in a 2.4GHZ wirelessly sets
On standby non-overlapping channel.Other physical layer parameters obey 802.11ac acquiescences.Because hardware environment is limited, for being related to exceed
Experimental bench node, is moved to different positions by 4 experiments of concurrent client, collects each subcarrier up channel matrix, and
CSI traces are fed back into USRP.
B.CSI is multiplexed compliance test result：
Performance of the CSI multiplexings on CSI feedback expense is reduced is first verified that, Fig. 3 a, 3b and 3c depict Grandet systems
The total degree of CSI feedback, implementation experiment is carried out by following benchmark respectively in continuous 100 wheel transmission：2nd, 3,4 antenna AP, 6/12 uses
Family (half user therein is moved with approaching people's speed of travel).It is observed that Grandet averagely only needs a wheel CSI feedback,
But standard 802.11ac agreements need 6/12 wheel CSI feedback under correlation circumstance, it is obvious that feedback cost can be by CSI
It is multiplexed and significantly reduces.
Fig. 4 a illustrate wherein 6 channel coherency times of user, and the handling capacity of network is checked in fig. 4b.Grandet
59.5% handling capacity is obtained than the 4 antenna AP and 8 user's topological structures under standard 802.11ac agreements more.We equally see
The handling capacity for observing Grandet increases with the growth of antenna amount, and by contrast, when number of users quantitative change is a lot, it subtracts
It is few insignificant.
C. throughput gain：
As shown in Fig. 5 a and Fig. 5 b, the relatively handling capacity of different MU-MIMO systems, with standard 802.11ac, OPUS,
Signpost is compared, and its gain is 1.78,1.37,1.23 times in Grandet methods uplink of the present invention, is in downlink transfer
2.03rd, 1.64,1.37 times.
Fig. 6 a illustrate that user scheduling mechanism of the present invention based on CML is showed, and are chosen the channel angle of user up
It is much like with downlink transmission process, because AP is always to the priority of the effective user Geng Gao of CSI in downlink transfer.Comment now
Estimate the influence of frame aggregation, from Fig. 6 b it can be seen that all user's selection mechanisms under network throughput with frame length (
It is maximum to 5.5ms in 802.11ac agreements) increase and increase, Grandet keeps highest handling capacity in all cases.
D.MAC layers of expense：
MAC expenses being considered now and checking whether Grandet is mensurable, the feedback in relatively more different MU-MIMO systems is opened
Pin, shown in Fig. 7 a, in order to verify the influence of user number, we are tested with the AP of 4 antennas, it is obvious that result is from figure
7b can be seen that Grandet and win other systems.
E. Throughput fairness：
In this miniature inspection program, the fair control assembly of Grandet is assessed, be each network topology operation 400
Wheel downlink transfer.As shown in Figure 8 a, the ratio of each user throughput and network throughput is illustrated, it is apparent that
The fair controlling mechanism of Grandet effectively maintains the handling capacity of each user.Additionally, referring to Fig. 8 b, this mechanism with do not have
The Grandet for having fair controlling mechanism is compared and is not almost caused throughput loss.
F.CSI is utilized：
For the CSI utilizing status of standard of comparison 802.11ac and Grandet, there is AP and 8 of 2/4 antenna at one
Tested on the base line topology of user, the new ideas for being called Packet Per CSI (PPCSI) are introduced, such as name institute
Show, test can be supported to the CSI data transfers how long that each is reported, this has with the signal coherence time and necessarily contacts.
We are that each user calculates average PPCSI and these results are illustrated in Fig. 9 a and 9b.Conveying length is fixed by the time
Justice.It is observed that：(i) in the 802.11ac agreements of standard, due to ignore CSI terms of validity PPCSI 2ms and 5.5ms it
Between.Grandet performances compared with stock situation enhance 10 times；(ii) with the increase of AP antenna amounts, the double increasings of PPCSI
It is long.This is primarily due to user and frequent within its coherence time can be scheduled.
Claims (9)
1. it is a kind of based on the term of validity check CSI feedback overhead reduction method, it is characterised in that comprise the following steps：
1) authenticating user identification and CML are initialized；
2) AP broadcast NDPA and NDP packets, state that this data transfer is transmitted or downlink transmission for up-link, if
Up-link is transmitted, and goes to step 3), if downlink transmission, go to step 4)；
3) up-link transmission：
3.1) intention of its uplink is stated in AP statements by NDPA and NDP bags；
3.2) user verifies that last time reports the validity of CSI after NDP bags are received, if last time reports CSI effectively, is dispensed into
Good groups；If it is invalid that last time reports CSI, bad groups are designated as, bad groups user recalculates new CSI according to the NDP bags for receiving；
3.3) in subsequent each round, the user for feeding back power is obtained by the CSI feedback of oneself to AP, if being located in good groups,
Only to AP feeding back ACK frames；
3.4) AP updates CSI Matrix Lists, repeat step 3.2 according to CSI the or ACK frames of feedback) until selected number of users
Amount reaches AP antenna amounts M, and M user carries out transmitting uplink data；
4) downlink transmission：
4.1) user verifies that last time reports the validity of CSI after NDP bags are received, if last time reports CSI effectively, is dispensed into
Good groups, if last time report CSI is invalid, are designated as bad groups；
4.2) good groups user to AP feeding back ACK frames, AP selects M concurrent user from good group users, if good group users
Quantity is less than M, then bad groups user recalculates new CSI and feed back to AP according to the NDP bags for receiving, and AP receives new in succession
CSI feedback reaches M until the quantity of user, and downlink transmission is carried out based on fair controlling mechanism.
2. it is according to claim 1 it is a kind of based on the term of validity check CSI feedback overhead reduction method, it is characterised in that
The step 1) in authenticating user identification and CML initialization include：User adds in Wi-Fi network and completes authentication first
With association process, then AP New any and noted down in CSI Matrix Lists, New any includes four Column Properties：ID, CSI squares
Battle array, reusable and coherence time.
3. it is according to claim 2 it is a kind of based on the term of validity check CSI feedback overhead reduction method, it is characterised in that
It is described reusable to include two values：0 and current time stamp, 0 represent current CSI matrixes lower whorl is transmitted it is invalid, once AP is received
To the CSI of new feedback, CSI Matrix Lists record the timestamp that current CSI is reached, and indicate that CSI can use from this moment on, can be again
It is changed into 0 with after associated user states that CSI is invalid；The coherence time is to report channel coherency time of upper a moment, i.e. last time
The term of validity of CSI matrixes, by it is reusable be changed into 0 when timestamp subtract the timestamp for recording before and obtain.
4. it is according to claim 1 it is a kind of based on the term of validity check CSI feedback overhead reduction method, it is characterised in that
The step 3) and step 4) in the Effective judgement of CSI be：Judge the CSI values H that last round of data transfer is usedlWith epicycle meter
Whether the CSI values Hn for obtaining is equal, if Hl=Hn, then replace CSI as feedback with ACK；Otherwise recalculate CSI and incite somebody to action
It feeds back to AP.
5. it is according to claim 1 it is a kind of based on the term of validity check CSI feedback overhead reduction method, it is characterised in that
The step 4.2) in the process of setting up of fair controlling mechanism include determining user's continuous service time, set up user scheduling mechanism
With the fair control in large scale network.
6. it is according to claim 5 it is a kind of based on the term of validity check CSI feedback overhead reduction method, it is characterised in that
The determination of user's continuous service time：The continuous service time to user i in scheduling every time is represented using ni, under
Row formula calculates continuous service time：
Wherein, large time slice are the coherence time of all users' maximums in CSI Matrix Lists, CoherenceTime
I is the coherence time of user i；In formulaUnder round.
7. it is according to claim 6 it is a kind of based on the term of validity check CSI feedback overhead reduction method, it is characterised in that
The upper limit of the large time slice is set to 30ms.
8. it is according to claim 5 it is a kind of based on the term of validity check CSI feedback overhead reduction method, it is characterised in that
The user scheduling mechanism is set up：All users are ranked up according to coherence time, user is adjusted according to order then
Degree.
9. it is according to claim 5 it is a kind of based on the term of validity check CSI feedback overhead reduction method, it is characterised in that
Fair control in the large scale network：Multiple AP is set first in extensive MU-MIMO networks, and root is then according to CSI squares
Coherence time distributes to different AP user in array table, and being connected to the user of same AP has identical CSI coherence times.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710036013.8A CN106712815B (en) | 2017-01-18 | 2017-01-18 | CSI feedback overhead reduction method based on validity period check |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710036013.8A CN106712815B (en) | 2017-01-18 | 2017-01-18 | CSI feedback overhead reduction method based on validity period check |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN106712815A true CN106712815A (en) | 2017-05-24 |
| CN106712815B CN106712815B (en) | 2020-06-19 |
Family
ID=58907631
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710036013.8A Expired - Fee Related CN106712815B (en) | 2017-01-18 | 2017-01-18 | CSI feedback overhead reduction method based on validity period check |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106712815B (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109510649A (en) * | 2018-12-28 | 2019-03-22 | 西安交通大学 | The data transmission method of MU-MIMO network based on CSI accuracy, validity and stability |
| CN109889247A (en) * | 2019-02-13 | 2019-06-14 | 中国人民解放军陆军工程大学 | Low overhead dynamical feedback safe transmission method and system suitable for narrowband Internet of Things |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2014008833A1 (en) * | 2012-07-09 | 2014-01-16 | 电信科学技术研究院 | Non-periodic channel state information transmission method, device and system |
| JP2016500942A (en) * | 2012-09-28 | 2016-01-14 | インターデイジタル パテント ホールディングス インコ | Method for WiFi beamforming, feedback and sounding (WiBEAM) |
| CN105637772A (en) * | 2013-10-17 | 2016-06-01 | 高通股份有限公司 | Methods and apparatus for channel state information feedback |
| CN106301720A (en) * | 2015-05-14 | 2017-01-04 | 北京三星通信技术研究有限公司 | The method and apparatus of transmitting uplink control information |
-
2017
- 2017-01-18 CN CN201710036013.8A patent/CN106712815B/en not_active Expired - Fee Related
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2014008833A1 (en) * | 2012-07-09 | 2014-01-16 | 电信科学技术研究院 | Non-periodic channel state information transmission method, device and system |
| JP2016500942A (en) * | 2012-09-28 | 2016-01-14 | インターデイジタル パテント ホールディングス インコ | Method for WiFi beamforming, feedback and sounding (WiBEAM) |
| CN105637772A (en) * | 2013-10-17 | 2016-06-01 | 高通股份有限公司 | Methods and apparatus for channel state information feedback |
| CN106301720A (en) * | 2015-05-14 | 2017-01-04 | 北京三星通信技术研究有限公司 | The method and apparatus of transmitting uplink control information |
Non-Patent Citations (1)
| Title |
|---|
| 蒋志平: "Communicating Is Crowdsourcing:Wi-Fi Indoor Localization with CSI-Based Speed Estimation", 《JOURNAL OF COMPUTER SCIENCE AND TECHNOLOGY》 * |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109510649A (en) * | 2018-12-28 | 2019-03-22 | 西安交通大学 | The data transmission method of MU-MIMO network based on CSI accuracy, validity and stability |
| CN109510649B (en) * | 2018-12-28 | 2021-02-09 | 西安交通大学 | Data transmission method of MU-MIMO network based on CSI accuracy, validity and stability |
| CN109889247A (en) * | 2019-02-13 | 2019-06-14 | 中国人民解放军陆军工程大学 | Low overhead dynamical feedback safe transmission method and system suitable for narrowband Internet of Things |
| CN109889247B (en) * | 2019-02-13 | 2022-07-05 | 中国人民解放军陆军工程大学 | Low-overhead dynamic feedback safe transmission method and system suitable for narrow-band Internet of things |
Also Published As
| Publication number | Publication date |
|---|---|
| CN106712815B (en) | 2020-06-19 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN103181206B (en) | Wireless communication system and wireless communications method | |
| CN106559879B (en) | The method and device that information is sent and determining, relationship determines | |
| US10257735B2 (en) | Wireless communication device and wireless communication method for controlling transmission timing based on communication quality | |
| CN104396296B (en) | Data transmission method, device and user equipment | |
| CN104253673B (en) | Uplink multi-users data transmission method and uplink multi-users input-output system | |
| CN102792628B (en) | Be used for transmission method and the device of the downlink OFDMA of the service groups with mixed-client type | |
| Zhou et al. | Signpost: Scalable MU-MIMO signaling with zero CSI feedback | |
| CN104427632B (en) | Data transmission scheduling method, device and system | |
| CN101917771A (en) | Method and apparatus for scheduling in a wireless network | |
| CN110392985A (en) | Distributed MIMO communication is triggered in radio node cluster | |
| CN109661784A (en) | Efficient channel for implicit MU-MIMO is estimated | |
| US9768992B2 (en) | System and method for randomized antenna allocation in asynchronous MIMO multi-hop networks | |
| CN105794296A (en) | Method and device for receiving multiuser uplink in wireless LAN | |
| US20120026928A1 (en) | Techniques for uplink multi-user mimo mac support | |
| CN107005288A (en) | Transmission antenna diversity scheme | |
| CN105101355B (en) | A kind of access point selection method based on user throughput estimation in wlan system | |
| CN107113640A (en) | For method, the related circuit of execution method and the application of the Radio Link for analyzing WI FI nodes | |
| CN102821478B (en) | A kind of starlike network-building method of broadband wireless access equipment based on cognition | |
| CN104883707B (en) | A kind of test system and method for testing of wireless aps antijamming capability | |
| CN102695182B (en) | A kind of communication processing method and base station | |
| CN106712815A (en) | Validity examination based CSI feedback overhead reduction method | |
| CN103313367A (en) | Uplink multi-user method applicable to wireless local area network | |
| US20170273092A1 (en) | Wireless communication device and wireless communication method | |
| CN106789767B (en) | Distributed C/S I estimation and dynamic updating method in a kind of WLAN | |
| CN103152807A (en) | Method for distributing power between multiple base stations and multiple antennae of TDD (Time Division Duplex) cooperative wireless network |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20200619 |