CN110071881A - A kind of any active ues detection of adaptive expense and channel estimation methods - Google Patents

A kind of any active ues detection of adaptive expense and channel estimation methods Download PDF

Info

Publication number
CN110071881A
CN110071881A CN201910345235.7A CN201910345235A CN110071881A CN 110071881 A CN110071881 A CN 110071881A CN 201910345235 A CN201910345235 A CN 201910345235A CN 110071881 A CN110071881 A CN 110071881A
Authority
CN
China
Prior art keywords
active ues
channel
pilot
frequency domain
channel estimation
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
Application number
CN201910345235.7A
Other languages
Chinese (zh)
Other versions
CN110071881B (en
Inventor
高镇
柯玛龙
肖振宇
吴泳澎
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Beijing Institute of Technology BIT
Original Assignee
Beijing Institute of Technology BIT
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Beijing Institute of Technology BIT filed Critical Beijing Institute of Technology BIT
Priority to CN201910345235.7A priority Critical patent/CN110071881B/en
Publication of CN110071881A publication Critical patent/CN110071881A/en
Application granted granted Critical
Publication of CN110071881B publication Critical patent/CN110071881B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

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
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L25/00—Baseband systems
    • H04L25/02—Details ; arrangements for supplying electrical power along data transmission lines
    • H04L25/0202—Channel estimation
    • H04L25/0224—Channel estimation using sounding signals
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L25/00—Baseband systems
    • H04L25/02—Details ; arrangements for supplying electrical power along data transmission lines
    • H04L25/0202—Channel estimation
    • H04L25/024—Channel estimation channel estimation algorithms
    • H04L25/0242—Channel estimation channel estimation algorithms using matrix methods

Landscapes

  • Engineering & Computer Science (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • Signal Processing (AREA)
  • Power Engineering (AREA)
  • Physics & Mathematics (AREA)
  • Mathematical Physics (AREA)
  • Mobile Radio Communication Systems (AREA)
  • Radio Transmission System (AREA)

Abstract

The invention discloses a kind of detection of any active ues of adaptive expense and channel estimation methods, utilize any active ues set and its channel estimated, signal errors assessment estimation quality is received by calculating, it can be according to any active ues quantity and channel circumstance in reality system, the time slot expense of pilot tone in adaptive adjustment frame structure, guarantee service quality with alap access delay, realizes super reliable any active ues detection and channel estimation.Present invention is alternatively directed to the enhancing mobile broadband scenes and the huge location communication scenes such as large-scale machines type communication in Future cellular networks; utilize the fragmentary characteristic of user uplink transmission flow and the sparsity of the virtual angle domain channel of extensive MIMO; rationally design the pilot signal for exempting to authorize multiple access access for broadband upstream; and advanced compressed sensing technology is introduced, user's access delay can be greatly reduced.

Description

A kind of any active ues detection of adaptive expense and channel estimation methods
Technical field
The present invention relates to any active ues in wireless communication and channel estimation field, more particularly to are supporting to enhance mobile broadband Huge location communication in, mass users uplink multi-address access scene under any active ues detection and channel estimation.
Background technique
With video flowing, the fast development of social networks and emerging technology of Internet of things, in following cellular network, The quantity and service traffics of user equipment all will be in explosive increase, it is therefore desirable to which base station is supported to be formed between mass users The huge connection of huge service traffics can be carried.The design of cellular network at present is mainly for mobile phone user, supported user Limited amount extremely challenges the reliable support of huge dimension uplink multi-address access for current network.
Tradition needs additional control signaling and scheduling of resource based on the random division multiple accessing agreement of authorization, and user equipment needs Network could be accessed by base station authorization.Physical Random Access Channel agreement in typical such as 4G-LTE system, is divided into uncontested Random division multiple accessing and competition random division multiple accessing.In uncontested random division multiple accessing, the specific pilot tone of each user of base station dispensing, in order to base It stands and distinguishes different users；And compete in random division multiple accessing, any active ues choose pilot tone simultaneously from predefined pilot set It sends, when two users choose same pilot, additional request collision is needed to handle.Unfortunately, in the communication of huge location, by Huge in number of users, the access way of this competitive resource will appear serious access request conflict, use so as to cause a large amount of It family can not fast access into network.If distributing orthogonal multiple access access-in resource to each user, and it will lead to great expense incurred.In addition, After user successfully accesses, it is also necessary to which the additional pilot tone that sends realizes uplink channel estimation.Therefore, logical for huge location Letter, traditional random multi-access channel scheme based on authorization will lead at higher access delay and the request collision of complexity Reason.On the other hand, it is that a large amount of user equipment is the sensing of low-power consumption battery limited that the following huge location, which communicates an important feature, Device etc. within the most of the time in a dormant state only can just wake up in the case where something outside triggers and send data. Therefore the typical feature of huge dimension multiple access access is large number of equipment not routinely to communication network uplink transmission data, but it is long when Between silence and burst transfer weave in.By the sparsity using this customer flow, there is scholar to propose to be based at present Compressed sensing exempts to authorize random multi-access channel scheme, and active equipment directly sends pilot tone and data to base station uplink, without Need the authorization by base station.The program can undoubtedly reduce access delay, but also bring new challenge, i.e. user needs basis Received signal detects active user, while carrying out channel estimation and Data Detection.
Using the fragmentary upstream traffic of user and compressed sensing technology, there are many any active ues detection schemes at present It is suggested, effectively reliably exempts to authorize random division multiple accessing to realize.The main thought of these schemes is to ask any active ues detection Topic is modeled as sparse signal reconfiguring problem.Specifically, Tsinghua University Wang Zhaocheng professor et al. proposes a kind of compressed sensing based Multi-user Detection Scheme realizes the detection of joint any active ues and data decoding.In order to further increase performance, Ore. Professor L.Wu of vertical university (Oregon State University) proposes base using the prior information for sending discrete symbols In AMP (approximate message transmission, Approximate Message Passing) algorithm and EM (expectation maximization, Expectation Maximization) algorithm any active ues detection scheme.Columbia Univ USA (Columbia University) X.Wang professor et al. points out uplink random multi-address signal there are block sparsity, and proposes a kind of TA-BSASP (the sparse adaptive subspace tracking of the block of thresholding auxiliary, Threshold Aided Block Sparsity Adaptive Subspace Pursuit) algorithm improves compressed sensing reconstruction property.In addition, the V.K.N.Lau of Hong Kong University of Science and Thchnology is taught Et al. the work of extensive random division multiple accessing is expanded into C-RAN (cloud wireless access network, Cloud Radio Access Network) in framework, reliable Multiple Access scheme under the framework is proposed.But above scheme all relies on perfect channel shape State information, and the acquisition of channel state information communicate especially for huge location, are a very challenging tasks.
Any active ues detection and channel estimation are realized in order to combine, Huawei Tech Co., Ltd (Canada) A.Bayesteh et al. proposes a kind of blind Detecting side of SCMA (Sparse Code multiple access, Sparse Code Multiple Access) Case come support huge location communicate in exempt from authorize random division multiple accessing, but the program only considers the scene of base station single antenna.For more days Professor V.K.N.Lau of linear system system, Hong Kong University of Science and Thchnology proposes for C-RAN network based on improvement BCS (Bayes's compression sense Know, Bayesian Compressive Sensing) Multiple Access scheme of algorithm, pass through and is received using up channel matrix difference Structural sparse between antenna improves any active ues detection and channel estimating performance.University of Toronto Professor W.Yu of (University of Toronto) by propose it is a kind of based on AMP algorithm any active ues detection and channel Algorithm for estimating is come the computation complexity that is further reduced in extensive random division multiple accessing, while it is logical in huge location to analyze multiaerial system Superiority in letter, however program needs are specifically distributed using channel and noise power is as prior information, in practical communication system There are still hinder in the application of system.
In addition, above-mentioned all extensive random division multiple accessing schemes consider the huge location communication in the narrowband of single carrier transmission, for The reliable support of extensive random division multiple accessing under eMBB (enhancing mobile broadband, Enhanced Mobile Broadband) scene, according to Be so one very challenge the problem of.Moreover, any active ues quantity and channel circumstance change constantly for real system, The degree of rarefication for leading to up channel matrix is time-varying, when degree of rarefication is smaller, as long as smaller pilot time slot expense Obtain reliable any active ues set and its channel；On the contrary, then needing biggish expense when degree of rarefication is bigger.It is above-mentioned all Scheme uses the pilot-frequency expense of regular length, undoubtedly will cause pilot tone waste or insufficient, it is therefore desirable to a kind of more flexible Mode rationally to control pilot-frequency expense.
Summary of the invention
In view of this, the present invention provides a kind of enlivening for adaptive expense for huge dimension random division multiple accessing under the scene of broadband User's detection and channel estimation methods, can adaptively adjust multiple access pilot tone according to practical any active ues quantity and channel circumstance and open Pin to guarantee reliable user's detection and channel estimating performance, and does not depend on the prior informations such as channel channel and noise；
Further, the present invention can use up channel space of matrices-frequency domain and the structuring of angle-frequency domain is dilute Property is dredged, access delay is greatly reduced.
In order to solve the above-mentioned technical problem, the present invention is implemented as follows:
A kind of any active ues detection of adaptive expense and channel estimation methods, comprising:
Step 1, in each time slot, any active ues send signal using the multiple access protocol uplink for exempting from authorization；
Step 2, according to initial slot expense G0, any active ues that collection base station end receives are in G0It is sent in a time slot Uplink pilot signal carries out any active ues detection and channel estimation using received signal, obtain any active ues set and Corresponding extensive mimo channel；
Step 3, any active ues set and respective channels obtained using estimation, reconstruct channel matrix, and calculate reception letter Number error；According to receiving signal errors assessment any active ues detection and channel estimation quality, and assessment result is broadcast to all User, if the result meets standard given in advance, any active ues stop sending pilot tone, only send out in time slot next Data are sent, to obtain reliable any active ues set and its channel；Otherwise, any active ues continue to send pilot tone, and base station end is more The reception signal for collecting a time slot re-starts any active ues detection and channel estimation, and re-execute the steps 3, until being received The reception signal of collection is enough to obtain reliable any active ues set and its channel.
Preferably, the corresponding frame structure of multiple access protocol for exempting from authorization used by step 1 includes time domain and frequency domain two A part；It is made of on frequency domain N number of subcarrier, is made of in time domain T time slot；Preceding G time slot for simultaneously send pilot tone with Data, and in N number of subcarrier of this G time slot, P pilot sub-carrier for sending pilot tone, use by (N-P) a data subcarrier In transmission data；Then (T-G) a time slot is only used for sending data, therefore all N number of carrier waves are data subcarrier.
Preferably, pilot signal transmitted by any active ues is non-orthogonal pilot signal, for each pilot sub-carrier, institute There is pilot signal of the user in pilot-frequency expense to generate by independent identically distributed standard multiple Gauss；On different pilot sub-carriers Pilot tone is different.
Preferably, any active ues detection and channel estimation of progress described in step 2 are as follows:
The sub-channel matrix of mimo channel vector composition based on all users has column sparsity, observes in different antennae The sparsity arrived is identical, and sub-channel matrix sparse pattern having the same corresponding to different pilot sub-carriers, from And any active ues test problems are modeled as spatial domain compressed sensing problem, constitute space-frequency domain model；
Based on extensive mimo channel, in virtual angle domain, there are cluster sparsities and different sub-channel matrix in frequency domain On also sparse pattern having the same, so that the compressed sensing that the channel estimation problems of any active ues are modeled as angle domain be asked Topic constitutes angle-frequency domain model；
Any active ues set is obtained using space-frequency domain model, is further enlivened by different confidence thresholdings Two different any active ues subsets of reliability in user's set；Any active ues set is substituted into angle-frequency domain model Channel estimation is carried out, and subtracts the corresponding signal of user in the high any active ues subset of reliability from receiving in signal, is connect Collection of letters residual error, then the residual error is substituted into space-frequency domain model and carries out remaining any active ues detection, and so on, alternately Ground carries out any active ues detection and channel estimation, until reaching given maximum number of iterations or receiving signal residual error less than thing First given thresholding.
Preferably, in the step 2, the modeling process of space-frequency domain are as follows:
For p-th of pilot sub-carrier, k-th of user that base station receives is defined in the signal of t-th of time slot are as follows:
WhereinFor p-th of pilot subchannel of k-th of user,It is k-th of user in p-th of pilot tone The uplink multi-address pilot tone transmitted in subchannel,It is the white Gaussian noise of p-th of pilot subchannel；Base station is equipped with extensive day Line, M are antenna for base station numbers；User uses single antenna；Total P pilot sub-carrier；
The factor of enlivening for further defining k-th of user is αk, 1 is taken when user enlivens, and 0 is taken when silent, then, base station end Total reception signal can be write as:
Wherein(·)TFor transposition symbol Number；K is total number of users；
So in continuous G time slot, the reception signal that base station end is collected into can be write as:
WhereinIt is empty Between domain uplink multi-address channel matrix,As it is desirable that G < < K, while channel matrix has column sparsity, There is common supported collection, therefore the model is space-frequency domain compressed sensing model between different lines；Known reception signalWith Pilot matrixEstimate XpSupported collection, any active ues set estimatedTo realize that any active ues detect.
Preferably, in the step 2, angle-frequency domain modeling process are as follows:
Consider channel transforming from a spatial domain to virtual angle domain, transformation relation are as follows:
Wherein wp,kIt is virtual angle domain channel；It is transformation matrix, is the uniform of half wavelength in antenna spacing It is discrete fourier matrix, () in the case of linear arrayHFor conjugate transposition symbol；Therefore, available angle-frequency domain mould Type:
Wherein,It is the access channel matrix of angle domain, (·)*It is conjugate of symbol；The model is used for channel estimation；Known reception signalIt is closed by the transformation of spatial domain and angle domain It is available equivalent reception signalIt utilizesPilot matrixAnd any active ues set of estimationIt can be with Estimation obtains the corresponding virtual angle domain channel of any active uesAgain by the channel conversion back to spatial domain, To realize the channel estimation of any active ues.
Preferably, using space-frequency domain model carry out any active ues detection, and using angle-frequency domain model into Row channel estimation is all made of improved AMP algorithm and resolves, and improved AMP algorithm learns hyper parameter using EM algorithm, and according to letter The sparsity structure optimization hyper parameter of road matrix updates rule；Wherein,
When improved AMP algorithm is applied to space-frequency domain model, sparse rateUpdate rule are as follows:
In formula (I), sparse rateIndicate that the probability of channel matrix (p, k, m) a element non-zero, p are pilot tone load The index of wave, k are user index, and m is the index of antenna for base station, and i represents the i-th iteration of algorithm, are gatheredHere (q, l, u) indicates the element index of three dimensional channel matrix, confidence The factorIt is intermediate variable defined in AMP algorithm；|·|cIndicate set element number；
When improved AMP algorithm is applied to angle-frequency domain model, the update rule of sparse rate are as follows:
In formula (II), set
The utility model has the advantages that
(1) present invention is estimated using any active ues set and its channel estimated by calculating reception signal errors Quality is counted, can be according to any active ues quantity and channel circumstance in reality system, the adaptive time slot for adjusting pilot tone in frame structure Expense guarantees service quality with alap access delay, guarantees super reliable any active ues detection and channel estimation.
(2) present invention be directed to broadband scene, propose it is a kind of suitable for multicarrier exempt from authorize multiple access protocol it is adaptive Frame structure, N number of subcarrier in preceding G time slot can carry pilot tone and data simultaneously, therefore user can award without base station Power sends pilot tone and data simultaneously, after base station carries out any active ues detection and channel estimation with pilot tone, may be implemented to data Decoding, between different pilot sub-carriers introduce pilot tone diversity, can to avoid complexity scheduling, substantially reduce access delay.
(3) the fragmentary characteristic for combining user uplink flow and the virtual angle domain of extensive mimo channel of the invention is sparse Property, the more measurement vector compaction perception of the distribution that any active ues detection and channel estimation are modeled as spatial domain and angle domain respectively Model, and it is alternately carried out any active ues detection and channel estimation, it can use two kinds of respective advantages of model, obtain enlivening use Family set and its corresponding channel.
(4) present invention is by being equipped with extensive antenna in base station, and utilizes the sparsity of the virtual angle domain of extensive MIMO, Access delay can further be reduced and improve any active ues detection and channel estimating performance.
(5) present invention learns the hyper parameter and noise variance of channel distribution by EM algorithm, convenient in systems in practice Application.
Detailed description of the invention
Fig. 1 is that the huge dimension multiple access in extensive mimo system accesses scene and one-ring channel model schematic diagram.
Fig. 2 is the structural sparse of space-frequency domain and angular frequency domain up channel matrix in the access of huge dimension multiple access Schematic diagram.
Fig. 3 is the flow chart of any active ues detection and channel estimation methods of the adaptive expense of the present invention.
Fig. 4 is that broadband of the present invention exempts to authorize the frame structure of multiple access protocol.
Fig. 5 is that the huge dimension Multiple Access scheme and three kinds of tradition based on spatial domain DMMV-AMP algorithm exempt to authorize the work of Multiple Access scheme The user's detection performance that jumps comparison.
Fig. 6 is that the huge dimension Multiple Access scheme and three kinds of tradition based on spatial domain DMMV-AMP algorithm exempt to authorize the letter of Multiple Access scheme Estimate performance comparison in road.
Fig. 7 is that any active ues detection performance of the present invention and three kinds of comparison schemes compare.
Fig. 8 is that the channel estimating performance of the present invention and three kinds of comparison schemes compare.
Fig. 9 is the enlivening under different active users of the invention from the Multiple Access scheme based on Turbo-DMMV-AMP algorithm User's detection and channel estimating performance.
Specific embodiment
The features of the present invention includes following aspects:
Firstly, receiving signal errors assessment user detection and channel estimation quality by calculating, and assessment result is broadcasted To all users, any active ues decide whether to continue to send pilot tone according to assessment result, to adaptively adjust pilot-frequency expense；? Beneficial to the scheme of adaptive pilot expense, the present invention can guarantee all any active ues reliably with access delay as small as possible Access network.
Secondly, fragmentary characteristic and extensive MIMO virtual angle domain channel of the present invention using user uplink transmission flow Sparsity is exempted from the frame structure for authorizing multiple access protocol by design multiple access pilot signal and broadband, any active ues of base station end is examined It surveys and channel estimation problems is modeled as space-frequency domain and the distributed more measurement vector compaction perception of angle-frequency domain respectively Problem；Then it utilizes proposed Turbo-DMMV-AMP algorithm to be alternately carried out any active ues detection and channel estimation, obtains Any active ues set and its channel vector.Space-frequency domain and angle-can be made full use of by having benefited from the alternate estimation mode The sparsity structure of frequency domain, pilot-frequency expense can be greatly reduced.
In addition, parameter and noise power that present invention introduces EM algorithms to be distributed real channel learn, make the algorithm It is more advantageous to the realization of real system.
With reference to the accompanying drawings and examples, the present invention will be described in detail.
The present invention considers that the typical huge dimension multiple access of uplink accesses scene in extensive mimo system, as shown in Figure 1.The system It is made of a base station and K user equipment, usual K is very big, and base station end is equipped with the uniform linear array of M root antenna and user End only considers single antenna without loss of generality.In order to support the service traffics of magnanimity to carry, using N number of subcarrier OFDM (just Hand over frequency division multiplexing, Orthogonal Frequency Division Multiplexing) transmission technology, P pilot tone is equably It is inserted into N number of subcarrier, for any active ues detection and channel estimation.For p-th of pilot sub-carrier, received in base station end K-th of user in t-th of time slot (t-th of OFDM symbol) send signal be
HereIndicate the corresponding subchannel of k-th of user, p-th of pilot sub-carrier,It is k-th of user The pilot signal for the uplink multi-address access transmitted in p-th of pilot subchannel,It is the additivity height of p-th of pilot subchannel This white noise.Scene is accessed for typical huge dimension multiple access, there was only the user of sub-fraction in given channel coherency time Equipment is active and needs multi-upstream access network.The factor of enlivening that we define user k is αk, when user enlivens 1 is taken otherwise to take 0；Meanwhile the collection for defining any active ues is combined intoThe quantity of any active ues is|·|cIndicate set element number.So, in the p pilot sub-carrier and t-th of time slot, base station, which receives, to be come Signal from all any active ues is
WhereinBy Channel Modeling AtHere ρkIt is large-scale fading caused by path fading and shadow fading,It is multipath fading.For The subchannel of p-th of pilot sub-carrier, k-th of user is
Wherein L is interference contribution quantity, βk,lIt is path complex gain,For the path delay of time,It is double-side band bandwidth.Base station Antenna-array response vector isHere It is kth The angle of arrival in a path user l, λ are to propagate wavelength,It is critical antenna interval.
Due to the fragmentary characteristic of user uplink transmission flow, only sub-fraction user is active, i.e. Ka< < K.Cause This, defines channel matrixThen channel vector corresponding to p-th of pilot sub-carrier and m root antennaBe it is sparse,
|supp{[Xp]:,m}|c=Ka< < K (4)
Moreover, the sparsity be all for all antennas it is identical,
supp{[Xp]:,1}=supp { [Xp?:,2}=...=supp { [ Xp?:,M}, (5)
Supp { } indicates the supported collection of vector or matrix, and this sparsity is called huge dimension access channel matrix by we Spatial domain structural sparse.BecauseAll be for all subchannels it is identical, thereforeIn frequency domain With common supported collection,
supp{X1}=supp { X2}=...=supp { XP} (6)
We call structural sparse united in (5) and (6)Space-frequency domain structural sparse, In order to illustrate this sparsity, it will be assumed that a total of K=10 user, wherein there is Ka=3 users be activation and And multi-upstream access network is needed, base station is equipped with M=16 root antenna, then,Structural sparse such as Fig. 2 (a) institute Show.On the other hand, base station is generally set up aloft, and the scatterer of surrounding is seldom, and user equipment is generally remotely located from base station Lower, surrounding scatterer is abundant, can be modeled with one-ring channel model to this characteristic.Consideration one is from base station distance For the user of R, it is in the circle of r, then the angle spread △ ≈ observed in base station that scatterer, which is mainly distributed on radius around it, Arctan (r/R) will be very small, because of R > > r.This will lead to the virtual angle domain channel of extensive MIMO, and there are sparsities. Specifically, the virtual angle domain channel of userWith spatial domain channelRelationship can be expressed as
Here ARIt is the transformation matrix depending on base-station antenna array geometry, when using the uniform of critical antenna interval When linear array, ARIt is a DFT matrix.Due to huge aerial array and minimum angle spread, i.e. △ < < M, channel Vector be it is sparse,
And it is this it is sparse be cluster, as shown in Fig. 2 (b).Moreover, because the space of channel passes in system bandwidth Broadcast that characteristic is closely similar, therefore the subchannel in different sub-carrier enjoys similar scatterer, causesOn frequency domain There is identical sparse pattern
We call structural sparse united in (8) and (9)Angle-frequency structure sparsity.Into One step, define virtual angle domain channel matrixWherein Consider the space-frequency domain structural sparse in (4) and (6), available | supp { [Wp]:,m}|c< < K, and
supp{W1}=supp { W2}=...=supp { WP} (10)
Fig. 2 (b) considers space-frequency domain and angle-frequency domain about the example of virtual angle domain channel matrix simultaneously Structural sparse.These above-mentioned sparse characteristics will be examined in the present invention for realizing any active ues of super reliable low time delay Survey and channel estimation.
Based on above-mentioned analysis, below with reference to Fig. 3 step by step to any active ues detection of the invention and channel estimation process It is described in detail.
Step 1, any active ues uplink send multiple access pilot signal
In each time slot, any active ues of access network is needed directly to send non-orthogonal multiple access pilot tone letter to base station Number, which is theoretical designed in advance based on DCS, and for known to base station end.System is accessed using authorization multiple access is exempted from Agreement.The present invention is further devised for broadband system to be exempted to authorize multiple access protocol.Presently, there are Multiple Access scheme examine Consider the huge location communication in narrowband of single carrier transmission, the huge connection that the service traffics of magnanimity can not be supported to carry, to huge in broadband system The reliable of dimension multiple access access supports it is the problem for having more challenge.In order to solve the problem and utilize one in broadband system A little characteristics, the present invention have carried out following steps.
Step 1.1, design broadband exempt to authorize the frame structure of multiple access protocol
For extensive MIMO-OFDM system, broadband designed by the present invention exempts to authorize the frame structure packet of multiple access protocol Containing two parts of time domain and frequency domain, as shown in Figure 4.It is made of on frequency domain N number of subcarrier, is made of in time domain T time slot.Preceding G A time slot is used for while sending pilot tone and data, and in N number of subcarrier, for P pilot sub-carrier for sending pilot tone, (N-P) is a Data subcarrier is for sending data.Then (T-G) a time slot is only used for sending data, therefore all N number of carrier waves are data Carrier wave.Since N number of subcarrier in preceding G time slot can carry pilot tone and data simultaneously, user can be without base station Authorization sends pilot tone and data simultaneously, and after base station carries out any active ues detection and channel estimation with pilot tone, logarithm may be implemented According to decoding, this frame structure can to avoid complexity scheduling, substantially reduce access delay.
Meanwhile time slot expense G used in pilot tone is to be commented according to base station any active ues detection and channel estimation quality Estimate what result adaptively adjusted.Adjustment process is referring to step 3.
Step 1.2, design broadband multiple access pilot signal
All designed multiple access pilot signals of userBy when base station end is modeled as the observation of compressed sensing Matrix.And observing matrix is particularly significant to compressed sensing reconstruction property, it is therefore desirable to be designed to multiple access pilot tone.In order to meet RIP (limited equidistant characteristics, Restricted Isometry Property) condition of observing matrix, for p-th of pilot tone Carrier wave, the pilot tone of all users sent in multiple time slots is distributed by independent identically distributed multiple Gauss to be generated.Meanwhile it is different Pilot tone on pilot sub-carrier must be different, to introduce the diversity between subcarrier, i.e. observing matrixFor different P is different.DCS theory points out that this species diversity can be further improved compressed sensing reconstruction property.
Step 2, any active ues detection and channel estimation
This step is according to initial slot expense G0, any active ues that collection base station end receives are in G0It is sent in a time slot Uplink signal carries out any active ues detection and channel estimation according to received signal and the pilot tone of known all users, Obtain any active ues set and corresponding extensive mimo channel.
The prior art is using Space category model while to carry out any active ues detection and channel estimation.And the present invention is in base station Any active ues detection and channel estimation problems are modeled as space-frequency domain respectively and angle-frequency domain DMMV is (distributed more Measure vector, Distributed Multiple Measurement Vector) compressed sensing problem.Then, according to space- Frequency domain model obtains any active ues set, and (according to the set, can further be obtained by the different confidence thresholding of setting can By spending two different any active ues subsets；One of subset user activity reliability is high, referring to step 2.3 to modules A Description)；Estimate any active ues set substitution angle-frequency domain model to obtain corresponding channel vector, according to the letter of estimation Road subtracts reception signal corresponding to the high any active ues of reliability from receiving, obtains and receive signal residual error, and should in signal Residual error substitutes into space-frequency domain model and estimates remaining any active ues, so that any active ues detection part needed to detect Any active ues are fewer and fewer；And so on, by being alternately carried out any active ues detection and channel estimation, obtain any active ues collection Conjunction and its channel vector.
Step 2.1, any active ues detection and channel estimation modeling
In base station end, the signal received in G continuous slot can be write as
Wherein, For exempting to authorize random division multiple accessing, in order to avoid complicated scheduling and caused access delay, base Needs stand according to reception signalAnd pilot matrixRapidly and accurately estimate any active ues setAnd its corresponding letter Road vectorThe problem, which can be equivalent to, estimates channel matrix based on model (11)In order to solve The above problem, traditional estimation method need G >=K such as MMSE linear estimator, moreover,It must be a unitary matrice.This It means that and needs to distribute orthogonal pilot tone to user, facilitate the user that identification of base stations is different, and the number of users in huge connection Huge, channel coherency time is limited, can not distribute orthogonal pilot tone for all users, i.e. G >=K cannot achieve.In order to solve this Problem, the present invention utilizeSpace-frequency domain structural sparse, in G < < K, (11) become a sky Between domain DMMV compressed sensing problem.So, it is known that receive signalAnd pilot matrixEstimate XpSupported collection, estimated Any active ues setTo realize that any active ues detect.
While angle-frequency structure sparsity in order to further utilize up channel matrix, according to extensive Mimo channel relationship (7) between spatial domain and virtual angle domain, Space category model (11) can further be write as
Wherein,It is virtual angle domain channel matrix.The model is virtual The corresponding DMMV compressed sensing problem of angle domain.Known reception signalIt can be with by the transformation relation of spatial domain and angle domain Obtain equivalent reception signalIt utilizesPilot matrixAnd any active ues set of estimationIt can estimate To the corresponding virtual angle domain channel of any active uesAgain by the channel conversion back to spatial domain, thus real The channel estimation of existing any active ues.
Because model (12) simultaneously consider up channel matrix space-frequency domain and angle-frequency structureization it is dilute Property is dredged, with the channel X in Space category model (11)pIt compares, channel WpIt is more sparse, but observed between different antennae Common sparsity is destroyed.Presently, there are compressed sensing based huge dimension Multiple Access scheme only consider XpSpatial domain structuring it is dilute Property is dredged, and utilizes (11) to realize and comes united any active ues detection and channel estimation.In order to make full use of XpThe sparse knot of enhancing Structure and WpIncome brought by the sparsity of enhancing, the present invention are carried out any active ues detection using (11) and are carried out pair using (12) The channel estimation answered.Specifically, the present invention improves AMP-NNSPL algorithm first, proposes to be used for Combined estimator sparse 3 D matrix DMMV-AMP algorithm, which is (11) and (12) expressed model out, can be used for solving the problems, such as compressed sensing (11) or (12) realize any active ues simultaneously and channel estimation (for AMP-NNSPL algorithm, referring specifically to document " translated name: benefit Sparse extensive MIMO-OFDM channel estimation is realized with approximate common supported collection ", author, English name and source are " Lin X,Wu S,et al.Estimation of sparse massive MIMO-OFDM channels with approximately common support[J].IEEE Communication Letters,2017,21(5):1179- 1182.").Later, it is based on DMMV-AMP algorithm, the invention proposes the schemes of alternately any active ues and channel estimation, claim For Turbo-DMMV-AMP algorithm.As the enhancing version of DMMV-AMP, Turbo-DMMV-AMP algorithm is by being alternately carried out Any active ues and channel estimation, to improve performance.
Step 2.2, DMMV-AMP algorithm
The present invention carries out any active ues detection using space-frequency domain model, and is carried out using angle-frequency domain model Channel estimation is all made of improved AMP algorithm and resolves, and improved AMP algorithm learns hyper parameter using EM algorithm, and according to channel Matrix (refers to spatial domain channel XpOr angle domain channel Wp) sparsity structure optimize hyper parameter.
In order to describe conveniently, the DMMV-AMP algorithm in the present invention is introduced by taking model (11) as an example here.We are fixed first Adopted xp,k,m=[Xp]k,m, then XpMinimum Mean Squared Error estimation, i.e. Posterior Mean can be expressed as
For simplified expression, we have ignored x herep,k,m, XpAndIn subscript index p and subscript G, wherein Edge posterior probability can be write as
p(xk,m| Y) and=∫ p (X | Y) dX\k,m, (14)
Here X\k,mIndicate setAnd combines posterior probability and calculated by bayes formula
WhereinIt is normalized constant factor.Under the hypothesis of white Gaussian noise, likelihood Function is
Here σ is noise power.In the present invention, it will be assumed that a flexible needle plate prior model,
The model can be matched with the true distribution of channel matrix well, here 0 < γk,m< 1 indicates sparse rate, i.e. xk,m Probability not equal to 0, δ () indicate dirichlet function, f (xk,m) be nonzero element in channel matrix distribution.
Since number of users is huge in the access of huge dimension multiple access, and it is directly proportional to number of users to further relate to dimension in formula (14) Multidimensional integral, therefore above-mentioned estimator is difficult to realize.In order to solve above-mentioned challenge, the main think of of DMMV-AMP algorithm in the present invention Think it is to obtain edge posterior probability p (x using AMP algorithmk,m| Y) approximate solution.Specifically, the algorithm is by following five part group At:
1, factor nodes information updating
The present invention indicates the factor relationship in formula (15) with factor graph, after AMP algorithm approximation, in factor section The message that point updatesWithIt can be expressed as
Here, i indicates algorithm iteration number,It is the posterior variance in formula (28), it is involved in the first iteration Be initialized as
2, variable node message updates
Accordingly, it can be expressed as in the message that variable node updates
3, Posterior Mean and posterior variance calculate
According to the update of factor nodes and variable node as a result, xk,mPosterior probability distribution can be approximated to be
In the present invention, we further assume that channel gain all obeys multiple Gauss distribution, i.e. f (xk,m)=CN (xk,m；μ, τ), then the posterior probability in formula (22) can be further simplified into
Wherein,
Finally, xk,mPosterior Mean and posterior variance can calculate separately for
Here, we are intermediate variableReferred to as confidence factor, when DMMV-AMP obtains reliable compressed sensing reconstruct When, we are just like drawing a conclusion
Therefore, which can serve to indicate that whether (k, m) a element of channel matrix is zero.
4, learn hyper parameter using EM algorithm
Above-mentioned formula (18)-(28) are the primary iteration step of DMMV-AMP algorithm, but need known real channel The design parameter and noise power of distribution, this be also currently based on AMP algorithm huge dimension Multiple Access scheme there are the drawbacks of.For side Just application in systems in practice, the present invention is using EM algorithm to hyper parameterLearnt, EM is calculated Method is made of two steps:
In formula (30),It indicates in known Y and θiIn the case of conditional expectation, i.e., this be contemplated to be about Joint posterior probability p (X | Y；θi).There are two hang-ups in above-mentioned EM algorithm: accessing in huge dimension multiple access in actual application In, p (X | Y；θi) computation complexity be difficult to accept；It is extremely complex to the combined optimization of four parameters in θ.In order to solve Above-mentioned problem, the present invention are obtained by AMP algorithmMeanwhile to joining in θ Several updates, we only update a parameter every time, and other parameters are fixed and are considered as constant.To formula (30) about spy Fixed parameter derivation and it is inverted be zero, the update rule of available hyper parameter is,
5, hyper parameter is optimized according to the sparsity structure of channel matrix and updates rule
It is worth noting that, sparse rate γp,k,mIndicate channel matrix XpOr Wp(k, m) a element non-zero probability. To γ in formula (34)p,k,mUpdate all p, k, m are independently carried out, do not account for the knot of up channel matrix Structure sparsity.From here, for convenience of describing, we rejoin the index of pilot sub-carrier.Since up channel matrix is deposited In structural sparse, therefore the corresponding sparse rate of Partial Elements is highly relevant, rather than is completely independent.For this purpose, the present invention draws Enter an index setTo correct the update rule of sparse rate, it is believed that be located at setIn channel element it is dilute The rate of dredging and xp,k,mIt is identical.
For DMMV compressed sensing model (11), the space-frequency domain structural sparse of channel matrix only considered, such as Formula (6) and Fig. 2 (a) are shown.In order to utilize this sparsity, the present invention proposes all channel members corresponding to the same user Element sparse rate having the same, i.e., for specific k, γp,k,mIt is all identical for arbitrary p and m.Therefore, by DMMV- When AMP algorithm is applied to model (10), the update rule of sparse rate is modified to
Wherein
In contrast, DMMV compressed sensing model (12) considers the space-frequency domain and angle-frequency of channel matrix simultaneously Rate domain structure sparsity, as shown in Fig. 2 (b).According to the characteristic that the extensive mimo channel cluster of virtual angle domain is sparse, Yi Jixin Road matrix WpStructural sparse, as shown in formula (8)-(10), it is considered herein that channel matrix element wp,k,mNeighbouring member Plain non-zero, then the element maximum probability is also non-zero, and otherwise, which is zero.Therefore, wp,k,mThere is phase with its neighbours Same sparse rate.The neighbours for defining the element are
When DMMV-AMP algorithm being then applied to model (12), the update rule of sparse rate is modified to
Step 2.3, Turbo-DMMV-AMP algorithm
Channel matrix in order to obtainEstimation, be defined asDirectly DMMV-AMP algorithm can be answered For model (11).Alternatively, DMMV-AMP algorithm is applied to model (12), estimation is obtainedPass through formula (7) again In transformation obtainIt is obtainingAfterwards, any active ues set and its channel vector can be obtained simultaneously, i.e.,Supported collection and corresponding element value.But above two method cannot all make full use of the sparsity structure of channel matrix. The present invention alternately utilizes model (11) and (12) to be iterated using two modules, is lived respectively based on DMMV-AMP algorithm Jump user's detection and channel estimation, to further decrease pilot-frequency expense, and this alternate estimation mode is named as Turbo-DMMV-AMP algorithm.The algorithm is made of following two module:
1, modules A: any active ues detector based on DMMV-AMP algorithm
In Turbo iteration each time, DMMV-AMP algorithm is applied to Space category model by modules A, obtains confidence factor(j refers to jth time Turbo iteration) obtains the different any active ues set of two reliabilities according to confidence factor, and by this Two set give module B as message transmission.
Specifically, in first time Turbo iteration (j=1), DMMV-AMP algorithm is applied to model (11) by modules A, is obtained The corresponding confidence factor of channel element.Because working as xp,k,mConfidence factor when ≠ 0Take 1, otherwise take 0, still can basis This characteristic of confidence factor designs any active ues detector.A threshold function r (x is defined first；ε), as | x | r (x when > ε； ε)=1, otherwise r (x；ε)=0.So, the present invention designs following any active ues detector according to the confidence factor of acquisition,
Here pthIt is set as 0.9, and 0 < ε < 1 is rationally arranged according to different demands, it is general that lower ε will lead to higher false-alarm Rate, and higher ε will lead to higher false dismissal probability.Later, the present invention obtains two reliably by the way that different thresholding ε is arranged Property different any active ues set: utilize thresholding εdet=0.4 obtains coarse any active ues set omega, utilizes thresholding εrel= 0.9 obtains reliable any active ues set Ξj, as shown by the following formula:
WhereinIt is to calculate gained according to formula (39),It can be found that ΞjIt is the subset of Ω.Then, modules A The two set are given to module B as message transmission.
2, module B: the channel estimator based on DMMV-AMP algorithm
Using the coarse any active ues set omega from modules A, module B is based on model (12) to the letter of user in the set Estimated that specifically, we should solve the problems, such as following compressed sensing in road
Wherein,WithIt is respectivelyAnd WpSubmatrix,AndIt is the set of all users.The channel estimator of this module Only consider the user in set omega, therefore channel matrix dimension corresponding to user uplink channel estimation reduces.Simultaneously as The structural sparse of extensive mimo channel angle-frequency domain, low-dimensional channel matrix [Wp]Ω,: it is still sparse.The present invention Mentioned DMMV-AMP algorithm is utilized to solve the problems, such as compressed sensing (41), estimates low-dimensional channel matrix, it is reliable to remove part Reception signal corresponding to any active ues comes from Ξj-1The reception signal of partial user, and signal residual error will be received As message transmission to modules A.It is calculated by following formula
Wherein, in set Γ element from set ΞjIn randomly select, and | Γ |c/|Ξj|c=0.8, one is only removed here Signal by any active ues is to avoid algorithm diverging.
In next Turbo iteration (j > 1), any active ues test problems in modules A become
Wherein,It is the reception signal residual error of jth time Turbo iteration,It is channel matrix Residual error,It is defined asAndHereIt is last Turbo Channel matrix estimated by iteration module B.The present invention executes two module iteration by alternate mode, because with iteration Progress,Become more and more sparse, while the channel of user is also constantly reevaluated in set omega, therefore this friendship Pilot-frequency expense is greatly reduced in the mode replaced.When iteration ends, any active ues set of estimation can be obtainedAnd Its corresponding channel vector
Step 3, the detection of assessment any active ues and channel estimation quality
This step estimates obtained any active ues set and respective channels using step 2, reconstructs channel matrix, and calculate and connect Receive signal errors；Any active ues detection and channel estimation quality are assessed according to signal errors is received, and assessment result is broadcast to All users, if the result meets standard given in advance, any active ues stop sending pilot tone, in time slot next Data are only sent, to obtain reliable any active ues set and its channel；Otherwise, any active ues continue to send pilot tone, base station The end reception signals for collecting a time slot re-start any active ues detection and channel estimation more, and re-execute the steps 3, until Collected reception signal is enough to obtain reliable any active ues set and its channel.
This step is specifically divided into three steps:
Uplink multi-address channel matrix is estimated in step 3.1, reconstruction
Using any active ues set and respective channels of estimation, the uplink multi-address channel matrix of estimation is rebuildSpecifically Process are as follows:And
Step 3.2 calculates the evaluated error for receiving signal
The evaluated error for receiving signal, process are calculated using the channel matrix of estimation are as follows:
Step 3.3, assessment estimation quality
According to the evaluated error assessment any active ues detection and channel estimation quality for receiving signal, and assessment result is broadcasted To all users.IfIt is unreliable then to estimate, any active ues continue to send pilot tone, and base station is more The reception signal for collecting a time slot re-starts any active ues detection and channel estimation, and repeats step 3, reliable until obtaining Any active ues set and channel estimation.Here ∈ is the standard given in advance for assessing estimation quality.
It is any active ues detection and the channel estimation methods of adaptive expense disclosed by the invention above.
With presently, there are based on exempt from authorization huge dimension multiple access schemes compared to (here consider be based on GSP algorithm, SOMP Algorithm, the joint any active ues and channel estimation methods of DSAMP algorithm, these schemes only consider spatial domain structural sparse), The advantage on access delay is being reduced in order to embody the present invention, is illustrating effect of the invention with Fig. 5~Fig. 9 here.Meanwhile being It embodies and utilizes virtual angle domain sparsity, any active ues detection and channel estimation estimation mode alternately and adaptive Performance gain brought by pilot-frequency expense scheme is answered, in addition proposes three kinds of comparison schemes here, is summarized are as follows:
(1) any active ues detection and channel estimation based on spatial domain DMMV-AMP algorithm: by DMMV-AMP algorithm application In spatial domain compressed sensing model (11), the estimation of spatial domain channel matrix is obtainedAnd its confidence factor, and utilize public affairs Detector shown in formula (31) carries out any active ues detection.Here channel square caused by the fragmentary characteristic of user uplink flow is only utilized Battle array space-frequency domain structural sparse.
(2) any active ues detection and channel estimation based on angle domain DMMV-AMP algorithm: by DMMV-AMP algorithm application In angle domain compressed sensing model (12), the estimation of angle domain channel matrix is obtainedAnd its confidence factor, and utilize public affairs Detector shown in formula (31) carries out any active ues detection.Here the space-frequency domain and angle-frequency of channel matrix are utilized simultaneously Rate domain structure sparsity.
(3) any active ues detection and channel estimation based on Turbo-DMMV-AMP algorithm: this programme joint utilizes model (11) and (12) it, is utilized respectively spatial frequency domain and angle-frequency structure sparsity is alternately carried out any active ues detection And channel estimation.This scheme is with unique difference of the invention, and the pilot time slot expense in this scheme is fixed, and this hair Pilot time slot expense in bright is adaptive according to any active ues detection and channel estimation quality assessment result.
Fig. 5 compared enlivening for scheme based on DMMV-AMP algorithm and other three kinds of traditional schemes based on Space category model User's detection performance, this performance pass through detection error probability PeIt measures, is defined as
From fig. 5, it can be seen that the scheme based on DMMV-AMP algorithm is wanted significantly better than other three kinds of traditional schemes.Cause This.When considering identical performance, DMMV-AMP algorithm can reduce access delay significantly, for example, for detecting error probability Less than 10-5In the case of, the best GSP algorithm of performance at least needs the pilot-frequency expense of G=72 in traditional scheme, and DMMV-AMP is calculated Method only needs the expense of G=58, it means that the algorithm can be reduced close to 19% access delay.Moreover, by matching in base station Standby more antennas use bigger(hereRefer to the pilot sub-carrier number for any active ues detection and channel estimation Amount), DMMV-AMP algorithm can obtain better performance, this is because bigger antenna for base station number orChannel square can be enhanced The structural sparse of battle array.Fig. 6 compared the MSE (mean square error, Mean Square Error) of channel estimation, furtherly Superiority of the DMMV-AMP algorithm in terms of improving channel estimating performance is illustrated.
For the present invention and three kinds of comparison schemes being mentioned, Fig. 7 and Fig. 8 compared respectively they any active ues detection and Channel estimating performance.As can be seen that the performance based on angle domain model (12) scheme can be significantly better than in pilot-frequency expense G > 18 Based on the performance of angle domain model (11) scheme, and it is opposite in G < 18.This is because space-is considered in model (12) simultaneously Frequency domain and angle-frequency domain structural sparse, therefore corresponding channel matrixThan in model (11)It is more sparse, but the sparsity structure in different antennae is destroyed.So being obtained in (12) reliableReconstruct institute The minimum observation (i.e. pilot-frequency expense) needed will be much smaller than model (11).And in observation deficiency situation, (11) believe in model Structural sparse can reconstruct to compressed sensing of road matrix on different antennas brings performance gain, therefore the base in low overhead Better performance can be obtained in Space category model (11).Above-mentioned conclusion explanation, using only Space category model or angle domain mould Type cannot all make full use of space-frequency domain and angle-frequency domain structural sparse, and be based on Turbo-DMMV-AMP The scheme of algorithm combines and two kinds of models is utilized, from Fig. 7 and Fig. 8 as can be seen that within the scope of all time slot expenses, the party Any active ues detection of case and channel estimating performance will be better than the schemes using only spatial domain or angle domain model.In addition Ground compares scheme comparison with three of the above, and any active ues detection of adaptive expense of the invention and channel estimation scheme can be certainly Adjustment pilot time slot expense is adapted to, guarantees reliable any active ues detection and channel estimating performance.The present invention is labelled in Fig. 8 not With ratio shared by pilot-frequency expense.
Fig. 9 compared the present invention and based on the non-adaptive Multiple Access scheme of Turbo-DMMV-AMP algorithm in different any active ues User's detection and channel estimating performance under several, and be labelled with average pilot time slot required for the present invention in each case and open Pin, it can be seen that with the increase of any active ues quantity, the detection error probability and channel estimation MSE of non-adaptive scheme are all It will rise, and the present invention can adaptively adjust pilot-frequency expense, guarantee that all any active ues can be correctly detected out, and Reliable channel estimation is obtained, embodies robustness of the present invention under practical any active ues quantity time-varying field scape, it is ensured that user Network can be accessed fast and reliablely.
In conclusion the above is merely preferred embodiments of the present invention, being not intended to limit the scope of the present invention. All within the spirits and principles of the present invention, any modification, equivalent replacement, improvement and so on should be included in of the invention Within protection scope.

Claims (7)

1. a kind of any active ues of adaptive expense detect and channel estimation methods characterized by comprising
Step 1, in each time slot, any active ues send signal using the multiple access protocol uplink for exempting from authorization；
Step 2, according to initial slot expense G0, any active ues that collection base station end receives are in G0The uplink sent in a time slot Pilot signal carries out any active ues detection and channel estimation using received signal, obtains any active ues set and correspondence Extensive mimo channel；
Step 3, any active ues set and respective channels obtained using estimation, reconstruct channel matrix, and calculate and receive signal mistake Difference；Any active ues detection and channel estimation quality are assessed according to signal errors is received, and assessment result is broadcast to all use Family, if the result meets standard given in advance, any active ues stop sending pilot tone, only send number in time slot next According to obtain reliable any active ues set and its channel；Otherwise, any active ues continue to send pilot tone, and base station end is collected more The reception signal of one time slot re-starts any active ues detection and channel estimation, and re-execute the steps 3, until collected Signal is received to be enough to obtain reliable any active ues set and its channel.
2. the method as described in claim 1, which is characterized in that the multiple access protocol for exempting from authorization used by step 1 is corresponding Frame structure include two parts of time domain and frequency domain；It is made of on frequency domain N number of subcarrier, is made of in time domain T time slot；Preceding G A time slot is used for while sending pilot tone and data, and in N number of subcarrier of this G time slot, and P pilot sub-carrier is for sending Pilot tone, (N-P) a data subcarrier is for sending data；Then (T-G) a time slot is only used for sending data, therefore all N number of loads Wave is data subcarrier.
3. method according to claim 2, which is characterized in that pilot signal transmitted by any active ues is non-orthogonal pilot letter Number, for each pilot sub-carrier, pilot signal of all users in pilot-frequency expense passes through the multiple height of independent identically distributed standard This is generated；Pilot tone on different pilot sub-carriers is different.
4. method as claimed in claim 3, which is characterized in that any active ues of progress described in step 2 detect and channel estimation Are as follows:
The sub-channel matrix of mimo channel vector composition based on all users has column sparsity, observes in different antennae Sparsity is identical, and sub-channel matrix sparse pattern having the same corresponding to different pilot sub-carriers, thus will Any active ues test problems are modeled as spatial domain compressed sensing problem, constitute space-frequency domain model；
Based on extensive mimo channel, in virtual angle domain, there are cluster sparsities and different sub-channel matrix on frequency domain Sparse pattern having the same, so that the channel estimation problems of any active ues to be modeled as to the compressed sensing problem of angle domain, structure Angled-frequency domain model；
Any active ues set is obtained using space-frequency domain model, any active ues are further obtained by different confidence thresholdings Two different any active ues subsets of reliability in set；Any active ues set is substituted into angle-frequency domain model to carry out Channel estimation, and the corresponding signal of user in the high any active ues subset of reliability is subtracted from receiving in signal, it obtains receiving letter Number residual error, then the residual error is substituted into space-frequency domain model and carries out remaining any active ues detection, and so on, alternately into The detection of row any active ues and channel estimation, until reach given maximum number of iterations or receive signal residual error be less than in advance to Fixed thresholding.
5. method as claimed in claim 4, which is characterized in that in the step 2, the modeling process of space-frequency domain are as follows:
For p-th of pilot sub-carrier, k-th of user that base station receives is defined in the signal of t-th of time slot are as follows:
WhereinFor p-th of pilot subchannel of k-th of user,It is that k-th of user believes in p-th of pilot tone The uplink multi-address pilot tone transmitted on road,It is the white Gaussian noise of p-th of pilot subchannel；Base station is equipped with extensive antenna, M It is antenna for base station number；User uses single antenna；Total P pilot sub-carrier；
The factor of enlivening for further defining k-th of user is αk, 1 is taken when user enlivens, and 0 is taken when silent, then, base station end is total Receiving signal can be write as:
Wherein(·)TFor transposition symbol；K is Total number of users；
So in continuous G time slot, the reception signal that base station end is collected into can be write as:
Wherein It is in spatial domain Row access channel matrix,As it is desirable that G < < K, while channel matrix has column sparsity, different lines Between have common supported collection, therefore the model be space-frequency domain compressed sensing model；Known reception signalWith pilot tone square Battle arrayEstimate XpSupported collection, any active ues set estimatedTo realize that any active ues detect.
6. method as claimed in claim 5, which is characterized in that in the step 2, angle-frequency domain modeling process are as follows:
Consider channel transforming from a spatial domain to virtual angle domain, transformation relation are as follows:
Wherein wp,kIt is virtual angle domain channel；It is transformation matrix, in the homogenous linear that antenna spacing is half wavelength It is discrete fourier matrix, () under array caseHFor conjugate transposition symbol；Therefore, available angle-frequency domain model:
Wherein,It is the access channel matrix of angle domain,(·)* It is conjugate of symbol；The model is used for channel estimation；Known reception signalIt can by the transformation relation of spatial domain and angle domain To obtain equivalent reception signalIt utilizesPilot matrixAnd any active ues set of estimationIt can estimate Obtain the corresponding virtual angle domain channel of any active uesAgain by the channel conversion back to spatial domain, thus Realize the channel estimation of any active ues.
7. such as method described in claim 5 or 6, which is characterized in that carry out any active ues inspection using space-frequency domain model It surveys, and channel estimation is carried out using angle-frequency domain model and is all made of improved AMP algorithm resolving, improved AMP algorithm benefit Learn hyper parameter with EM algorithm, and hyper parameter is optimized according to the sparsity structure of channel matrix and updates rule；Wherein,
When improved AMP algorithm is applied to space-frequency domain model, sparse rateUpdate rule are as follows:
In formula (I), sparse rateIndicate that the probability of channel matrix (p, k, m) a element non-zero, p are the rope of pilot sub-carrier Draw, k is user index, and m is the index of antenna for base station, and i represents the i-th iteration of algorithm, is gatheredHere (q, l, u) indicates the element index of three dimensional channel matrix, confidence The factorIt is intermediate variable defined in AMP algorithm；|·|cIndicate set element number；
When improved AMP algorithm is applied to angle-frequency domain model, the update rule of sparse rate are as follows:
In formula (II), set
CN201910345235.7A 2019-04-26 2019-04-26 Active user detection and channel estimation method with adaptive overhead Active CN110071881B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201910345235.7A CN110071881B (en) 2019-04-26 2019-04-26 Active user detection and channel estimation method with adaptive overhead

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201910345235.7A CN110071881B (en) 2019-04-26 2019-04-26 Active user detection and channel estimation method with adaptive overhead

Publications (2)

Publication Number Publication Date
CN110071881A true CN110071881A (en) 2019-07-30
CN110071881B CN110071881B (en) 2020-12-15

Family

ID=67369075

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201910345235.7A Active CN110071881B (en) 2019-04-26 2019-04-26 Active user detection and channel estimation method with adaptive overhead

Country Status (1)

Country Link
CN (1) CN110071881B (en)

Cited By (25)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110784286A (en) * 2019-11-01 2020-02-11 重庆邮电大学 Multi-user detection method of non-orthogonal multiple access system based on compressed sensing
CN111711584A (en) * 2020-05-22 2020-09-25 西北工业大学 Sparse underwater acoustic channel estimation method based on proportional normalization minimum mean square error
CN111865845A (en) * 2020-06-04 2020-10-30 国网江西省电力有限公司供电服务管理中心 MIMO user detection and channel estimation device and method
CN111901889A (en) * 2020-07-13 2020-11-06 北京科技大学 Distributed uplink multi-user random access method
CN112260730A (en) * 2020-10-15 2021-01-22 东南大学 C-V2V broadband large-scale MIMO pilot frequency multiplexing channel acquisition method
CN112291176A (en) * 2020-11-09 2021-01-29 安徽大学 Dynamic multi-user detection method and dynamic multi-user detection device thereof
CN112564749A (en) * 2020-11-26 2021-03-26 北京邮电大学 Channel reconstruction method, device and equipment of non-crohnok structure
CN112887233A (en) * 2021-01-21 2021-06-01 中国科学技术大学 Sparse Bayesian learning channel estimation method based on 2-dimensional cluster structure
CN112929307A (en) * 2021-02-26 2021-06-08 广东工业大学 Distributed channel estimation method under authorization-free communication
CN113067668A (en) * 2021-02-04 2021-07-02 宁波大学 Method for performing activity and signal joint detection aiming at MMV model
CN113194499A (en) * 2021-04-02 2021-07-30 西安交通大学 Authorization-free self-adaptive transmission method based on user estimation
CN113438746A (en) * 2021-08-27 2021-09-24 香港中文大学（深圳） Large-scale random access method based on energy modulation
TWI741619B (en) * 2020-05-26 2021-10-01 國立中山大學 Method of parameter estimation for a mimo system based on deep learning
CN114024638A (en) * 2021-11-04 2022-02-08 哈尔滨工业大学 Uplink authorization-free system receiving method based on spatial modulation
CN114039639A (en) * 2021-11-04 2022-02-11 电子科技大学 Combined channel estimation and user detection method in large connection scene
CN114285523A (en) * 2021-12-24 2022-04-05 西安交通大学 Large-scale MTC authorization-free multi-user detection method and system facing multi-service requirements
CN114500322A (en) * 2021-12-31 2022-05-13 上海交通大学 Method for equipment activity detection and channel estimation under large-scale authorization-free access scene
CN114520700A (en) * 2022-01-13 2022-05-20 重庆邮电大学 Low-complexity signal detection method in machine type communication under low-resolution ADC
CN114584437A (en) * 2022-05-06 2022-06-03 北京理工大学 Semi-blind message detection method and device of single handshake protocol
CN114826480A (en) * 2022-03-11 2022-07-29 南京邮电大学 Signal detection method and device of non-orthogonal multiple access system in large frequency offset environment
CN115002929A (en) * 2022-05-16 2022-09-02 西南交通大学 Active user detection method based on joint estimation of preamble and arrival angle
CN115865296A (en) * 2022-11-23 2023-03-28 东南大学 Orthogonal pilot frequency sequence active detection method based on covariance
WO2023060370A1 (en) * 2021-10-11 2023-04-20 Nokia Shanghai Bell Co., Ltd. Methods, devices, and computer readable media for adaptive user equipment device activity detection
CN117915397A (en) * 2024-03-20 2024-04-19 银河航天(西安)科技有限公司 Method, device and storage medium for transmitting telemetry transport frame based on virtual channel
CN115865296B (en) * 2022-11-23 2024-07-02 东南大学 Orthogonal pilot sequence active detection method based on covariance

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104113396A (en) * 2013-04-17 2014-10-22 上海贝尔股份有限公司 Channel detecting method and device in large-scale antenna system
CN106789770A (en) * 2016-12-26 2017-05-31 电子科技大学 The nonopiate multiple-user access methods of extensive MIMO based on compressed sensing
CN107743103A (en) * 2017-10-26 2018-02-27 北京交通大学 The multinode access detection of MMTC systems based on deep learning and channel estimation methods
CN109327850A (en) * 2018-11-16 2019-02-12 安徽大学 Multi-user detection method of non-orthogonal multiple access system based on gradient tracking and multi-step quasi-Newton method technology

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104113396A (en) * 2013-04-17 2014-10-22 上海贝尔股份有限公司 Channel detecting method and device in large-scale antenna system
CN106789770A (en) * 2016-12-26 2017-05-31 电子科技大学 The nonopiate multiple-user access methods of extensive MIMO based on compressed sensing
CN107743103A (en) * 2017-10-26 2018-02-27 北京交通大学 The multinode access detection of MMTC systems based on deep learning and channel estimation methods
CN109327850A (en) * 2018-11-16 2019-02-12 安徽大学 Multi-user detection method of non-orthogonal multiple access system based on gradient tracking and multi-step quasi-Newton method technology

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
XIAO XU 等: ""Active User Detection and Channel Estimation in Uplink CRAN Systems"", 《IEEE ICC 2015 - WIRELESS COMMUNICATIONS SYMPOSIUM》 *
ZHEN GAO 等: ""Compressive Sensing Techniques for Next-Generation Wireless Communications"", 《IEEE WIRELESS COMMUNICATIONS JUNE 2018》 *

Cited By (39)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110784286A (en) * 2019-11-01 2020-02-11 重庆邮电大学 Multi-user detection method of non-orthogonal multiple access system based on compressed sensing
CN110784286B (en) * 2019-11-01 2022-05-03 重庆邮电大学 Multi-user detection method of non-orthogonal multiple access system based on compressed sensing
CN111711584B (en) * 2020-05-22 2022-06-07 西北工业大学 Sparse underwater acoustic channel estimation method based on proportional normalization minimum mean square error
CN111711584A (en) * 2020-05-22 2020-09-25 西北工业大学 Sparse underwater acoustic channel estimation method based on proportional normalization minimum mean square error
TWI741619B (en) * 2020-05-26 2021-10-01 國立中山大學 Method of parameter estimation for a mimo system based on deep learning
CN111865845A (en) * 2020-06-04 2020-10-30 国网江西省电力有限公司供电服务管理中心 MIMO user detection and channel estimation device and method
CN111901889A (en) * 2020-07-13 2020-11-06 北京科技大学 Distributed uplink multi-user random access method
CN111901889B (en) * 2020-07-13 2021-09-07 北京科技大学 Distributed uplink multi-user random access method
CN112260730B (en) * 2020-10-15 2021-09-24 东南大学 C-V2V broadband large-scale MIMO pilot frequency multiplexing channel acquisition method
CN112260730A (en) * 2020-10-15 2021-01-22 东南大学 C-V2V broadband large-scale MIMO pilot frequency multiplexing channel acquisition method
CN112291176B (en) * 2020-11-09 2022-03-18 安徽大学 Dynamic multi-user detection method and dynamic multi-user detection device thereof
CN112291176A (en) * 2020-11-09 2021-01-29 安徽大学 Dynamic multi-user detection method and dynamic multi-user detection device thereof
CN112564749A (en) * 2020-11-26 2021-03-26 北京邮电大学 Channel reconstruction method, device and equipment of non-crohnok structure
CN112887233A (en) * 2021-01-21 2021-06-01 中国科学技术大学 Sparse Bayesian learning channel estimation method based on 2-dimensional cluster structure
CN113067668A (en) * 2021-02-04 2021-07-02 宁波大学 Method for performing activity and signal joint detection aiming at MMV model
CN113067668B (en) * 2021-02-04 2022-05-20 宁波大学 Method for performing activity and signal joint detection aiming at MMV model
CN112929307A (en) * 2021-02-26 2021-06-08 广东工业大学 Distributed channel estimation method under authorization-free communication
CN113194499A (en) * 2021-04-02 2021-07-30 西安交通大学 Authorization-free self-adaptive transmission method based on user estimation
CN113438746A (en) * 2021-08-27 2021-09-24 香港中文大学（深圳） Large-scale random access method based on energy modulation
WO2023060370A1 (en) * 2021-10-11 2023-04-20 Nokia Shanghai Bell Co., Ltd. Methods, devices, and computer readable media for adaptive user equipment device activity detection
CN114039639A (en) * 2021-11-04 2022-02-11 电子科技大学 Combined channel estimation and user detection method in large connection scene
CN114024638A (en) * 2021-11-04 2022-02-08 哈尔滨工业大学 Uplink authorization-free system receiving method based on spatial modulation
CN114039639B (en) * 2021-11-04 2023-04-25 电子科技大学 Combined channel estimation and user detection method under large connection scene
CN114285523A (en) * 2021-12-24 2022-04-05 西安交通大学 Large-scale MTC authorization-free multi-user detection method and system facing multi-service requirements
CN114285523B (en) * 2021-12-24 2023-09-15 西安交通大学 Multi-service-demand-oriented large-scale MTC (machine type communication) unlicensed multi-user detection method and system
CN114500322A (en) * 2021-12-31 2022-05-13 上海交通大学 Method for equipment activity detection and channel estimation under large-scale authorization-free access scene
CN114500322B (en) * 2021-12-31 2023-08-25 上海交通大学 Method for detecting device activity and estimating channel under unauthorized large-scale access scene
CN114520700A (en) * 2022-01-13 2022-05-20 重庆邮电大学 Low-complexity signal detection method in machine type communication under low-resolution ADC
CN114520700B (en) * 2022-01-13 2024-03-22 陈卫锋 Low-complexity signal detection method in machine type communication under low-resolution ADC
CN114826480A (en) * 2022-03-11 2022-07-29 南京邮电大学 Signal detection method and device of non-orthogonal multiple access system in large frequency offset environment
CN114826480B (en) * 2022-03-11 2023-07-18 南京邮电大学 Signal detection method and device of non-orthogonal multiple access system in large frequency offset environment
CN114584437A (en) * 2022-05-06 2022-06-03 北京理工大学 Semi-blind message detection method and device of single handshake protocol
CN114584437B (en) * 2022-05-06 2022-07-01 北京理工大学 Semi-blind message detection method and device of single handshake protocol
CN115002929A (en) * 2022-05-16 2022-09-02 西南交通大学 Active user detection method based on joint estimation of preamble and arrival angle
CN115002929B (en) * 2022-05-16 2024-05-28 西南交通大学 Active user detection method based on joint estimation of leading and arrival angle
CN115865296A (en) * 2022-11-23 2023-03-28 东南大学 Orthogonal pilot frequency sequence active detection method based on covariance
CN115865296B (en) * 2022-11-23 2024-07-02 东南大学 Orthogonal pilot sequence active detection method based on covariance
CN117915397A (en) * 2024-03-20 2024-04-19 银河航天(西安)科技有限公司 Method, device and storage medium for transmitting telemetry transport frame based on virtual channel
CN117915397B (en) * 2024-03-20 2024-05-31 银河航天(西安)科技有限公司 Method, device and storage medium for transmitting telemetry transport frame based on virtual channel

Also Published As

Publication number Publication date
CN110071881B (en) 2020-12-15

Similar Documents

Publication Publication Date Title
CN110071881A (en) A kind of any active ues detection of adaptive expense and channel estimation methods
Ke et al. Compressive sensing-based adaptive active user detection and channel estimation: Massive access meets massive MIMO
Guo et al. Exploiting multiple antennas for cognitive ambient backscatter communication
US11671151B2 (en) Efficient peak-to-average-power reduction for OFDM and MIMO-OFDM
WO2019041470A1 (en) Large-scale mimo robust precoding transmission method
CN105577587B (en) Mimo system channel estimation methods and device under a kind of 60GHz indoor scenes
US20120002567A1 (en) Cross-layer optimization for next-generation wifi systems
CN108832976A (en) A kind of uplink channel estimation method of extensive mimo system
CN106028451A (en) User grouping mechanism applied to NOMA
US20120177001A1 (en) Space division multiple access transmission method of statistical characteristic mode
CN111181671A (en) Deep learning-based downlink channel rapid reconstruction method
Ding et al. Machine learning enabled preamble collision resolution in distributed massive MIMO
CN107302800A (en) Apparatus and method for mixing multiple access wireless communication system
Fu et al. Active user identification based on asynchronous sparse Bayesian learning with SVM
Abebe et al. MIMO-based reliable grant-free massive access with QoS differentiation for 5G and beyond
US11647468B2 (en) Transmission power allocation method based on user clustering and reinforcement learning
Wang et al. Energy-efficient collection scheme based on compressive sensing in underwater wireless sensor networks for environment monitoring over fading channels
Zhang et al. Variational Bayesian inference clustering-based joint user activity and data detection for grant-free random access in mMTC
WO2012106963A1 (en) Method and device for eliminating interference and noise
Jespersen et al. Deep learning for synchronization and channel estimation in NB-IoT random access channel
CN104702540B (en) A kind of signal for cell-edge terminals receives optimization method
Zhang et al. Selection transmitting/maximum ratio combining for timing synchronization of MIMO-OFDM systems
Nan et al. Weighted compressive sensing based uplink channel estimation for time division duplex massive multi‐input multi‐output systems
Jiang et al. AcsiNet: Attention-based deep learning network for CSI prediction in FDD MIMO systems
Li et al. Wavelet transform DC-GAN for diversity promoted fingerprint construction in indoor localization

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant