CN106788882A - The incoherent multi-user test method and device of nonopiate multi-pulse modulation - Google Patents
The incoherent multi-user test method and device of nonopiate multi-pulse modulation Download PDFInfo
- Publication number
- CN106788882A CN106788882A CN201611032864.7A CN201611032864A CN106788882A CN 106788882 A CN106788882 A CN 106788882A CN 201611032864 A CN201611032864 A CN 201611032864A CN 106788882 A CN106788882 A CN 106788882A
- Authority
- CN
- China
- Prior art keywords
- user
- signal
- matrix
- information
- module
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L1/00—Arrangements for detecting or preventing errors in the information received
- H04L1/004—Arrangements for detecting or preventing errors in the information received by using forward error control
- H04L1/0045—Arrangements at the receiver end
- H04L1/0047—Decoding adapted to other signal detection operation
- H04L1/0048—Decoding adapted to other signal detection operation in conjunction with detection of multiuser or interfering signals, e.g. iteration between CDMA or MIMO detector and FEC decoder
-
- 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
- H04L25/0246—Channel estimation channel estimation algorithms using matrix methods with factorisation
-
- 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
- H04L25/03—Shaping networks in transmitter or receiver, e.g. adaptive shaping networks
- H04L25/03006—Arrangements for removing intersymbol interference
Landscapes
- Engineering & Computer Science (AREA)
- Computer Networks & Wireless Communication (AREA)
- Signal Processing (AREA)
- Power Engineering (AREA)
- Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- Radar Systems Or Details Thereof (AREA)
Abstract
The present invention discloses the incoherent multi-user test method and device of a kind of nonopiate multi-pulse modulation, and the method includes：Receive by multi-user's superposed signal of nonopiate multi-pulse modulation；Set up the pseudo- linear model for receiving signal；Signal will be received channel correlation matrix is obtained by one group of matched filter；Make QR decomposition to correlation matrix, correlation matrix is converted into upper triangular matrix structure；Between each user by feed back successively adjudicated user profile to eliminate user between multi-access inference；Respectively eliminated with the correlation of indoor each signal by decorrelation inversion operation with indoor；Big judgement is selected the signal after detection, selects maximum to be exported as judgement from M kind possible outcomes.Decision signal of the present invention by feedback power compared with large user, the detection performance of the weaker user of power is greatly improved, and each user effectively eliminates multi-access inference by decision-feedback, even if interference user still has fine stability when more, for wireless communication technology field.
Description
Technical field
The invention belongs to wireless communication technology field, more particularly to nonopiate multi-pulse modulation Multiuser Detection, specifically
It is the incoherent multi-user test method and device of a kind of nonopiate multi-pulse modulation, can be used to improve the inspection of the weaker user of power
Performance is surveyed, even and if disturb number of users increase, detecting system is remained to good stability.
Background technology
Nonopiate multi-pulse modulation (nonorthogonal multipulse modulation, NMM) belongs to a kind of non-thread
Property modulation technique, due to using noncoherent detection in receiving terminal, it is not necessary to the phase information of carrier wave so that receive end detecting device
It is more easily implemented.Multi-user's NMM systems under additive white Gaussian noise channel, each user is possible by sending M kinds respectively
Signal waveform transmits log2The information of M-bit, and these signal waveforms can be unequal nonopiate, energy.So,
The nonorthogonality of information is sent due to each user, multi-access inference will be produced when reaching receiving terminal when superposition, have a strong impact on NMM
The capacity and performance of system.And the increase of the quantity and power with interference user, multi-access inference problem will become more tight
Weight, especially when the unbalanced power between user, it is necessary to alleviate or eliminate the influence of multi-access inference by certain means.
Multiuser Detection is to suppress multi-access inference, a kind of effective way of lifting NMM power system capacities.Traditional sensing techniques exist
After receiving signal overmatching wave filter, directly carry out selecting big decision process recovering original signal, by multi-access inference directly as height
This noise.And the concept of Multiuser Detection is, by using cross-correlation information between the user included in multi-access inference, effectively to reduce
Even eliminate the influence of interference.
With traditional CDMA (code division multiple access, CDMA) system unlike, NMM systems
System not only needs to eliminate the multi-access inference between user when Multiuser Detection is carried out, and with it is indoor also need to carry out it is incoherent
Detection.By extending the signal space of all sending signals of all users, NMM can be regarded as a kind of pseudo- linear modulation.It is based on
This pseudo- linear modulation, it is proposed that the incoherent decorrelation detector of multi-access inference can be completely eliminated, the detector is rearmounted
Three kinds of non-coherent decision criterions：Optimal, asymptotic optimization and generalized likelihood ratio test.Then also proposed performance more excellent
Incoherent least mean-square error (minimum mean squared error, MMSE) detector, has been similarly obtained the detector
Three kinds of non-coherent decision criterions：Maximum criterion, asymptotic optimization and generalized likelihood ratio test.Although decorrelation,
The computational complexity of MMSE both linear detectors is not high, but because decorrelation matrix inversion operation is exaggerated noise work(
Rate, and MMSE still has the multi-access inference of remnants so that the performance of detector is unsatisfactory.And someone further studies therefore,
A kind of non-coherent decision feedback multi-user detector, the detector can be greatly enhanced the detection of the weaker user of power
Can, but due to being required for updating channel correlation matrix again after the judgement for completing each user, and fortune of inverting is carried out to it
Calculate, computational complexity is very high.
To sum up, in the incoherent multi-user test method of existing NMM systems, otherwise noise power is exaggerated, or still
There is the multi-access inference of remnants, otherwise although detection performance is fine, its computational complexity is too high.Therefore, it is necessary to propose one kind
Computational complexity is not high, but performance remains unchanged very well, can accomplish the multi-user test method of the compromise in complexity and performance
With device.
The content of the invention
Defect it is an object of the invention to overcome prior art, it is proposed that a kind of nonopiate multi-pulse modulation it is incoherent
Multi-user test method and device.It can be greatly enhanced the detection of the weaker user of power while with lower complexity
Performance, performance will be considerably beyond existing linear detector, it might even be possible to close to the detection performance bound of single user.Even and if being
Interference number of users in system gradually increases, and the present invention still has good stability.
The present invention is a kind of incoherent multi-user test method of nonopiate multi-pulse modulation, it is characterised in that included
Following steps：
Step 1：Multi-user's superposed signal is received, the pseudo- linear model for receiving signal is set up, received by nonopiate many arteries and veins
Multiple users' superposed signal waveform r (t) of modulated, number of users is K wherein in modulating system, and each user carries out M and enters respectively
Multi-pulse modulation processed, and the signal space opened based on all possible sending signal waveform of all users, are set up and receive letter
Number extension signal space pseudo- linear model.
Step 2：The sufficient statistic y after matched filtering is obtained, based on pseudo- linear model, the multi-user's superposition that will be received
Signal, the sufficient statistic y for receiving signal is obtained by one group of matched filter matched with each subscriber signal waveform
=RAb+n, construction multiuser channel correlation matrix R；Wherein, A is the diagonal matrix being made up of the amplitude information of all users, b
The symbolic information of all users is represented, n is σ for the covariance matrix of zero-mean2The Gaussian random variable of R, σ2It is noise variance.
Step 3：Correlation matrix R is converted into upper triangular matrix structure, QR decomposition is made to correlation matrix R and is obtained R=QU, its
Middle Q is orthogonal matrix, and U is upper triangular matrix, with the conjugate transposition premultiplication of Q with the output y of matched filter, obtains vectorial z=
QHY=UAb+QHN, upper triangular matrix U is converted to by correlation matrix R.
Step 4：Multi-access inference between user is eliminated by decision-feedback, the characteristics of based on upper triangular matrix structure, to QR points
All matrixes after solution carry out piecemeal treatment by user, and each user is from bottom to up in turn using the feedback information for having adjudicated user
Interference Cancellation is carried out, is used to eliminate multi-access inference between user, specifically each has been detected with feedback discriminative information per family to be checked
User is surveyed to eliminate multi-access inference, what user to be detected eliminated is all multi-access inferences for having detected user；
Step 5：The correlation for eliminating user itself is operated by decorrelation, is respectively carried out by matrix inversion operation with indoor
Decorrelation is operated, and further eliminates the correlation of the possible sending signal waveform of all M kinds of each user itself；
Step 6：Each user's primary signal is recovered by selecting big judgement, the subscriber signal after decorrelation select and is sentenced greatly
Certainly, select maximum to be exported as the judgement of user from M kind possible outcomes, and backfeed loop is given by the decision signal, according to
The multi-access inference that the user causes to follow-up user to be detected is eliminated secondaryly, completes noncoherent detection.
Of the invention or a kind of incoherent multi-user detection device of nonopiate multi-pulse modulation, it is characterised in that according to
The handling process direction of detection signal is connected with matched filter module, QR split-matrixes conversion module, decision-feedback in turn and does
Disturb cancellation module, decorrelation inversion operation module, select big judging module.Each user is obtaining judgement letter by selecting big judging module
, it is necessary in turn feed back to decision feedback interference cancellation module after number, the information that follow-up user to be detected is obtained using feedback,
The multi-access inference that user causes to user to be detected has been adjudicated in elimination, and each module is described below：
Matched filter module, input is, by multiple user's superposed signals of nonopiate multi-pulse modulation, to be output as
Sufficient statistic is tieed up with the MK obtained after filtering, the module is obtained with one group of matched filter matched with each signal waveform
Take the sufficient statistic y=RAb+n for receiving signal, construction multiuser channel correlation matrix R；
QR split-matrix conversion modules, are input into the sufficient statistic to be obtained after matched filtering, are output as being expressed as
The MK dimensional vectors of three-legged structure, the module makees QR decomposition to correlation matrix R and obtains R=QU, and wherein Q is orthogonal matrix, and U is upper
Triangular matrix, with the conjugate transposition premultiplication of Q with the output y of matched filter, obtains vectorial z=QHY=UAb+QHN, by Correlation Moment
Battle array R is converted to upper triangular matrix U；
Decision feedback interference cancellation module, is input into the MK dimensional vectors to be obtained after QR operation splittings, and feedback is obtained
The information for having adjudicated user, be output as eliminate multi-access inference after only the M dimensional vectors comprising each user's self information, the module
The characteristics of based on upper triangular matrix structure, all matrixes after being decomposed to QR carry out piecemeal treatment by user, each user from it is lower to
On in turn carry out Interference Cancellation using the feedback information for having adjudicated user, be used to eliminate multi-access inference between user；
Decorrelation inversion operation module, is input into have eliminated the expectation user profile of multi-access inference, is output as decorrelation
What is obtained after operation treats discriminative information, and the module carries out decorrelation operation to each user by matrix inversion operation, further
Eliminate the correlation between all possible M kinds sending signal of each user itself；
Big judging module is selected, is input into expect that user's treats decision signal, be output as the original letter of expectation user for recovering
Number, the module is carried out selecting big judgement to the user profile after decorrelation, and maximum is selected from M kind possible outcomes as user's
Judgement output, and gives backfeed loop by the decision signal, and in turn eliminate that the user causes to follow-up user to be detected is more
Location is disturbed, so as to complete noncoherent detection.
The present invention compared with prior art, has the following advantages that：
1, present invention greatly enhances the detection performance of the weaker user of power.Decomposed by QR and obtain correlation matrix R's
Upper triangular matrix structure, due to not having multi-access inference, and the larger user of power positioned at the user of upper triangular matrix bottom
First it is detected, its detection better performances, when feeding back discriminative information of these power compared with large user successively by order from bottom to up,
During being iterated Interference Cancellation successively, the detection performance of the weaker user of power can be greatly enhanced, its performance is much excellent
In linear detectors such as existing decorrelation, MMSE, it might even be possible to approach the detection performance bound of single user.
2, it is of the invention when disturbing number of users more, still with good stability.Due to upper triangular matrix structure
Feature, matrix bottom user does not exist multi-access inference, and secondary lower section user only exists the interference ... ... of user, and interference is used
Amount is reduced, and the multi-access inference of these users can be eliminated by decision-feedback loop.Therefore, the system can be very
Eliminate well due to the multi-access inference that multi-user's superposition is caused, when disturbing number of users more, still with good stability,
Will not cause that detection performance substantially deteriorates with the increase of interference number of users.
3, the present invention can be effectively reduced computational complexity, and fortune is decomposed due to need to only carry out a QR after matched filtering
Calculate, and existing non-coherent decision feedback detector unlike, it is not necessary to update Correlation Moment again after each user judgement
Battle array information, its computational complexity is relatively low；And with it is indoor carry out decorrelation operate when, matrix inversion dimension only has M ranks, much low
In non-coherent decision feedback in matrix dimensionality for MK ranks correlation matrix R inversion operation.Therefore, the present invention can effectively drop
Low computational complexity, accomplishes the compromise in complexity and performance.
Brief description of the drawings
Fig. 1 is the nonopiate multi-pulse modulation system model figure in the embodiment of the present invention；
Fig. 2 is the schematic flow sheet of incoherent multi-user test method of the invention；
Fig. 3 is that upper triangular matrix piecemeal of the present invention processes schematic diagram；
Fig. 4 is the structural representation of incoherent multi-user detection device of the invention；
Fig. 5 is the ary modulation system of 4 user 4, simulation curve figure when each user power is unequal；
Fig. 6 is the ary modulation system of 6 user 4, simulation curve figure when each user power is unequal.
Specific embodiment
The present invention is explained in more detail below in conjunction with the accompanying drawings.
Embodiment 1
Nonopiate multi-pulse modulation technology can be unequal nonopiate, energy due to the sending signal of each user, when
Superposition ground will produce multi-access inference when reaching receiving terminal.In existing multiuser detection, although incoherent decorrelating detection
Multi-access inference is completely eliminated, but is also exaggerated noise simultaneously.Although incoherent MMSE detections make mean square error reach most
It is small, but still remnants have multiple access interference term.And incoherent decision-feedback can be close to list by the iterative feedback performance between user
User circle, but its computation complexity it is too high be not easy to realize.For various problems present in prior art, the present invention is proposed
A kind of incoherent multi-user test method of nonopiate multi-pulse modulation, the present invention is decomposed by QR and is converted to correlation matrix
Triangular matrix structure, multi-access inference then making decisions feedback successively from bottom to up each user to eliminate user, then
Operated by decorrelation and further eliminated with the correlation between indoor signal, the property of the weaker user of power can be greatly enhanced
Can, and because iterative process completely eliminates multi-access inference, in the more system of interference user, the present invention still has fine
Stability.
The incoherent multi-user test method of nonopiate multi-pulse modulation of the invention, the nonopiate multi-pulse modulation of use
System model it is as shown in Figure 1, it is assumed that in system have K user, each user respectively to should have M bar Line independents, have
There is a signal waveform that unit energy and equiprobability send, and the sending signal between each user is unequal non-orthogonal, energy
, each user will be changed into k roads signal by signal all the way by serioparallel exchange first, whereinThen this k roads signal is sent
Enter in mode selector, a sending signal as each user may be accordingly selected in signal waveform from M bars, and to the hair
The number of delivering letters sends after carrying out chip shaping.Assuming that channel is additive white Gaussian noise channel, the sending signal of each user exists
By after channel, synchronously reaching receiver.
The present invention is a kind of incoherent multi-user test method of nonopiate multi-pulse modulation, because nonopiate multiple-pulse is adjusted
The sending signal of each user is non-orthogonal in system, there is certain correlation, when reaching receiver when sending signal superposition,
The multi-access inference between user can be caused, the detection performance of system is had a strong impact on, power system capacity is reduced.Can be fine by the present invention
Ground eliminates multi-access inference between user, referring specifically to Fig. 2, including has the following steps：
Step 1：Multi-user's superposed signal is received, the pseudo- linear model for receiving signal is set up.
Receiving terminal is received by multiple users' superposed signal waveform r (t) of nonopiate multi-pulse modulation, and the signal is Fig. 1
Shown sending signal, has K user in system, each user carries out M system multi-pulse modulations respectively, i.e., may from M kinds
Selection one kind of signal waveform equal probability is transmitted, it is assumed that channel model is additive white Gaussian noise signal, each user's mistake
Receiver is synchronously reached after channel.
And the signal space that the present invention is opened based on all possible sending signal waveform of all users, by multiple-pulse
Modulation is equivalent to pseudo- linear modulation, sets up the pseudo- linear model for receiving signal in extension signal space.
Step 2：Obtain the sufficient statistic y after matched filtering.
Pseudo- linear model based on above-mentioned extension, the multi-user's superposed signal that will be received, by a group and each user
The matched filter that signal waveform matches receives the sufficient statistic y=RAb+n of signal to obtain, and constructs multiuser channel
Correlation matrix R；Wherein, A is the diagonal matrix being made up of the amplitude information of all users, and b represents the symbolic information of all users,
N is σ for the covariance matrix of zero-mean2The Gaussian random variable of R, σ2It is noise variance.It is K, each user point for number of users
The system for not carrying out M system multi-pulse modulations, then this group of matched filter is by all possible common MK transmission of all users
Signal is constituted, and this group of MK matched filter is that the pseudo- linear model based on above-mentioned extension is obtained.In the present invention
The sufficient statistic for receiving signal is also the output of matched filter.
Step 3：Correlation matrix R is converted into upper triangular matrix structure, referring to Fig. 3.
Make QR decomposition to correlation matrix R and obtain R=QU, wherein Q is orthogonal matrix, and U is upper triangular matrix, with the conjugation of Q
Transposition premultiplication obtains vectorial z=Q with the output y of matched filterHY=UAb+QHN so that correlation matrix R is converted to triangle
Matrix structure U.
The characteristics of present invention is based on the upper triangular matrix structure, the user positioned at matrix bottom is dry in the absence of multiple access
Disturb, when each user in turn feeds back the information for having adjudicated user to carry out Interference Cancellation from bottom to up, in theory can be with complete
Totally disappeared except multi-access inference.
Step 4：By multi-access inference between decision-feedback elimination user.
The characteristics of present invention is based on upper triangular matrix structure, all matrixes after being decomposed to QR are carried out at piecemeal by user
Reason, each user in turn carries out Interference Cancellation using the feedback information for having adjudicated user from bottom to up, is used to eliminate many between user
Location is disturbed, and specifically each to have detected and give user to be detected to eliminate multi-access inference with feedback discriminative information per family, to be detected
What user eliminated is all multi-access inferences for having detected user.Multi-access inference is eliminated with user profile has been adjudicated in the present invention,
By eliminating multi-access inference, multi-user's superposed signal is broken down into each mutually incoherent single user signal, subsequently to carry out
Respectively with indoor noncoherent detection.
Press the order arrangement of power non-decreasing in advance due to each user, in turn detected from bottom to up when by decision-feedback
During each user, power larger user be first detected, and its detection Performance comparision is good, while the discriminative information is fed back to subsequently treating
During detection user, it is also possible to improve the detection performance of the weaker user of power.
Step 5：The correlation for eliminating user itself is operated by decorrelation.
Although the multi-access inference between user has been eliminated, what is carried out due to each user is multi-pulse modulation, is sent
Signal is any one that M kinds may be in signal waveform, by the M dimensional vectors that are obtained after decision-feedback still comprising this M kind
Correlation between signal, respectively carries out decorrelation operation with indoor by matrix inversion operation, further eliminates each user certainly
The correlation of the possible sending signal waveform of all M kinds of body.
The present invention not only eliminates the multi-access inference between user, at the same also eliminate each user itself be possible to send
Correlation between signal.
Step 6：Each user's primary signal is recovered by selecting big judgement.
Each subscriber signal after to decorrelation carries out selecting big judgement, and each user is both needed to be selected most from M kind possible outcomes
Big value is exported as the judgement of user, completes the noncoherent detection to the user.And user's decision signal is given feed back to
Road, in turn eliminates the multi-access inference that the user causes to follow-up user to be detected, so as to complete the incoherent inspection of each user
Survey.
Selecting big judgement is specifically, the M dimensional vectors that each user is obtained after decorrelation operation, in M element therein
Maximum be judged to 1, remaining M-1 element is judged to 0, and the position for then occurring in M dimensional vectors according to maximum is reflected
It is corresponding log to penetrate2M-bit information, the bit information is the final expectation user's raw information for recovering, and completes anon-normal
Hand over the incoherent Multiuser Detection of multi-pulse modulation.
Each user's superposed signal waveform that the present invention obtains receiving terminal, multi-access inference is eliminated by its point by Interference Cancellation
It is each orthogonal single user signal to solve, and carries out noncoherent detection to each single user signal, and conclusive judgement is recovered
The raw information of each user.
The present invention is converted to upper triangular matrix structure by by correlation matrix R so that be superimposed upon the interference letter on each user
Number it is reduced, especially the user positioned at matrix bottom does not exist multi-access inference, based on the feature, to having pressed power in advance
Size be incremented by the user of arrangement, feedback interference is in turn made decisions from bottom to up and offsets to eliminate multi-access inference, due to
Interference the user that number of users is less and power is larger be first detected, its detection better performances, when the discriminative information of these users
Be fed back on the weaker user of power carry out interference eliminate when, the performance of the weaker user of power can be greatly enhanced.
Embodiment 2
The incoherent multi-user test method of nonopiate multi-pulse modulation is with embodiment 1, the foundation wherein described in step 1
By the pseudo- linear model of the reception signal of nonopiate multi-pulse modulation, specifically include：
Step 1.1：Receive by multi-user's superposed signal of nonopiate multi-pulse modulation, wherein in all of K user
Each user transmits log by sending M kinds non-orthogonality signal2The information of M-bit, by after additive white Gaussian noise channel, K
The superposed signal of individual user synchronously reaches receiver, and it receives signal and is expressed as：
Wherein z (t) is that one-sided power spectrum density is σ2Noise, skj(t), j ∈ 1,2 ..., and M } it is expressed as the of user k
J Line independent, with unit energy and wait the waveform of general transmission, EkjAnd φkjThe energy of j-th signal of user k is represented respectively
Amount and phase.
Step 1.2：The pseudo- linear model for receiving signal in extension signal space is set up, by extending all letters of all users
The signal space of number waveform, will entirely participate in the symbolic information of all K users in nonopiate multi-pulse modulation system with one
The column vector of MK dimensions is represented：The wherein information b of user kkThe M dimensional vectors being expressed as：
And bkEquiprobably take possible one kind of M kinds in above-mentioned set.
Multi-pulse modulation technology is regarded as a kind of pseudo- linear modulation, and the pseudo- linear model for obtaining receiving signal is as follows：
Wherein, SkIt is the N × M rank matrixes being made up of the M signal waveform of user k, andAndIt is the N × MK rank matrixes being made up of the signal waveform of all K users,
MK × MK rank block diagonal matrix that expression is made up of range value, wherein each M × M rank matrix-block is to represent the M of user
The range value that sending signal is constituted is planted, and ψ is Gaussian random variable, its average is 0, the covariance square of the Gaussian random variable
Battle array is σ2I, wherein I are unit matrix.
Pseudo- linear model of the invention, is dexterously regarded as pseudo- linear modulation by nonopiate multi-pulse modulation, by the model,
Multiuser detection proposed in conventional CDMA system just may apply in nonopiate multi-pulse modulation system.It is existing
Incoherent multi-user test method, is all based on what pseudo- linear model was proposed, including irrelevant decorrelation, incoherent MMSE detections
Method.
Embodiment 3
The incoherent multi-user test method of nonopiate multi-pulse modulation with embodiment 1-2, wherein described in step 2
With wave filter group, constituted with the wave filter that sending signal waveform matches that is possible to of all users by MK, reception signal
Overmatching wave filter, the MK dimension sufficient statistics for obtaining：
Y=SHR=RAb+n
Sufficient statistic y is the output of matched filter group.
Wherein R=SHS is the MK × MK rank correlation matrixes being made up of signal waveform, and matrix A is by the amplitude of all users
The diagonal matrix of information structure, vectorial b is the MK dimensional vectors being made up of the symbolic information of all users, and n is zero-mean
Covariance matrix is σ2The Gaussian random variable of R.
The present invention with unlike conventional CDMA system, when conventional CDMA system carries out Multiuser Detection, it is assumed that system use
Amount is K, then only need to the K matched filter matched with sending signal, and a kind of incoherent multi-user of the invention examines
In survey, then the MK wave filter that sending signal matches that is possible to all users is needed, its dimension is bigger.Simultaneously because
Need to know that all users' is possible to sending signal waveform in advance, therefore present invention is mainly applied to cooperation communication system
In concentration receiver, such as in the base station of multiple access communication.
Embodiment 4
The incoherent multi-user test method of nonopiate multi-pulse modulation is with embodiment 1-3, sentencing wherein described in step 4
Certainly feedback interference is offset and is included：
4.1：User is arranged in order by the order of power non-decreasing, i.e. E1≤E2≤…≤EK。
4.2：Output to matched filter carries out partitioning of matrix treatment by user, and per M, the vector or matrix of dimension are to divide
It it is one piece, it represents a certain user as an entirety.
4.3：Feedback has adjudicated the information of user to eliminate multi-access inference, spy of the present invention based on upper triangular matrix structure
Point, referring to Fig. 3, in units of user's block, since power is compared with large user, in turn feeds back the letter for having adjudicated user from bottom to up
Breath carries out Interference Cancellation operation, is used to eliminate multi-access inference between user；Specific practice is that prominent user need not be carried out
Interference Cancellation can directly adjudicate output, and then other each users are in turn fed back by order from bottom to up and all adjudicated user
Information, with offset adjudicated the multi-access inference that user causes to user to be detected.
Due to positioned at upper triangular matrix bottom user do not exist multi-access inference, each user afterwards, by matrix from
Under supreme in turn each user by the multi-access inference comprising all users in matrix lower section, and pass through decision-feedback, from lower past
On in turn each user can obtain all information for having adjudicated user in matrix lower section, so, for each use to be detected
Family, when the influence of noise is ignored, the information obtained using feedback can fully eliminate the multi-access inference between user.And
The user that power is larger, interference user is less is first detected judgement, its detection better performances, and lower-powered user passes through
Using user's discriminative information of these detection better performances, it is also possible to be greatly enhanced the performance of the weaker user of power.
Embodiment 5
The incoherent multi-user test method of nonopiate multi-pulse modulation is with embodiment 1-4, the solution wherein described in step 5
Related inversion operation, refers to after the multi-access inference between eliminating user by decision-feedback, to obtain one and only expect to use comprising each
The M dimensional vectors of family self information, the M dimensional vectors include the information for expecting the possible sending signal of all M kinds of user, each to send
Between signal have certain correlation, decorrelation operation is carried out by matrix inversion operation, so as to obtain expect user wait sentence
Certainly information bk.Then by M dimensional vectors bkCarry out selecting big judgement, select the maximum in M value to be exported as judgement, just may be used
Recover the primary signal for expecting user.
The M dimensional vectors of each user obtained for decision-feedback treatment, can be decomposed into a M ranks matrix and a table
Show that each user treats the product of the M dimensional vectors of discriminative information.Operated by decorrelation inversion operation, to each user M dimensional vectors
With above-mentioned M ranks inverse of a matrix matrix, can obtain each user M dimensions treats judgement output to premultiplication.
Embodiment 6
Of the invention or a kind of incoherent multi-user detection device of nonopiate multi-pulse modulation, referring to Fig. 4, according to detection
The handling process direction of signal be connected with turn matched filter module, QR split-matrixes conversion module, decision-feedback interference support
Disappear module, decorrelation inversion operation module, select big judging module.Each user is obtaining decision signal by selecting big judging module
Afterwards, it is necessary in turn feed back to one backfeed loop of decision feedback interference cancellation module composition, the court verdict of each user is anti-
Present as the Interference Cancellation parameter of next user to be detected, the detection of all users to be detected is carried out successively, that is, subsequently
User to be detected using the information that obtains of feedback, the multi-access inference that user causes to user to be detected, each mould have been adjudicated in elimination
Block is described below：
Matched filter module, input is, by multiple user's superposed signals of nonopiate multi-pulse modulation, to be output as
Sufficient statistic is tieed up with the MK obtained after filtering, the module is obtained with one group of matched filter matched with each signal waveform
Take the sufficient statistic y=RAb+n for receiving signal, construction multiuser channel correlation matrix R；
QR split-matrix conversion modules, are input into the sufficient statistic to be obtained after matched filtering, are output as being expressed as
The MK dimensional vectors of three-legged structure, the module makees QR decomposition to correlation matrix R and obtains R=QU, and wherein Q is orthogonal matrix, and U is upper
Triangular matrix, with the conjugate transposition premultiplication of Q with the output y of matched filter, obtains vectorial z=QHY=UAb+QHN, by Correlation Moment
Battle array R is converted to upper triangular matrix U；
Decision feedback interference cancellation module, is input into the MK dimensional vectors to be obtained after QR operation splittings, and feedback is obtained
The information for having adjudicated user, be output as eliminate multi-access inference after only the M dimensional vectors comprising each user's self information, the module
The characteristics of based on upper triangular matrix structure, all matrixes after being decomposed to QR carry out piecemeal treatment by user, each user from it is lower to
On in turn carry out Interference Cancellation using the feedback information for having adjudicated user, be used to eliminate multi-access inference between user；
Decorrelation inversion operation module, is input into have eliminated the expectation user profile of multi-access inference, is output as decorrelation
What is obtained after operation treats discriminative information, and the module carries out decorrelation operation to each user by matrix inversion operation, further
Eliminate the correlation between all possible M kinds sending signal of each user itself；
Big judging module is selected, is input into expect that user's treats decision signal, be output as the original letter of expectation user for recovering
Number, the module is carried out selecting big judgement to the user profile after decorrelation, and maximum is selected from M kind possible outcomes as user's
Judgement output, and gives backfeed loop by the decision signal, and in turn eliminate that the user causes to follow-up user to be detected is more
Location is disturbed, so as to complete noncoherent detection.
Embodiment 7
The overall of the incoherent multi-user detection device of nonopiate multi-pulse modulation is constituted with embodiment 6, wherein, judgement is anti-
Feedback Interference Cancellation module, including K-1 backfeed loop submodule, this K-1 submodule need before to meet user K, user K- respectively
1 ... user's 2 selects big judging module, constitutes decision-feedback loop.These submodules are by feedback user K, user K- respectively
The decision signal of 1 ... user 2, is used to eliminate these users to the multi-access inference caused by user to be detected.
During module specific works, the user K positioned at upper triangular matrix bottom due in the absence of the multi-access inference between user,
Decision-feedback loop is not needed, directly passed through decorrelation inversion operation module, selected the judgement letter that big judging module obtains user K
Number, the signal is then fed back into the K-1 backfeed loop submodule, the multi-access inference that user K is caused to user K-1 is eliminated,
Then the signal of user K-1 passed through into decorrelation inversion operation module again, select the judgement letter that big judging module obtains user K-1
Number, the K-1 backfeed loop submodule ... ... is fed back to followed by by the decision signal of user K, user K-1, the rest may be inferred,
The signal that user has been adjudicated by turn feeding back carries out Interference Cancellation, can eliminate the multi-access inference between all users.
The characteristics of based on upper triangular matrix structure, the user K positioned at matrix bottom does not need decision-feedback loop, can be with
Directly carry out decorrelation to operate and adjudicate output, and be located at the user 1 of matrix the top due to there are other all users to it
The multi-access inference that causes is, it is necessary to all feedback informations for having adjudicated user, and user 1 is last signal to be detected, its
Decision signal need not be fed back again.Therefore, by upper triangular matrix structure, computation complexity is obtained further the present invention
Simplify.
Embodiment 8
The overall of the incoherent multi-user detection device of nonopiate multi-pulse modulation is constituted with embodiment 6-7, wherein, select big
Judging module, described to select big judging module, its input is the only M dimensional vectors b comprising user's self informationk, it is necessary to bkMiddle M
Individual element carries out selecting big judgement, and maximum is judged to 1, and remaining is all 0, and so, the total M kinds of court verdict of the column vector can
Energy：bk={ bk1,…bkm,…bkM}∈{[1,0…0]T,…,[0,0…1]T, then occur in M dimensional vectors according to maximum
Position be mapped as corresponding log2M-bit information
Expectation user's raw information that as conclusive judgement is obtained, completes the incoherent of whole nonopiate multi-pulse modulation
Multiuser Detection process.
The incoherent multi-user detection device of the nonopiate multi-pulse modulation of the present invention is in the non-of nonopiate multi-pulse modulation
On the basis of relevant multi-user test method, based on the scheme proposed when implementing.
A complete example is given below, then the method for the present invention is further explained：
Embodiment 9
The incoherent multi-user test method of nonopiate multi-pulse modulation is with embodiment 1-5
What Fig. 2 showed the incoherent multi-user test method of nonopiate multi-pulse modulation in the present embodiment realizes flow chart element
Figure, and the partitioning of matrix treatment and specific calculating of Fig. 3 are combined, realize describing in detail to of the invention.As shown in Fig. 2 including
Following steps：
Step 101, receives by multiple users' superposed signal waveform r (t) of nonopiate multi-pulse modulation.
Consider multi-user's NMM modulation schemes, each user passes by sending one of M kind non-orthogonality signals in K user
Defeated log2The information of M-bit.By synchronously reaching receiver after additive white Gaussian noise channel, it receives signal and is：
Wherein z (t) is that one-sided power spectrum density is σ2Noise.skj(t), j ∈ 1,2 ..., and M } it is expressed as the of user k
J Line independent, with unit energy and wait the waveform of general transmission.EkjWith φkjThe energy of j-th signal of user k is represented respectively
Amount and phase.
Step 102, the signal space that the sending signal waveform based on all users is opened obtains receiving signal in extension
The pseudo- linear model of signal space.
In this step, the pseudo- linear model for receiving signal in extension signal space includes：
By extending the signal space of all signal waveforms of all users, by the symbolic information of all K users in system
The column vector tieed up with a MK is represented：The wherein information b of user kkThe M dimensional vectors being expressed as：
bkDeng generally taking possible one kind of M kinds in above-mentioned set.
Multi-pulse modulation technology can be regarded as a kind of pseudo- linear modulation, and the pseudo- linear model for obtaining receiving signal is as follows：
Wherein, SkIt is the N × M rank matrixes being made up of the M signal waveform of user k, andAndIt is the N × MK rank matrixes being made up of the signal waveform of all K users.
MK × MK rank block diagonal matrix that expression is made up of range value, wherein each M × M rank matrix-block is to represent the M of user
Plant the range value that sending signal is constituted.And it is 0 that ψ is average, covariance matrix is σ2The Gaussian random variable of I.
Step 103, will receive signal and is obtained by one group of matched filter matched with each signal waveform fully
Statistic y, and construct channel correlation matrix R.
In this step, matched filter group is matched with the sending signal waveform that is possible to of all users by MK
Wave filter is constituted, and receives signal overmatching wave filter, the MK dimension sufficient statistics for obtaining：
Y=SHR=RAb+n.
Wherein R=SHS is the MK × MK rank correlation matrixes being made up of signal waveform, and matrix A is by the amplitude of all users
The diagonal matrix of information structure, vectorial b is the MK dimensional vectors being made up of the information symbol of all users, and n is zero-mean
Covariance matrix is σ2The Gaussian random variable of R.
Step 104, QR decomposition is made to correlation matrix R and obtains R=QU, and wherein Q is orthogonal matrix, and U is upper triangular matrix, and
With the conjugate transposition premultiplication of Q with the output of matched filter, correlation matrix R is converted into upper triangular matrix structure, i.e. z=QHy
=UAb+QHn。
Wherein, A is the diagonal matrix being made up of the amplitude information of all users, and b represents the symbolic information of all users, n
For the covariance matrix of zero-mean is σ2The Gaussian random variable of R.
Step 105, the characteristics of based on upper triangular matrix structure, is carried out leading between piecemeal treatment, each user to matrix by user
Crossing feedback and having adjudicated the information of user carries out Interference Cancellation, is used to eliminate multi-access inference.
In this step, it is necessary first to which user power is pressed the order arrangement of non-decreasing, i.e. E1≤E2≤…≤EK。
Because the judgement output of each user is a M dimensional vectors bk, when feedback is made decisions, ignore noise item
Influence, partitioning of matrix treatment can be carried out by user to the output of matched filter, per M dimension vector or matrix be to represent one
Individual user.
The structural representation of specific partitioning of matrix treatment is as shown in Figure 3：
After piecemealWherein zk=[zk1,zk2,…,zkM]TIt is the column vector of M dimensions.Additionally, will
K-th row block in upper triangular matrix U is (k-1) * M+1 rows to kth * M rows, the row block of kth+1~K be kth * M+1 arrange to
K*M is arranged, and the matrix-block for being constituted is denoted as UK-k.Similarly by Ak+1,…,AKThe M that sub-block is constituted × (K-k) M ranks matrix-block note
Make AK-k, by bk+1,…,bK(K-k) M × 1 dimensional vector that sub-block is constituted is denoted as bK-k.By leading diagonal in matrix U, matrix A
On k-th M × M rank submatrix, U is denoted as respectivelyk、Ak。
After the completion of the fast treatment of matrix point, by order from bottom to up, the information that will adjudicate user of iteration is anti-successively
It is fed in the detection of active user to eliminate multi-access inference.It is embodied as by the vector that is obtained after feedback：
After carrying out phase shift to above formula, obtain：
As k=K, the multi-access inference that the other users that equation right-hand member do not have are caused.And as k ≠ K, it is assumed that in the absence of error
Propagate, the field feedback adjudicated is completely correct, and equation right-hand member can also be completely eliminated power it is caused compared with large user
Multi-access inference.
Step 106, respectively with indoor by decorrelation matrix inversion operation, each user to having eliminated multi-access inference,
Carry out noncoherent detection.
After the multi-access inference between eliminating user by decision-feedback, for each user, one is obtained only comprising itself
The M dimensional vectors U of informationkAkbk, and the column vector cannot directly carry out selecting judgement greatly and obtain output valve, it is necessary to first pass through matrix
Inversion operation obtains information symbol bk。
The concrete operations for carrying out decorrelation matrix inversion operation to each user are as follows：
And then eliminate with the correlation between indoor each signal.
Step 107, carries out selecting big judgement to the signal after detection, selects a kind of defeated as judgement from M kind possible outcomes
Go out.
In this step, due to bkOnly to include the M dimensional vectors of user's self information, the total M kinds of the column vector may be tied
Really：bk∈{[1,0…0]T,…,[0,0…1]T, it is necessary to bkIn each element bkmCarry out selecting big judgement, it is output as
The log corresponding to judgement maximum for arriving2User's raw information of M-bit
Expectation user's raw information that as conclusive judgement is obtained, completes the incoherent of whole nonopiate multi-pulse modulation
Multiuser Detection process.
The present invention is made iteratively the Interference Cancellation of each user successively by upper triangular matrix structure, efficiently solve by
The weaker user of power is excessively influenceed to detect the technical problem of performance in interference user.
Embodiment 10
The incoherent multi-user test method and device of nonopiate multi-pulse modulation with embodiment 1-9, in order to verify this hair
The bright detection performance that can be greatly enhanced the weaker user of power, below to the embodiment of the present invention in incoherent Multiuser Detection side
Method carries out emulation and compares with decorrelation of the prior art, MMSE multi-user test methods, referring in particular to Fig. 5.
Based on Monte Carlo simulation flowcharts, channel is additive white Gaussian noise channel, the hair of each user in system
It is 31 balance gold sequences that the number of delivering letters uses code length, and respectively may send sequence then by same gold sequences with indoor M bars
The cyclically shifted sequences of row are constituted.The transmit power of each user with unequal, but can be required with indoor M bar sending signals
Power is consistent.Assuming that each user is arranged in order by the order of power non-decreasing, i.e. E1≤E2≤…≤EK, user
Order during detection in reverse order is detected that first the signal of detection user K, finally detects the signal of user 1, wherein user successively
1 signal power is most weak.It is assumed that user 1 contrasts multi-user test method proposed by the present invention and prior art to expect user
In decorrelation, the detection performance to the weaker user of power of MMSE multi-user test methods, and with the inspection under single user channel
Performance is surveyed as referring to boundary.
Fig. 5 gives 4 user 4 ary modulation system, i.e., K=4, M=4 in nonopiate multi-pulse modulation, each user's
Transmit power is unequal, wherein interference user power is expect user power 10 times, has obtained expectation user 1 in all kinds of detections
The relation curve that bit error probability under device changes with signal to noise ratio.It can be seen that decorrelation detector carry out it is many
When user detects, due to the excessive reason of interference user power, noise is also seriously amplified while multi-access inference is eliminated, its
Detection performance is much worse than preferable single user circle；And MMSE detectors still have the multiple access interference term of remnants, also due to interference
User power is excessive so that remaining multi-access inference is also more, and its detection performance is simply slightly better than decorrelation, and this explanation exists
Detection performance both when interference user power is excessive is all undesirable；And a kind of incoherent Multiuser Detection proposed by the present invention
Method, due to detection power, larger user is first detected, and the probability of its judgement error is smaller, and then this is sentenced by iterative feedback
When certainly signal to the weaker user of power to carry out Interference Cancellation, the performance of the weaker user of power can be greatly enhanced, contrast can
See, with the increase of signal to noise ratio, bit error probability will be far below decorrelation, MMSE detectors to the present invention, 10-4Bit-errors
During probability, compared with decorrelation, MMSE detectors, the present invention has the nearly gain of 2dB, it might even be possible to close to the inspection of single user
Survey boundary.
Embodiment 11
, with embodiment 1-9, simulated conditions are with real for the incoherent multi-user test method and device of nonopiate multi-pulse modulation
Example 10 is applied, in order to verify that the present invention still has good stability when disturbing user more, Fig. 6 is each when more to interference user
Multiuser detection has carried out emulation to the detection performance of the weaker user of power and has compared.
Fig. 6 gives 6 user 4 ary modulation system, i.e., K=6, M=4 in nonopiate multi-pulse modulation, relative to implementation
Example 10, further increases by two interference users in this example.The transmit power of each user is unequal, and interference user power is remained as
Expect user power 10 times, have obtained expecting what bit error probability of the user 1 under all kinds of detectors changed with signal to noise ratio
Relation curve.The simulation curve of comparison diagram 5 can be seen that with the increase of interference number of users, decorrelation, the inspection of MMSE detectors
Further to deteriorate when surveying performance than 44 ary modulation of user, have mistake higher general under identical error rate condition
Rate, and more to be worse than single user circle；And detection tool is carried out using a kind of incoherent multi-user test method of the invention
There is good stability, detection performance when still with 44 ary modulation of user is suitable.10-4During bit error probability, with solution
Related, MMSE detectors are compared, and performance gain of the invention has been expanded to 4dB, illustrate that detection stability of the invention is higher.
In brief, the incoherent multi-user test method and device of nonopiate multi-pulse modulation disclosed by the invention, connect
Receive by multiple user's superposed signal waveforms of nonopiate multi-pulse modulation；Set up the pseudo-wire for receiving signal in extension signal space
Property model；Signal will be received by one group of matched filter matched with each signal waveform, channel correlation matrix is constructed；It is right
Correlation matrix makees QR decomposition, and is converted to correlation matrix with the output of matched filter with the conjugate transposition premultiplication of matrix Q
Triangular matrix structure；The information for having adjudicated user between each user by feeding back successively carries out Interference Cancellation, between being used to eliminate user
Multi-access inference；It is each that noncoherent detection is carried out by decorrelation inversion operation with indoor, eliminate with the correlation of indoor each signal；It is right
Signal after detection carries out selecting big judgement, selects a kind of conduct judgement to export from M kind possible outcomes.The present invention is by feeding back work(
Rate can be greatly enhanced the detection performance of the weaker user of power compared with the decision signal of large user, and each user by according to
Secondary ground feedback iteration all completely eliminates multi-access inference, even if the present invention still has very well in the more system of interference user
Stability.For wireless communication field, such as the concentration receiver of cooperation communication system.
Claims (8)
1. a kind of incoherent multi-user test method of nonopiate multi-pulse modulation, it is characterised in that including having the following steps：
Step 1：Multi-user's superposed signal is received, the pseudo- linear model for receiving signal is set up, is received and is adjusted by nonopiate multiple-pulse
Multiple users' superposed signal waveform r (t) of system, number of users is K wherein in modulating system, and it is many that each user carries out M systems respectively
Impulse modulation, and the signal space opened based on all possible sending signal waveform of all users, are set up reception signal and existed
Extend the pseudo- linear model of signal space；
Step 2：The sufficient statistic y after matched filtering is obtained, based on pseudo- linear model, the multi-user's superposition letter that will be received
Number, the sufficient statistic y=for receiving signal is obtained by one group of matched filter matched with each subscriber signal waveform
RAb+n, construction multiuser channel correlation matrix R；Wherein, A is the diagonal matrix being made up of the amplitude information of all users, b tables
Show the symbolic information of all users, n is σ for the covariance matrix of zero-mean2The Gaussian random variable of R, σ2It is noise variance；
Step 3：Correlation matrix R is converted into upper triangular matrix structure, QR decomposition is made to correlation matrix R and is obtained R=QU, wherein Q
It is orthogonal matrix, U is upper triangular matrix, with the conjugate transposition premultiplication of Q with the output y of matched filter, obtains vectorial z=QHy
=UAb+QHN, upper triangular matrix U is converted to by correlation matrix R；
Step 4：Multi-access inference between user is eliminated by decision-feedback, the characteristics of based on upper triangular matrix structure, after being decomposed to QR
All matrixes carry out piecemeal treatment by user, each user is in turn carried out using the feedback information for having adjudicated user from bottom to up
Interference Cancellation, is used to eliminate multi-access inference between user, and specifically each has been detected with feedback discriminative information per family to use to be detected
To eliminate multi-access inference, what user to be detected eliminated is all multi-access inferences for having detected user at family；
Step 5：The correlation for eliminating user itself is operated by decorrelation, solution phase is respectively carried out by matrix inversion operation with indoor
Operation is closed, the correlation of the possible sending signal waveform of all M kinds of each user itself is further eliminated；
Step 6：Each user's primary signal is recovered by selecting big judgement, the subscriber signal after decorrelation is carried out to select big judgement,
Select maximum to be exported as the judgement of user from M kind possible outcomes, and backfeed loop is given by the decision signal, in turn
The multi-access inference that the user causes to follow-up user to be detected is eliminated, noncoherent detection is completed.
2. the incoherent multi-user test method of nonopiate multi-pulse modulation according to claim 1, it is characterised in that institute
State and receive multi-user's superposed signal and set up the pseudo- linear model for receiving signal, refer to：
Step 1.1：Receive by multi-user's superposed signal of nonopiate multi-pulse modulation, wherein each user in all K users
Log is transmitted by sending one of M kind non-orthogonality signals2The information of M-bit, by after Gaussian white noise channel, K user
Superposed signal synchronously reach receiver, its receive signal be expressed as：
Wherein z (t) is that one-sided power spectrum density is σ2Noise, skj(t), j ∈ 1,2 ..., and M } it is expressed as j-th of user k
Line independent, with unit energy and wait the waveform of general transmission, EkjAnd φkjThe energy of j-th signal of user k is represented respectively
With phase；
Step 1.2：The pseudo- linear model for receiving signal in extension signal space is set up, by extending all signal waves of all users
The signal space of shape, the symbolic information b of all K users in system is represented with the column vector that a MK is tieed up：The wherein information b of user kkThe M dimensional vectors being expressed as：
bkEquiprobably take possible one kind of M kinds in above-mentioned set；
A kind of pseudo- linear modulation is regarded as by by multi-pulse modulation technology, the pseudo- linear model for obtaining receiving signal is as follows：
Wherein, SkIt is the N × M rank matrixes being made up of the M signal waveform of user k, andAndIt is the N × MK rank matrixes being made up of the signal waveform of all K users；
MK × MK rank block diagonal matrix that expression is made up of range value, wherein each M × M rank matrix-block is to represent the M of user
Plant the range value that sending signal is constituted；And it is 0 that ψ is average, covariance matrix is σ2The Gaussian random variable of I, I is unit square
Battle array.
3. the incoherent multi-user test method of nonopiate multi-pulse modulation according to claim 1, it is characterised in that institute
Matched filter group is stated, is made up of the wave filter that sending signal waveform matches that is possible to of MK and all users, reception
Signal overmatching wave filter group, obtains the sufficient statistic of MK dimensions：
Y=SHR=RAb+n
Wherein R=SHS is the MK × MK rank correlation matrixes being made up of signal waveform, and matrix A is by the amplitude information structure of all users
Into diagonal matrix, vectorial b is the MK dimensional vectors being made up of the symbolic information of all users, and n for zero-mean covariance
Matrix is σ2The Gaussian random variable of R.
4. the incoherent multi-user test method of nonopiate multi-pulse modulation according to claim 1, it is characterised in that institute
Stating decision feedback interference cancellation includes：
4.1：User is arranged in order by the order of power non-decreasing, i.e. E1≤E2≤…≤EK；
4.2：Output to matched filter carries out partitioning of matrix treatment by user, and per M, the vector or matrix of dimension are to be divided into one
Block, it represents a certain user as an entirety；
4.3：Feedback has adjudicated the information of user to eliminate multi-access inference, the characteristics of based on upper triangular matrix structure, with user's block
It is unit, since power is compared with large user, in turn feeds back from bottom to up and adjudicated the information of user and carry out Interference Cancellation operation,
It is used to eliminate multi-access inference between user；Specific practice is that prominent user need not carry out Interference Cancellation and can directly adjudicate defeated
Go out, then other each users are in turn fed back the information for having adjudicated user by order from bottom to up, user has been adjudicated to offset
The multi-access inference caused to user to be detected.
5. the incoherent multi-user test method of nonopiate multi-pulse modulation according to claim 1, it is characterised in that institute
Decorrelation inversion operation is stated, refers to after the multi-access inference between eliminating user by decision-feedback, to obtain one only comprising each phase
Hope that the M dimensional vectors of user's self information, the M dimensional vectors include the information for expecting the possible sending signal of all M kinds of user, respectively
There is certain correlation between sending signal, it is necessary to carry out decorrelation operation by matrix inversion operation, so as to obtain expecting to use
Discriminative information b is treated at familyk；Then by M dimensional vectors bkCarry out selecting big judgement, select the maximum in M value defeated as adjudicating
Go out, just can recover the primary signal for expecting user.
6. a kind of incoherent multi-user detection device of nonopiate multi-pulse modulation, it is characterised in that according to the place of detection signal
Reason flow direction is connected with matched filter module, QR split-matrixes conversion module, decision feedback interference cancellation module, solution in turn
Related inversion operation module, select big judging module；Each user after selecting big judging module to obtain decision signal, it is necessary to successively
Ground feeds back to decision feedback interference cancellation module, and use has been adjudicated in the information that follow-up user to be detected is obtained using feedback, elimination
The multi-access inference that family is caused to user to be detected, each module is described below：
Matched filter module, input is, by multiple user's superposed signals of nonopiate multi-pulse modulation, to be output as matching filter
The MK dimension sufficient statistics obtained after ripple, the module is connect with one group of matched filter matched with each signal waveform to obtain
The sufficient statistic y=RAb+n of the collection of letters number, construction multiuser channel correlation matrix R；
QR split-matrix conversion modules, are input into the sufficient statistic to be obtained after matched filtering, are output as that upper triangle can be expressed as
The MK dimensional vectors of structure, the module makees QR decomposition to correlation matrix R and obtains R=QU, and wherein Q is orthogonal matrix, and U is upper triangle
Matrix, with the conjugate transposition premultiplication of Q with the output y of matched filter, obtains vectorial z=QHY=UAb+QHN, by correlation matrix R
Be converted to upper triangular matrix U；
Decision feedback interference cancellation module, is input into the MK dimensional vectors to be obtained after QR operation splittings, and feedback is obtained
The information of user is adjudicated, is output as eliminating a M dimensional vector comprising each user's self information after multi-access inference, the module is based on
The characteristics of upper triangular matrix structure, all matrixes after being decomposed to QR carry out piecemeal treatment by user, each user from bottom to up according to
Interference Cancellation is carried out using the feedback information for having adjudicated user secondaryly, is used to eliminate multi-access inference between user；
Decorrelation inversion operation module, is input into have eliminated the expectation user profile of multi-access inference, is output as decorrelation operation
What is obtained afterwards treats discriminative information, and the module carries out decorrelation operation to each user by matrix inversion operation, further eliminates
Correlation between all possible M kinds sending signal of each user itself；
Big judging module is selected, is input into expect that user's treats decision signal, be output as the expectation user's primary signal for recovering, should
Module is carried out selecting big judgement to the user profile after decorrelation, and maximum is selected from M kind possible outcomes as the judgement of user
Output, and backfeed loop is given by the decision signal, in turn eliminate the user and the multiple access that follow-up user to be detected causes is done
Disturb, so as to complete noncoherent detection.
7. the incoherent multi-user detection device of nonopiate multi-pulse modulation according to claim 6, it is characterised in that institute
Decision feedback interference cancellation module, including K-1 backfeed loop submodule are stated, this K-1 submodule needs before to meet user respectively
K, user K-1 ... users 2 select big judging module, constitute decision-feedback loop.These submodules are by feedback user respectively
K, the decision signal of user K-1 ... users 2, are used to eliminate these users to the multi-access inference caused by user to be detected；
During module specific works, the user K positioned at upper triangular matrix bottom is due in the absence of the multi-access inference between user, being not required to
Decision-feedback loop is wanted, directly passed through decorrelation inversion operation module, selected the decision signal that big judging module obtains user K, connect
And the signal is fed back into the K-1 backfeed loop submodule, eliminate the multi-access inference that user K is caused to user K-1, then again
By the signal of user K-1 by decorrelation inversion operation module, select big judging module and obtain the decision signal of user K-1, then connect
And the decision signal of user K, user K-1 is fed back into the K-1 backfeed loop submodule ... ..., the rest may be inferred, by successively
The signal that ground feedback has adjudicated user carries out Interference Cancellation, can eliminate the multi-access inference between all users.
8. the incoherent multi-user detection device of nonopiate multi-pulse modulation according to claim 6, it is characterised in that institute
State and select big judging module, its input is the only M dimensional vectors b comprising user's self informationk, it is necessary to bkIn M element carry out
Big judgement is selected, maximum is judged to 1, remaining is all 0, so, the total M kinds of court verdict of the column vector may：bk=
{bk1,…bkm,…bkM}∈{[1,0…0]T,…,[0,0…1]T, the position for then occurring in M dimensional vectors according to maximum
It is mapped as corresponding log2M-bit information
Expectation user's raw information that as conclusive judgement is obtained, completes the incoherent multiplex of whole nonopiate multi-pulse modulation
Family detection process.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201611032864.7A CN106788882B (en) | 2016-11-16 | 2016-11-16 | Noncoherent multi-user detection method and device for non-orthogonal multi-pulse modulation |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201611032864.7A CN106788882B (en) | 2016-11-16 | 2016-11-16 | Noncoherent multi-user detection method and device for non-orthogonal multi-pulse modulation |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN106788882A true CN106788882A (en) | 2017-05-31 |
| CN106788882B CN106788882B (en) | 2020-01-31 |
Family
ID=58970672
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201611032864.7A Active CN106788882B (en) | 2016-11-16 | 2016-11-16 | Noncoherent multi-user detection method and device for non-orthogonal multi-pulse modulation |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106788882B (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110062384A (en) * | 2019-04-28 | 2019-07-26 | 西安交通大学 | A kind of wireless monitoring method based on signal rotation |
| CN110198177A (en) * | 2019-05-07 | 2019-09-03 | 中国人民解放军陆军工程大学 | A kind of multi-sequence joint frequency-hopping communication method of combination pseudorandom feature code |
| CN111404579A (en) * | 2020-02-17 | 2020-07-10 | 济南浪潮高新科技投资发展有限公司 | Multi-sequence frequency hopping communication method, transmitter and receiver |
| CN114339828A (en) * | 2021-12-07 | 2022-04-12 | 重庆邮电大学 | Design method of non-orthogonal time reversal uplink multiple access system |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2004040789A2 (en) * | 2002-10-29 | 2004-05-13 | Axiocom Inc. | Cascade filter receiver for ds-cdma communication systems |
| CN101286779A (en) * | 2008-06-02 | 2008-10-15 | 西安电子科技大学 | Method for eliminating multi-user interference in multiuser MIMO system |
| CN102437979A (en) * | 2011-05-24 | 2012-05-02 | 中兴通讯股份有限公司 | Interference elimination method and device |
-
2016
- 2016-11-16 CN CN201611032864.7A patent/CN106788882B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2004040789A2 (en) * | 2002-10-29 | 2004-05-13 | Axiocom Inc. | Cascade filter receiver for ds-cdma communication systems |
| CN101286779A (en) * | 2008-06-02 | 2008-10-15 | 西安电子科技大学 | Method for eliminating multi-user interference in multiuser MIMO system |
| CN102437979A (en) * | 2011-05-24 | 2012-05-02 | 中兴通讯股份有限公司 | Interference elimination method and device |
Non-Patent Citations (2)
| Title |
|---|
| M.K. VARANASI等: "Noncoherent decision-feedback multiuser detection", 《IEEE TRANSACTIONS ON COMMUNICATIONS》 * |
| 周欣等: "非正交多脉冲调制信号的非相干MMSE多用户检测", 《中国测试技术》 * |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110062384A (en) * | 2019-04-28 | 2019-07-26 | 西安交通大学 | A kind of wireless monitoring method based on signal rotation |
| CN110198177A (en) * | 2019-05-07 | 2019-09-03 | 中国人民解放军陆军工程大学 | A kind of multi-sequence joint frequency-hopping communication method of combination pseudorandom feature code |
| CN110198177B (en) * | 2019-05-07 | 2021-10-08 | 中国人民解放军陆军工程大学 | Multi-sequence combined frequency hopping communication method combined with pseudo-random feature codes |
| CN111404579A (en) * | 2020-02-17 | 2020-07-10 | 济南浪潮高新科技投资发展有限公司 | Multi-sequence frequency hopping communication method, transmitter and receiver |
| CN114339828A (en) * | 2021-12-07 | 2022-04-12 | 重庆邮电大学 | Design method of non-orthogonal time reversal uplink multiple access system |
| CN114339828B (en) * | 2021-12-07 | 2023-07-21 | 重庆邮电大学 | Design method of non-orthogonal time reversal uplink multiple access system |
Also Published As
| Publication number | Publication date |
|---|---|
| CN106788882B (en) | 2020-01-31 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106788882A (en) | The incoherent multi-user test method and device of nonopiate multi-pulse modulation | |
| CN101427507B (en) | Method and apparatus for use of space time trellis codes based on channel phase feedback | |
| CN102999783A (en) | Method for generating MIMO (multiple-input and multiple-output) radar orthogonal polyphase code signals on the basis of genetic-tabu hybrid algorithm | |
| CN101771442B (en) | Method, system and device for selecting precoding vector and mobile terminal | |
| CN104486044A (en) | Broadband module mixing pretreatment method for large-scale MIMO system | |
| CN103490804A (en) | Method for selecting multi-user MIMO system antenna based on priority genetic simulated annealing | |
| CN105450275A (en) | Optimal energy efficiency-based antenna selection method for multi-user and large-scale antenna relay system | |
| US6452916B1 (en) | Space-time spreading method of CDMA wireless communication | |
| CN105392192A (en) | Energy-efficiency-optimization-based power distribution method in multi-user large-scale antenna relay system | |
| CN105763300A (en) | Receiver design method for spatial pattern segmentation multiple access technology | |
| CN101227254B (en) | Method for detecting V-BLAST in MIMO system | |
| CN106341169A (en) | Antenna selection method for the uplink of multi-user large-scale MIMO system | |
| CN105743559A (en) | Hybrid beam-forming and space-time coding multi-user downlink transmission method in Massive MIMO (Multiple Input Multiple Output) system | |
| CN106357309A (en) | Method of large scale MIMO linear iterative detection under non-ideal channel | |
| CN105162503A (en) | Multi-user beam forming and antenna selection joint-design method in massive multiple-input multiple-output (MIMO) system | |
| CN105007248B (en) | A kind of downlink precoding method of MIMO Full-duplex cellular systems | |
| CN106972877A (en) | The signal BDS extensive MIMO beam selection methods of multiple cell mmWave are found based on wave beam | |
| CN104639473B (en) | A kind of mimo channel method of estimation based on fully-complementary sequence and compressed sensing | |
| CN106301495A (en) | Generalized spatial modulation system antenna selecting method based on maximum channel capacity criterion | |
| CN107181705A (en) | A kind of half-blind channel estimating method and system | |
| CN101471714B (en) | Method for obtaining descending beam shape-endowing weight value based on intelligent antenna system | |
| CN105207704A (en) | Precoding-scheme-based downlink precoding method for multi-user multi-input and multi-output system | |
| CN104349510A (en) | Method and device of implementing wireless body area network | |
| CN108761438A (en) | Radar-communication integration Design of Signal method based on MPSK-DSSS | |
| CN105007096B (en) | Nonopiate code word based on DS CDMA systems is with frequency multi-beam separation method |
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 |