The content of the invention
To solve the deficiencies in the prior art, it is an object of the invention to provide a kind of MIMO blind Channel Estimations fuzziness is removed
Method, the method are based on Orthogonal Space-Time Block Code (Orthogonal Space-Time Block Code, OSTBC), can
The ambiguity issue of blind Channel Estimation in the non-cooperation MIMO communications of effectively solving.
In order to realize above-mentioned target, the present invention is adopted the following technical scheme that:
A kind of MIMO blind Channel Estimation fuzziness minimizing technologies based on Orthogonal Space-Time Block Code, it is characterised in that bag
Include following steps:
(1), set up, analyze and process MIMO space-time system models:
(1) one is set up with NtIndividual transmitting antenna and NrThe MIMO space-time systems of individual reception antenna, transmitting antenna transmitting
Each frame signal is made up of training sequence, Cyclic Prefix and three part of data, and aforementioned training sequence is time domain orthogonal training sequence,
Channel is flat slow change rayleigh fading channel, it is assumed that continuous NfThe channel status of frame signal experience is identical;
(2) analyze time domain orthogonal training sequence signal:
Assume the time domain orthogonal training sequence signal N of k moment any antennast× 1 dimension complex vector strK () represents, then connects
The collection of letters number is expressed as:
ytr(k)=Hstr(k)+v(k)
In formula, H is Nr×NtDimension rayleigh fading channel response matrix, ytrK () is Nr× 1 ties up received signal vector, and v (k) is
Nr× 1 dimension noise vector, it is that 0, variance is that aforementioned noise obeys averageGauss distribution;
(3) before transmission data signal is grouped:
Make s (k)=[s1(k) s2(k) … sN(k)]TFor armed k-th packet being made up of N number of symbol,
And each symbol independent same distribution;
(4) process k-th packet s (k):
S (k) is mapped as into N through Space Time Codingt× L dimensions encoder matrix C (k):
In formula, AnAnd BnRespectively corresponding to nth symbol snReal part (the s of (k)Rn(k)) and imaginary part (sIn(k)) coding
Matrix, numbers of the L for encoder matrix time slot then receive signal and are expressed as:
Y (k)=HC (k)+V (k)
In formula, Y (k) is Nr× L ties up receipt signal matrix, and V (k) is Nr× L ties up noise matrix, and Y (k), V (k) obey average
For 0, variance it isGauss distribution;
(2), channel estimation and decoding:
(1) export the estimation signal of first data block:
First, joint utilizes blind and based on time domain orthogonal training sequence channel estimation methods, to first data block
Carry out blind Channel Estimation;Then, maximum likelihood decoding is done using signal and channel joint two-dimensional search to first data block, it is defeated
Go out the estimation signal of first data block;
(2) export the estimation signal of subsequent data blocks:
First, the time domain orthogonal training sequence estimated value for being obtained using zero forcing equalization is carried out based on time domain to subsequent data blocks
The channel estimation of quadrature training sequence, makes the estimation channel of subsequent data blocks have and first data block identical phase mode
Paste;Then, maximum likelihood decoding is done using sending signal linear search to subsequent data blocks, exports the estimation letter of subsequent data blocks
Number.
The aforesaid MIMO blind Channel Estimation fuzziness minimizing technologies based on Orthogonal Space-Time Block Code, it is characterised in that
When blind Channel Estimation being carried out to first data block, do time domain orthogonal training sequence using chu sequences, sent out using Alamouti
Send scheme.
The aforesaid MIMO blind Channel Estimation fuzziness minimizing technologies based on Orthogonal Space-Time Block Code, it is characterised in that
When blind Channel Estimation being carried out to first data block, do time domain orthogonal training sequence, adopt using QPSK modulation sequences
Alamouti delivery plans.
The invention has benefit that:By combining two-dimensional search using signal and channel, using orthogonal space-time block codes
Dependency between middle signal, eliminates the intrinsic phase ambiguity of channel estimation results on the basis of blind Channel Estimation, reduces letter
The scope that road may be solved, so as to the phase place provides convenient for determination decoded signal;Additionally, this scheme is by different pieces of information block point
Not Li Yong blind and channel estimation methods based on time domain orthogonal training sequence, by the different pieces of information block decoding data for once sending
Phase deviation unification to a fixed value, solve in non-cooperative communication as the ambiguity issue of blind Channel causes decoding wrong
Problem by mistake;The method of the present invention can apply to various multi-antenna signals, the blind recognition of collaboration communication signal, blind Detecting.
Specific embodiment
Make specific introduction to the present invention below in conjunction with the drawings and specific embodiments.
First, set up, analyze and process MIMO space-time system models
1st, MIMO space-time system models are set up
One is set up with NtIndividual transmitting antenna and NrThe MIMO space-time systems of individual reception antenna, it is every that transmitting antenna is launched
One frame signal is made up of training sequence, Cyclic Prefix and three part of data, and channel is flat slow change rayleigh fading channel, it is assumed that even
Continuous NfThe channel status of frame signal experience is identical, i.e., one data block is by NfFrame is constituted, as shown in Figure 1.
Training sequence in wireless communication system is just being divided into time domain orthogonal training sequence, frequency domain quadrature training sequence and code domain
Training sequence is handed over, in the space-time system model set up by the present invention, training sequence adopts time domain orthogonal training sequence.
2nd, analyze time domain orthogonal training sequence signal
Assume the time domain orthogonal training sequence signal N of k moment any antennast× 1 dimension complex vector strK () represents, then connects
The collection of letters number is expressed as:
ytr(k)=Hstr(k)+v(k)
In formula, H is Nr×NtDimension rayleigh fading channel response matrix, ytrK () is Nr× 1 ties up received signal vector, and v (k) is
Nr× 1 dimension noise vector, wherein, it is that 0, variance is that noise obeys averageGauss distribution.
3rd, MIMO space-time system models are processed
Data signal is due to needing to carry out Space Time Coding, therefore is first grouped before transmission, makes s (k)=[s1(k) s2
(k) … sN(k)]TFor armed k-th packet being made up of N number of symbol, and each symbol independent same distribution.
Next k-th packet s (k) is processed, specifically, s (k) N is mapped as into through Space Time Codingt× L dimensions are compiled
Code Matrix C (k):
In formula, AnAnd BnRespectively corresponding to nth symbol snReal part (the s of (k)Rn(k)) and imaginary part (sIn(k)) coding
Matrix, numbers of the L for encoder matrix time slot.Then receive signal to be expressed as:
Y (k)=HC (k)+V (k)
In formula, Y (k) is Nr× L ties up receipt signal matrix, and V (k) is Nr× L ties up noise matrix, and Y (k), V (k) obey average
For 0, variance it isGauss distribution.
Additionally, because channel becomes slowly, the channel of adjacent data block n+1 and data block n follows following relation:
HnMultiple Gauss distribution is obeyed with △ H, β is constant coefficient, the less channel variation of β value is less.
2nd, channel estimation and decoding
In conventional Blind channel estimation algorithm, the phase ambiguity that each channel estimation is produced is different from and is random
, this brings very big difficulty to being correctly decoded.
Next, with reference to Fig. 2, channel estimation involved in the present invention and decoding scheme is discussed in detail.
1st, channel estimation and decoding are carried out to first data block
Channel estimation part:Joint utilizes blind and based on time domain orthogonal training sequence channel estimation methods, to first
Individual data block carries out blind Channel Estimation, by the phase ambiguity unification of first data block a to fixed value, this fixed value by
Semantic judgement determines.
Decoded portion:Due to the intrinsic fuzziness of blind Channel Estimation, there is multigroup solution in channel estimation.The present invention is in lsb decoder
Divide by combining two-dimensional search using signal and channel, using the dependency between signal in Space Time Coding, to first data
Block does maximum likelihood decoding, and the scope that channel may be solved further is reduced on the basis of channel estimation, so as to for final phase place
The determination provides convenient of fuzziness.
For the decoded signal phase place deflection unification for being caused all data blocks by blind Channel Estimation fuzziness, we only exist
The two-dimensional search of signal and channel is carried out in the maximum likelihood decoding of first data block, detailed process is shown in Fig. 3.
When carrying out blind Channel Estimation to first data block, as the blind Channel Estimation is using time domain orthogonal training sequence
Complete, and for different time domain orthogonal training sequences, the performance of scheme has difference.
Processing scheme when occurring in that numerous multi-antenna spaces for different communication environment in recent years.Wherein, Space-Time Block Coding
(STBC) scheme improves connective stability and improves data transmission rate by providing diversity gain, is great representational sky
When scheme.However, its performance is largely dependent upon the accuracy of channel estimation.
By taking classical Alamouti schemes as an example, its encoder matrix can be expressed asTherefore, under
Face we by taking two transmitting antennas as an example will discuss respectively and time domain orthogonal training sequence is done by chu sequences and QPSK modulation sequences
The channel estimation and decoding problem of Alamouti delivery plans.
(1), by the use of chu sequences as time domain orthogonal training sequence
When noise is not considered, there is 180 ° or 0 ° of phase ambiguity in whole mimo system channel matrix.Specifically push away
Lead process as follows:
Make h1,h2The true subchannel of two transmitting antennas, s are represented respectively1,s2Actual transmission signal is represented, then receives letter
Number it is:
OrderRepresent channel estimation value,Final decoded result is represented, if the subchannel estimated value of two antennas is all
There are 180 ° of phase place deflections, when being decoded with such channel, we can obtain:
OrderThen α1=α2=-1, as a result identical with signal is received, error vector is 0, so situation
Under, decoded signal can occur 180 ° of phase place deflections.
From deriving, for Alamouti delivery plans, decoded portion has certain selection, makes to phase ambiguity
The decoded result for obtaining finally has been only possible to two kinds, respectively (s1 s2) and (- s1 -s2), wherein, (s1 s2) it is correct decoding.
(2), time domain orthogonal training sequence is done using QPSK modulation sequences
When noise is not considered, there is 0 °, 180 °, 90 ° or -90 ° of phase ambiguity in whole mimo system channel matrix.
Specific derivation is as follows:
Make h1,h2The true subchannel of two transmitting antennas, s are represented respectively1,s2Actual transmission signal is represented, then receives letter
Number it is:
OrderRepresent channel estimation value,Final decoded result is represented, when being decoded with such channel, α1With
α21, -1, i or-i are can use respectively, and removing is correctly decoded (α1=α2=situation 1), also three kinds situations we need to consider:
If 1. the subchannel estimated value of two antennas all has 180 ° of phase place deflections, we can obtain:
OrderThen α1=α2=-1, as a result identical with signal is received, error vector is 0, so situation
Under, decoded signal can occur 180 ° of phase place deflections.
If 2. the first sub-channels estimated value has 90 ° of phase place deflections, there are -90 ° of phase places in the second sub-channels estimated value
Deflection, we can obtain:
OrderThen α1=-i, α2=i, as a result identical with signal is received, error vector is 0, so situation
Under, there are -90 ° of phase place deflections in the decoded signal of antenna 1, the decoded signal of antenna 2 occurs 90 ° of phase place deflections.
If 3. the first sub-channels estimated value has -90 ° of phase place deflections, there are 90 ° of phase places in the second sub-channels estimated value
Deflection, we can obtain:
OrderThen α1=i, α2=-i, as a result identical with signal is received, error vector is 0, so situation
Under, there are 90 ° of phase place deflections in the decoded signal of antenna 1, the decoded signal of antenna 2 occurs -90 ° of phase place deflections.
It is to sum up aforementioned, when being carried out to Alamouti delivery plans using QPSK modulation sequences as time domain orthogonal training sequence
When channel estimation and decoding, decoded result has following four possible:(s1 s2)、(-s1 -s2)、(-is1 is2)、(is1 -is2)。
2nd, channel estimation and decoding are carried out to subsequent data blocks
Channel estimation part:The time domain orthogonal training sequence estimated value obtained using zero forcing equalization is carried out to subsequent data blocks
Based on the channel estimation of time domain orthogonal training sequence, the estimation channel of subsequent data blocks is made to have and first data block identical
Phase ambiguity, so as to the phase ambiguity unification of different pieces of information block that once will be sent is to a fixed value.
Decoded portion:Maximum likelihood decoding is done using sending signal linear search to subsequent data blocks, follow-up data is exported
The estimation signal of block.
It can be seen that, after channel is further determined that (after channel estimation and decoding are carried out to first data block), to follow-up
Data block decoding we use general maximum likelihood algorithm, i.e., maximum likelihood solution is done using sending signal linear search
Code, identical so as to ensure that the phase place deflection that different pieces of information block decoded signal is carried, the detailed process is shown in Fig. 4.
3rd, scheme emulation and performance evaluation
In order to obtain estimating performance, we are respectively with the time domain orthogonal training sequence by chu Sequence compositions and by QPSK tune
The time domain orthogonal training sequence of Sequence composition processed, estimates channel, phase ambiguity real channel by direct SVD decomposition methods
Eliminate, count the bit error rate, it is as a result specific as follows:
1st, chu sequences are as time domain orthogonal training sequence, using Alamouti delivery plans
Simulated conditions:A MIMO space-time system with two eight receipts is set up, using Alamouti delivery plans, is sent
And receive two data blocks, each 4 frame of data block.In sending data, time domain orthogonal training sequence uses chu sequences, is accorded with by 32
Number composition, data division is made up of 64 symbols.1000 Monte Carlo simulations are carried out altogether.
The planisphere for receiving signal is shown in Fig. 5.
The planisphere of equalizing signal is shown in Fig. 6.
Decoding performance curve chart is shown in Fig. 7.
As can be seen here:Using chu sequences as time domain orthogonal training sequence, under Alamouti space time coding schemes, institute
Carrying algorithm can be preferably balanced by channel effect aliasing reception signal together, sends letter so as to correctly recover system
Number, and the bit error rate reduces with the increase of signal to noise ratio, reaches 10 in 12dB-2。
2nd, QPSK modulation sequences are as time domain orthogonal training sequence, using Alamouti delivery plans
Simulated conditions:A MIMO space-time system with two eight receipts is set up, using Alamouti delivery plans, is sent
And receive two data blocks, each 4 frame of data block.In sending data, time domain orthogonal training sequence uses QPSK modulation sequences, by
32 symbol compositions, data division are made up of 64 symbols.1000 Monte Carlo simulations are carried out altogether, count the bit error rate.
The planisphere for receiving signal is shown in Fig. 8.
The planisphere of equalizing signal is shown in Fig. 9.
Decoding performance curve chart is shown in Figure 10.
As can be seen here:Using QPSK modulation sequences as time domain orthogonal training sequence, in Alamouti space time coding schemes
Under, carried algorithm preferably balanced can receive channel effect aliasing reception signal together, send out so as to correctly recover system
The number of delivering letters, and the bit error rate reduces with the increase of signal to noise ratio, reaches 10 in 15dB-2。
During by the use of QPSK modulation sequences as time domain orthogonal training sequence, by the decoding performance of Alamouti delivery plans with
The decoding performance of the completely known system of channel information is compared, and sees Figure 11.
As can be seen here:Under the scene that this patent considers, receiving terminal cannot know time domain orthogonal training sequence particular content
And under channel condition information, therefore the bit error rate performance and cooperative communication scene of blind Channel Estimation and decoding, it is known that real channel
The bit error rate performance for being decoded has certain gap, but also can realize system proper communication to a certain extent.
To sum up aforementioned, the method for the present invention combines two-dimensional search by adopting signal and channel first, using Space Time Coding
Dependency between middle signal, eliminates the intrinsic phase ambiguity of channel estimation results on the basis of blind Channel Estimation, reduces letter
The scope that road may be solved, so as to the phase place provides convenient for determination decoded signal.
Additionally, the solution of the present invention is blind and based on time domain orthogonal training sequence by being utilized respectively in different pieces of information block
Channel estimation methods, by the phase deviation unification of the different pieces of information block decoding data for once sending to a fixed value.
Sum it up, the method for the present invention solves the shadow that the intrinsic phase ambiguity of blind Channel Estimation is caused to system decoding
Ring.
The method of the present invention can apply to various multi-antenna signals, the blind recognition of collaboration communication signal, blind Detecting.
It should be noted that above-described embodiment the invention is not limited in any way, all employing equivalents or equivalent change
The technical scheme obtained by the mode changed, all falls within protection scope of the present invention.