Summary of the invention
The purpose of the present invention is to solve the decline problems that CFO deviation in LTE network leads to OFDM performance, to evade
The spectrum efficiency for preferably utilizing entire communication system in a multi-path environment, provides higher cell traffic throughput, promotes visitor
Family perception, proposes a kind of CFO estimation method based on more noises.
In order to achieve the above object, the CFO estimation method based on more noises designed by the present invention, including the touching of more noises
Hair, capture composite signal, estimation deflection factor, estimation CFO and etc.: a variety of interchannel noises are added to transmitting signal, to difference
Reception signal synthesized, and then frequency deviation is estimated.
Preferred scheme the following steps are included:
Step 1: more noise triggerings;
Step 1-1: setting constant Ng, it is N=N for lengthg 2Transmitting signal x (n) (n=0..N-1), receive letter
Number y (n)=x (n) * δ (ε)+w (n), wherein δ is the signal deflection factor as caused by CFO, it is the function of ε, and ε is to be evaluated
CFO, w (n) be interchannel noise;
Step 1-2: setting amount of noise k generates unitary matrice training sequence
Uk={ u1(n), u2(n) ... uk(n) } interchannel noise W, is generated at randomk={ w1(n), w2(n) ... wk(n) } it, utilizes
The channel model that step 1-1 is established generates corresponding reception signal respectively
Yk={ y1(n), y2(n) ... yk(n)};
Step 2: capture composite signal;
Step 2-1: for WkAnd Yk, calculate composite signal Zk=Yk-Wk={ z1(n), z2(n) ... zk(n) }, wherein zi
(n)=yi(n)-wi(n), i=0 ... k;
Step 2-2: unitary matrice U is utilizedk, capture the deflection variable for leading to CFOWherein U* indicates the conjugate transposition operation of U matrix,
rfi(n)=ui*(yi(n)-wi(n)), i=0 ... k;
Step 3: estimation deflection factor
Step 3-1: deflection inequality is calculated
Step 3-2: deflection factor is calculated
Step 4: estimation CFO;
CFO is estimated using the deflection factor that step 3-2 is generated, is obtained
Wherein, Re () indicates to seek the real part functions of imaginary number, and j indicates imaginary unit.
The resulting CFO estimation method based on more noises of the present invention generates different reception signals using a variety of noises.
It is synthesized using the reversible property of unitary matrice transposition to signal is received, to obtain transmitting and receive the frequency deflection of signal.
The resulting CFO evaluation method based on more noises of the present invention, may be implemented same transmitting signal and makes an uproar in different background
The estimation to frequency deflection is realized in training under sound using the correlation of channel.
Embodiment 1:
The CFO estimation method based on more noises of this example description, including triggered including more noises, composite signal is captured,
Estimation deflection factor, estimation CFO and etc..
Preferred scheme the following steps are included:
Step 1: more noise triggerings;
Step 1-1: setting constant Ng, it is N=N for lengthg 2Transmitting signal x (n) (n=0..N-1), receive letter
Number y (n)=x (n) * δ (ε)+w (n), wherein δ is the signal deflection factor as caused by CFO, it is the function of ε, and ε is to be evaluated
CFO, w (n) be interchannel noise;
Step 1-2: setting amount of noise k generates unitary matrice training sequence
Uk={ u1(n), u2(n) ... uk(n) } interchannel noise W, is generated at randomk={ w1(n), w2(n) ... wk(n) } it, utilizes
The channel model that step 1-1 is established generates corresponding reception signal respectively
Yk={ y1(n), y2(n) ... yk(n)};
Step 2: capture composite signal;
Step 2-1: for WkAnd Yk, calculate composite signal Zk=Yk-Wk={ z1(n), z2(n) ... zk(n) }, wherein zi
(n)=yi(n)-wi(n), i=0 ... k;
Step 2-2: unitary matrice U is utilizedk, capture the deflection variable for leading to CFOWherein U* indicates the conjugate transposition operation of U matrix,
rfi(n)=ui*(yi(n)-wi(n)), i=0 ... k;
Step 3: estimation deflection factor
Step 3-1: deflection inequality is calculated
Step 3-2: deflection factor is calculated
Step 4: estimation CFO;
CFO is estimated using the deflection factor that step 3-2 is generated, is obtained
Wherein, Re () indicates to seek the real part functions of imaginary number, and j indicates imaginary unit.
Below with N=4, NgThis method is specifically described for=2, representative basis data are as shown in table 1.
Step 1: more noise triggerings;
Step 1-1: a variety of channel noise signals are generated using additive white Gaussian noise
w1(n)=awgn (rand (2,2)+i*rand (2,2), 15)=[9.0949,8.5906;3.6409 2.6249],
w2(n)=awgn (rand (2,2)+i*rand (2,2), 15)=[4.4900,8.2094;5.1739 1.7341],
W2={ w1(n), w2(n)};
Step 1-2: unitary matrice training sequence u is generated1(n)=[- 0.5257, -0.8507;- 0.8507,0.5257],
u2(n)=[- 0.7678,0.6407;- 0.6407, -0.7678], U2={ u1(n), u2(n) } channel model, is utilized
It generates respectively and receives signal y1(n)=[7.7186-0.0025i, 7.2142-0.0108i;3.3159+0.0016i 2.3000-
0.0004i],
y2(n)=[4.3630+0.0019i, 8.0824+0.0012i;3.7654-0.0023i 0.3257-0.0107i],
Y2={ y1(n), y2(n)};
Step 2: capture composite signal;
Step 2-1: composite signal is calculated
Z2=Y2-W2=
{ [- 1.3764-0.0025i, -1.3763-0.0108i;- 0.3249+0.0016i, -0.3249-0.004i];[-
0.1270+0.0019i, -0.1270+0.0012i;- 1.4085-0.0023i, -1.4085-0.0107i] };
Step 2-2: capture leads to the deflection variable of CFO
SRf=ΣkRfk=ΣkUk **Zk=[2.0000-0.0000i, 2.0000+0.0119i;2.0000+0.0023i-
1.4085-0.0107i]};
Step 3: estimation deflection factor
Calculate deflection inequalityDeflection factor
Step 4: estimation CFO;
CFO is estimated using the deflection factor that step 3-2 is generated, is obtained
We estimate MNE algorithm designed by the present invention and traditional estimation method based on CP based on leading Moose
Method and based on pilot tone Classen estimation etc. three kinds it is deemed-to-satisfy4 can be carried out comparison, referring to shown in attached drawing 2.It is tied by comparison
Fruit can be seen that for Moose and Classen estimation method under noise jamming, performance shake less, and is based on the estimation side CP and MNE
Method shake is relatively large, main reason is that MNE considers the reason of more noises.But under the premise of identical Signal to Noise Ratio (SNR),
MNE has smaller least mean-square error MSE.
Wherein, the table referred in embodiment is as follows:
Table 1