A kind of pulse interference suppression method of ofdm system
Technical field
The present invention relates to a kind of pulse interference suppression technology of communications field, more particularly, to a kind of pulse of ofdm system
Disturbance restraining method.
Background technology
Due to OFDM (Orthogonal Frequency Division Multiplexing, OFDM) technology
With preferable anti-multipath jamming ability, therefore it is by standards such as IEEE802.11, IEEE802.16 and IEEE802.20
Adopt, and be widely used in various digital broadband communication systems.However, in many practical communication systems, when one
Ofdm signal not only contains Gaussian Background noise but also also many artificially generated by after transmission in signal is received
Impulsive noise, these impulsive noises are entirely different with the statistical property of Gaussian Background noise, cause traditional to be made an uproar based on background
Performance deep fades of the ofdm system of sound design under impulse noise interference environment.Therefore, effective impulsive noise is designed to do
It is very necessary that suppressing method is disturbed for lifting the performance of ofdm system.
Widely studied at present on the impulse noise mitigation problem in ofdm system.Himal A.Suraweera
Et al. filter amplitude by increasing a kind of nonlinear filter for presetting decision threshold before traditional OFDM receiver
Larger impulsive noise, realizes the suppression to impulse noise interference.Pablo Torio et al. then propose a kind of based on cell
The ofdm system impulse noise suppression method of intertexture.Additionally, Takuya Kitamura et al. have studied the arteries and veins based on copy signal
Noise jamming suppressing method is rushed, symbol judgement is carried out by iteration and copy signal is deleted the elimination realized to impulsive noise.
However, research all assumes that impulsive noise is to obey a certain specific noise statisticses distributed model above, this causes in treatment band
, it is necessary to first pass through a data-aided training process to obtain specific impulsive noise system when having the signal of impulse noise interference
Meter distributed constant, not only increases system complexity, and when the noise statisticses distributed model for assuming can not reflect actual arteries and veins
When rushing noise statisticses, the performance of such method can deep fades.For this problem, Lin and Evans et al. are made an uproar using pulse
Sparse characteristic of the sound in time domain, it is proposed that a kind of Impulsive Noise Mitigation Method based on compressed sensing technology, the method is not required to
The statistical model of impulsive noise is preset, and can be by management loading (Sparse Bayesian
Learning, SBL) restructing algorithm from compared with low-dimensional known signal vector in estimate higher-dimension impulsive noise vector, but due to
SBL algorithms are directly maximized in each iterative process when hyper parameter is solved by maximum likelihood method using conditional expectation
All hyper parameters to be estimated are updated simultaneously, this causes that whole SBL convergences of algorithm are slower, and conditional expectation can be caused to maximize
Step is difficult to, so cause whole method time complexity is higher, practicality is not strong.
The content of the invention
The technical problems to be solved by the invention are to provide a kind of pulse interference suppression method of ofdm system, and it can have
The time estimated for impulsive noise is reduced to effect, and impulsive noise can be effectively reduced to OFDM symbol error probability
Influence.
The present invention solve the technical scheme that is used of above-mentioned technical problem for:A kind of pulse interference suppression side of ofdm system
Method, it is characterised in that comprise the following steps:
1. in transmitting terminal, ofdm signal vector to be sent is designated as X, it is assumed that there is M known pilot in X, then by X
M known pilot constitute vector be designated as XT;Then inverse discrete Fourier transformer inverse-discrete and addition Cyclic Prefix are successively carried out to X,
Obtain transmission signal vector;Wherein, the dimension of X is N × 1 dimension, N>2,1<M<N, XTDimension be M × 1 dimension, T represented by X
The indexed set that the index of M known pilot is constituted, the length of T is M;
2. assume the status information of channel for transmission signal, it is known that then transmitting terminal by transmission sending signal to
Measure to receiving terminal, the impulsive noise vector sum ambient noise vector that transmission signal vector is produced in transmitting procedure in channel
Correspondence is designated as e and n, wherein, unknown impulsive noise vector e and the dimension of unknown ambient noise vector n are N × 1 dimension;
3. in receiving terminal, the received signal vector to receiving successively is removed Cyclic Prefix and discrete Fourier becomes
Change, obtain pending received signal vector, be designated as Y, Y after X superimposed pulse noise vector e and ambient noise vector n by forming;
Then extracted in X from YTIt is middle to be superimposed the signal vector formed after impulsive noise vector e and ambient noise vector n, note
It is YT;Further according to the status information of channel, Y is removedTIn with XTCorresponding known signal vector, obtains only comprising unknown
The known vector Z of impulsive noise vector e and unknown two compositions of ambient noise vector n;Wherein, the dimension of Y is N × 1 dimension, YT
Dimension be M × 1 dimension, the dimension of Z is M × 1 dimension;
4. the estimate problem that unknown impulsive noise vector e will be asked for is converted into one estimate of e of solution so that l2-
l1Norm type object functionThe minimum problem of value, then willInF (e) is expressed as with functional form,Wherein, symbol " | | | |2" it is to ask for l2Norm
Symbol, symbolTo ask for l2Norm square, symbol " | | | |1" it is to ask for l1Norm sign, FTRepresent known M × N
The DFT matrix of dimension, τ represents regularization parameter, and f () is function representation mode;
5. the estimate for solving an e by iterative manner causes l2-l1Norm type object function
Value it is minimum, detailed process is:
5. the value for -1, setting regularization parameter τ is τ=0.1 | | (FT)HZ||∞, the initial value for making the estimate of e is e0, e0
=((FT)HFT)-1(FT)HZ, and make k represent iterations, the initial value of k is 1, wherein, symbol " | | | |∞" infinite to ask for
Norm sign, (FT)HIt is FTAssociate matrix, ((FT)HFT)-1It is (FT)HFTInverse matrix;
5. the evaluation points α of a kth time iteration -2, is definedk,Make againWherein, (Sk)HIt is SkAssociate matrix, Sk=ek-ek-1, ekObtained after representing kth time iteration
E estimate, ek-1The estimate of the e obtained after -1 iteration of kth is represented,Table
Show and ask for f (e after kth time iterationk) first derivative values,F (e after -1 iteration of kth are asked in expressionk-1) single order lead
Numerical value,
5. -3, according to αkAnd uk, the estimate of the e obtained after kth time iteration is calculated, it is designated as ek,Wherein, max () is to take max function, symbol " | | " it is the symbol that takes absolute value
Number,Represent ukIn i-th element, 1≤i≤N;
5. -4, judgeWhether set up, if set up, perform step 5. -5;Otherwise, k=k+1 is made,
Step is then back to 5. -2 to continue executing with, wherein, symbol " | | | |2" it is to ask for l2Norm sign, ε is the minimum normal of setting
Number, "=" in k=k+1 is assignment;
5. -5, terminate iterative process, obtain the final estimate of e, be designated asWherein,In
"=" is assignment;
6. the final estimate of e is subtracted with YThe frequency-domain expression obtained by DFT, that is, pressed down
Received signal vector after impulse noise interference processed.
5. described step takes ε=10 in -4-5。
Compared with prior art, the advantage of the invention is that:
1) the inventive method will ask for the vectorial estimate problem of unknown impulsive noise and be converted into solution one so that l2-
l1The minimum problem of the value of norm type object function, can so be effectively prevented from the optimal sparse solution of direct estimation impulsive noise
When the NP-hard problems that occur.
2) the inventive method solves an optimization subproblem to obtain one so that l by iterative manner2-l1Norm type mesh
The minimum optimal solution of the value of scalar functions, not only significantly reduces the time estimated for impulsive noise, and be effectively reduced
Influence of the impulsive noise to OFDM symbol error probability, so just improves practicality.
Brief description of the drawings
Fig. 1 is the system block diagram that pulse interference suppression is realized using the inventive method;
Fig. 2 is the timing of signal to noise ratio one, is being entered with the existing impulse noise suppression method based on SBL using the inventive method
The contrast schematic diagram of the average operating time of the CPU consumed during row ofdm system impulse noise mitigation;
Fig. 3 is OFDM systems when the inventive method changes with the existing impulse noise suppression method based on SBL with signal to noise ratio
The comparison schematic diagram of the symbol error probability of system.
Specific embodiment
The present invention is described in further detail below in conjunction with accompanying drawing embodiment.
A kind of pulse interference suppression method of ofdm system proposed by the present invention, Fig. 1 is given using the inventive method reality
The system block diagram of existing pulse interference suppression, the inventive method specifically includes following steps:
1. in transmitting terminal, ofdm signal vector to be sent is designated as X, it is assumed that there is M known pilot in X, then by X
M known pilot constitute vector be designated as XT;Then inverse discrete Fourier transformer inverse-discrete and addition Cyclic Prefix are successively carried out to X,
Obtain transmission signal vector;Wherein, the dimension of X is N × 1 dimension, N>2,1<M<N, XTDimension be M × 1 dimension, T represented by X
The indexed set that the index of M known pilot is constituted, the length of T is M.
In the specific implementation, N typically can value be 16,64,128,512 or 1024 etc., and M typically takes the half of N, i.e.,
2. assume the status information of channel for transmission signal, it is known that then transmitting terminal by transmission sending signal to
Measure to receiving terminal, the impulsive noise vector sum ambient noise vector that transmission signal vector is produced in transmitting procedure in channel
Correspondence is designated as e and n, wherein, unknown impulsive noise vector e and the dimension of unknown ambient noise vector n are N × 1 dimension.
3. in receiving terminal, the received signal vector to receiving successively is removed Cyclic Prefix and discrete Fourier becomes
Change, obtain pending received signal vector, be designated as Y, Y after X superimposed pulse noise vector e and ambient noise vector n by forming;
Then extracted in X from YTIt is middle to be superimposed the signal vector formed after impulsive noise vector e and ambient noise vector n, note
It is YT;Further according to the status information of channel, Y is removedTIn with XTCorresponding known signal vector, obtains only comprising unknown
, can be referred to as vector Z herein unknown by the known vector Z of impulsive noise vector e and unknown two compositions of ambient noise vector n
Projection of the impulsive noise vector in known pilot;Wherein, the dimension of Y is N × 1 dimension, YTDimension be M × 1 dimension, the dimension of Z
For M × 1 is tieed up.
4. the estimate problem that unknown impulsive noise vector e will be asked for is converted into one estimate of e of solution so that l2-
l1Norm type object functionThe minimum problem of value, then willInF (e) is expressed as with functional form,Wherein, symbol " | | | |2" it is to ask for l2Norm
Symbol, symbolTo ask for l2Norm square, symbol " | | | |1" it is to ask for l1Norm sign, FTRepresent known M × N
The DFT matrix of dimension, τ represents regularization parameter, and the value of τ is the normal number for setting, and f () is function representation side
Formula.
5. an optimization subproblem is solved by iterative manner to obtain an estimate of e so that l2-l1Norm type target
FunctionValue it is minimum, detailed process is:
5. the value for -1, setting regularization parameter τ is τ=0.1 | | (FT)HZ||∞, the initial value for making the estimate of e is e0, e0
=((FT)HFT)-1(FT)HZ, and make k represent iterations, the initial value of k is 1, wherein, symbol " | | | |∞" infinite to ask for
Norm sign, (FT)HIt is FTAssociate matrix, ((FT)HFT)-1It is (FT)HFTInverse matrix.
5. the evaluation points α of a kth time iteration -2, is definedk,Make againWherein, (Sk)HIt is SkAssociate matrix, Sk=ek-ek-1, ekObtained after representing kth time iteration
E estimate, ek-1The estimate of the e obtained after -1 iteration of kth is represented,Table
Show and ask for f (e after kth time iterationk) first derivative values,F (e after -1 iteration of kth are asked in expressionk-1) single order lead
Numerical value,
5. -3, according to αkAnd uk, the estimate of the e obtained after kth time iteration is calculated, it is designated as ek,Wherein, herein willReferred to as optimization is asked
Topic, max () is to take max function, symbol " | | " it is the symbol that takes absolute value,Represent ukIn i-th element, 1≤i≤
N。
5. -4, judgeWhether set up, if set up, perform step 5. -5;Otherwise, k=k+1 is made,
Step is then back to 5. -2 to continue executing with, wherein, symbol " | | | |2" it is to ask for l2Norm sign, ε is the minimum normal of setting
Number, takes ε=10 in the present embodiment-5, "=" in k=k+1 is assignment.
5. -5, terminate iterative process, obtain the final estimate of e, be designated asWherein,In
"=" is assignment.
6. the final estimate of e is subtracted with YThe frequency-domain expression obtained by DFT, that is, pressed down
Received signal vector after impulse noise interference processed.
By it is following emulation with further illustrate ofdm system of the invention pulse interference suppression method feasibility and
Validity.
It is the OFDM Systems modulated using QPSK to choose analogue system, and is madeSimulated environment is
MATLAB2011b, the computer for using is 2.16GHz to possess Intel Pentium Dual Cores processor, the interior 2.96GB, dominant frequency of saving as
Computer, its operating system be Windows XP SP3, the unit (decibel) of signal to noise ratio is represented with dB.Employed in emulation
Impulsive noise (overlaps index and takes 0.2,0.01) power specific factor takes to produce according to Myddelton Class A models.
Fig. 2 is given when N takes 16,64,128,256,512 and 1024 respectively, and signal to noise ratio be -5dB when, using this hair
Bright method is consumed with the existing impulse noise suppression method based on SBL when ofdm system impulse noise mitigation is carried out
The comparable situation of the average operating time of CPU.From figure 2 it can be seen that the average fortune of the CPU consumed using the inventive method
The row time will be significantly less than the average operating time of the CPU consumed using the existing impulse noise suppression method based on SBL,
I.e. the inventive method substantially reduces the time of impulse noise mitigation interference, and practicality is higher.
Fig. 3 is given as N=256, under different state of signal-to-noise, using the inventive method and existing based on SBL
Impulse noise suppression method after, the symbol error probability situation of change of ofdm system.Set send OFDM symbol sum as
5000, signal to noise ratio excursion is -10~30dB.From figure 3, it can be seen that with the existing impulse noise mitigation based on SBL
Method is compared, and when signal to noise ratio is less than 10dB, the performance of the inventive method is slightly below the existing impulse noise mitigation based on SBL
Method, and the performance of the inventive method is then substantially better than the existing impulsive noise suppression based on SBL when signal to noise ratio is more than 10dB
Method processed, and ofdm system using the inventive method has more than relative to the ofdm system for not carrying out impulse noise mitigation
The performance gain of 5dB.