CN102447657A - Digital diminishing and compensating method for IQ (Image Quality) imbalance - Google Patents

Abstract

The invention provides a digital diminishing and compensating method for IQ (Image Quality) imbalance. The digital diminishing and compensating method comprises the following steps of: (a) preprocessing data of a sampling point so as to eliminate pilot frequency; (b) updating gain factors and phase factors; and (c) performing IQ correction based on the updated gain factors and phase factors.

Description

Subdue compensation method to the unbalanced numeral of IQ

Technical field

The present invention relates to the analog radio frequency and the digital baseband design combined optimization technology of the SoC chip in the mobile communication, relate in particular to a kind of unbalanced numeral of IQ that is directed against and subdue compensation method.

Background technology

The framework of zero intermediate frequency formula has been adopted in the increasing communications field, can practice thrift device cost like this, but the corresponding problems such as IQ imbalance of having brought.At first provide the unbalanced formulate of IQ below.

Suppose that before the transmitting terminal up-conversion certain some signal indication is (a+j*b), after eating dishes without rice or wine, can be expressed as (supposing that transmitting terminal does not have the IQ imbalance) again through up-conversion:

Re{(a+j*b)*(coswt+j*sin?wt)}

＝a*coswt-b*sinwt (1)

At receiving terminal, carry out down-conversion, exist this moment IQ uneven, establish one the road and multiply by (1+ β) cos wt.One the tunnel takes advantage of-sin (wt+ φ).If g is the uneven factor of gain, φ is the unbalance in phase factor, does not consider the carrier phase deviation between the transmitter and receiver here, because this deviation can be through the channel estimating compensation, so locate not list in ofdm system.

Real part separates:

$(a*\cos \mathrm{wt}-b*\sin \mathrm{wt})*(1+\mathrm{\β})\cos \mathrm{wt}$

$=(1+\mathrm{\&beta;})*a*\frac{1+\mathrm{<figure-callout\; id="2"\; label="cos"\; filenames="HSA00000297770600032.png"\; state="\{\{state\}\}">cos</figure-callout>}2\mathrm{wt}}{2}-(1+\mathrm{\&beta;})*b*\frac{\sin 2\mathrm{wt}}{2}---\left(2\right)$

After filter filtered 2 frequency multiplication components, (2) formula can be expressed as:

$I=(1+\mathrm{\&beta;})*a*\frac{1}{2}---\left(3\right)$

Imaginary component leaves:

$-(a*\cos \mathrm{wt}-b*\sin \mathrm{wt})*\sin (\mathrm{wt}+\mathrm{\&phi;})$

$=-\frac{1}{2}*a*\sin (2\mathrm{wt}+\mathrm{\&phi;})-\frac{1}{2}*a*\sin \mathrm{\&phi;}-\frac{1}{2}*b*\cos (2\mathrm{wt}+\mathrm{\&phi;})+\frac{1}{2}*b*\cos \mathrm{\&phi;}---\left(4\right)$

After filter filtered the separation of 2 frequencys multiplication, (4) formula can be expressed as

$Q=\frac{1}{2}*(b*\cos \mathrm{\&phi;}-a*\sin \mathrm{\&phi;})---\left(5\right)$

Can find out that from formula 3 and formula 5 the IQ imbalance will be brought infringement to the amplitude and the phase place of signal, when modulation system was got over high-order, the infringement that is brought was bigger.Therefore the IQ imbalance is the problem that must solve in the zero intermediate frequency reciver.

In addition, equally also there is the unbalanced phenomenon of IQ in the front end of transmitter.The imbalance of transmitter and receiver might occur strengthening stack, and this can produce tangible influence to systematic function.

Uneven correction of IQ commonly used and method of estimation mainly are divided into the method for frequency domain and time domain.The method that frequency domain is handled relates to the process of matrix inversion through regular meeting, so amount of calculation is bigger.Then amount of calculation is less relatively for the method for time domain, but estimated time is longer.

At present, still do not have and carry out in the MB-OFDM system that IQ is uneven to be corrected and estimate related patent U.S. Patent No..

Summary of the invention

To the above-mentioned deficiency of prior art, the present invention combine MB-OFDM the signal characteristic novelty a kind of estimation and backoff algorithm based on the IQ balance of time domain of novelty proposed.The present invention is intended to and can maximum reduces the infringement that the IQ imbalance causes system.In addition, this method can also be used for the IQ imbalance of compensation transmitter.

In addition, through correcting receiver, the scheme of correcting transmitter again, the imbalance that can correct transmitter and receiver earlier.

Particularly, the invention provides a kind of unbalanced numeral of IQ that is directed against and subdue compensation method, comprising: (a). the sampling number certificate is carried out preliminary treatment, to eliminate pilot tone; (b). upgrade gain factor and phase factor; And (c). carry out the IQ correction based on gain factor after upgrading and phase factor.

According to a preferred embodiment, to subdue in the compensation method in above-mentioned numeral, said pretreated step comprises that the signal with 6 the OFDM symbols of being separated by in the said sampling number certificate subtracts each other.

According to a preferred embodiment, subdue in the compensation method in above-mentioned numeral, before said preliminary treatment, also comprise: initialization gain factor and phase factor; And carry out IQ according to initialized gain factor and phase factor and proofread and correct.

According to a preferred embodiment, subdue in the compensation method in above-mentioned numeral, repeat said step (a)～(c), through recurrence, estimate the uneven factor of gain and the unbalance in phase factor.

According to a preferred embodiment, to subdue in the compensation method in above-mentioned numeral, said sampling number is according to being expressed as: I+j*Q, said IQ checking list is shown:

I′＝I，

Q′＝Q*G _N+I*P _N，

Wherein, G _NThe gain factor of expression after N renewal, P _NThe phase factor of expression after N renewal.

According to a preferred embodiment, to subdue in the compensation method in above-mentioned numeral, said renewal gain factor and phase factor are based on that real part energy, imaginary part energy and the real imaginary part product of the certain symbol that adds up carry out.

Should be appreciated that the above generality of the present invention is described and following detailed description all is exemplary and illustrative, and be intended to further explanation is provided for as claimed in claim the present invention.

Description of drawings

Comprise that accompanying drawing is for providing the present invention further to be understood, they are included and are constituted the application's a part, and accompanying drawing shows embodiments of the invention, and play the effect of explaining the principle of the invention with this specification.In the accompanying drawing:

Fig. 1 schematically shows the unbalanced numeral of IQ that is directed against of the present invention and subdues the flow chart of compensation method.

Fig. 2 shows the flow chart according to a preferred embodiment.

Fig. 3 schematically shows application numeral of the present invention and subdues the system module Organization Chart of compensation method.

Fig. 4 shows the situation that the signal of 6 the OFDM symbols of being separated by in the pre-treatment step subtracts each other.

Fig. 5 A and 5B show the schematic and phase estimation deviation sketch map of gain estimation deviation respectively.

Fig. 6 schematically shows the unbalanced strategic process figure of IQ that uses loop to return (Loopback) scheme compensation transmitter and receiver.

Embodiment

Now with embodiments of the present invention will be described by referring to the drawings in detail.

Fig. 1 shows the flow chart of the most basic technical scheme of the present invention.As shown in Figure 1; Of the present inventionly subdue compensation method and comprise: sampling number according to carrying out preliminary treatment (step 101), is upgraded gain factor and carries out IQ correction (step 103) with phase factor (step 102) and based on gain factor after upgrading and phase factor to the unbalanced numeral of IQ.Wherein, carry out pretreated main purpose and be that wherein pilot tone is the known signal that is distributed in the OFDM data symbol to eliminate the influence of pilot tone (pilot).Particularly, this pretreated step can through with sampling number according in the signal of 6 the OFDM symbols of being separated by subtract each other and realize.In fact, pilot tone can have influence on the statistical property of I and Q signal, and two topmost factors that influence the value of pilot tone are tfc (time-frequency-code, temporal frequency code) and tds (time-domain-spreading, time domain expansion).Though the pairing pilot tone of different tfc and tds has nothing in common with each other, they all have a common characteristic, and promptly each 6 OFDM symbol will repeat.Therefore, the signal subtraction of six symbols of will being separated by just can eliminate pilot tone.

Fig. 2 shows a preferred embodiment of the present invention.This embodiment has used the principle of method shown in Figure 1.Wherein, method shown in Figure 2 comprises:

Step 201, initialization gain factor and phase factor;

Step 202 is carried out the IQ correction according to initialized gain factor and phase factor;

Step 203 is carried out preliminary treatment to the sampling number certificate;

Step 204 is upgraded gain factor and phase factor;

Step 205 is carried out the IQ correction based on gain factor after upgrading and phase factor;

Step 206 judges whether to have carried out N IQ and proofreaies and correct, and wherein N is a preset value;

Step 207 after executed N time, estimates the uneven factor of gain and the unbalance in phase factor through recurrence.

Below, in conjunction with concrete applied environment and give concrete data and briefly introduce a kind of implementation according to principle of the present invention.But what those having ordinary skill in the art will appreciate that is that following manner does not constitute any restriction to basic principle of the present invention.

At first, the following step position of module in virtual system of being correlated with is as shown in Figure 3.

Suppose initial gain factor G ₀Value and phase factor P ₀Value is initialized as 1 and 0 respectively.

Then, time domain data is carried out the 1st IQ that takes turns correct, sampling number is then corrected formula and is according to being I+j*Q before suppose to correct:

I′＝I

Q′＝Q*G _N+I*P _N

Wherein, G _NThe gain factor of expression after N renewal, P _NThe phase factor of expression after N renewal.Data after the correction are used as follow-up other resume module.

Before doing the uneven estimation of IQ, I ' and Q ' are done preliminary treatment.In order to shield the differentia influence of tfc and tds, be example, wherein R with 12 symbol datas _n[l] is the l point data of n symbol.

Multiply by the sign bit factor of this symbol earlier.

R1 _n[l]＝p[n]×R _n[l]

R1n [l] is a primary signal, after preliminary treatment, just draws R2.6 symbols of being separated by again subtract each other, and only get odd number symbols here.

R2[165·n+l]＝R1[n][l]-R1[n+6][l]

This step please refer to Fig. 4, and Fig. 4 only shows R1.Fig. 4 shows the frequency-hopping mode of three kinds of different TFC (time-frequency code, Time Frequency Code), and it is intended to show the tfc of appearance whatsoever, can eliminate pilot tone through preprocess method of the present invention.

Then, can select the add up real part energy of certain symbol of following equality for use, the imaginary part energy, real imaginary part product:

$A=\underset{k=0}{\overset{\mathrm{width}}{\mathrm{\&Sigma;}}}{\left(\mathrm{real}\left({R2}_k\right)\right)}^2$

$B=\underset{k=0}{\overset{\mathrm{width}}{\mathrm{\&Sigma;}}}{\left(\mathrm{imag}\left({R2}_k\right)\right)}^2$

$C=\underset{k=0}{\overset{\mathrm{width}}{\mathrm{\&Sigma;}}}(\mathrm{real}\left({R2}_k\right)*\mathrm{imag}\left({R2}_k\right))$

E _g＝sign(A-B)

S _g＝[abs(A-B)/A]＞hold1

E _p＝-sign(C)

S _p＝[abs(C)/A]＞hold2

Wherein hold1 and hold2 can choose a smaller value, to be used for the judging convergence sign, for example choose 1/64.

Real part energy, imaginary part energy and real imaginary part product according to the above-mentioned certain symbol that adds up upgrade gain factor and phase factor.This updating steps can be selected following mode for use:

G _N+1＝G _N+E _g×U _g

P _N+1＝P _N+E _p×U _p

Wherein, Ug is the step-length of gain adjustment, and Up is the step-length of phase place adjustment.It is different that Ug and Up get when restraining attitude and stable state fast, when restraining attitude fast, gets 1/128.When stable state, get 1/512.The step-length of convergence attitude is bigger fast, can accelerate convergence rate.If select little step-length (for example 1/512) at the very start for use, will cause convergence time longer.For example, State Control is: when Sg is continuous when being 02 times, thinking that gain has got into stable state, when Sp is continuous when being 02 times, thinking that phase place has got into stable state.The window that adds up of convergence attitude and stable state is long all different with factor renewal step-length fast.

Just can use G subsequently _N+1And P _N+1The IQ that carries out a new round corrects.

Finally, through recurrence, just can estimate the uneven factor of gain and the unbalance in phase factor.

With the said method is example, and Fig. 5 A has provided under different signal to noise ratio (snr)s with Fig. 5 B, the estimation effect that has or not the algorithm pre-treatment step to be brought.In said accompanying drawing, gain deviation directly calculates with original range value, and phase deviation is calculated with the tangent value of this phase place.

In addition, the inventor of this case proves in the following manner: transmitter IQ imbalance+delay+unbalanced mixed effect of receiver IQ still is equivalent to an IQ imbalance.Specific as follows:

At first define some marks:

TX IQ is uneven: g _t, φ _t

RX IQ is uneven: g _r, φ _r

Carrier frequency: ω=2 π f _c

Baseband signal: a+j*b

Defining ∏ { acos (ω t) }=∏ { asin (ω t) }=a again is the low pass operation, and the process of carrier wave coefficient is extracted in expression.

Then the RF signal of receiving terminal is:

S _tx(t)＝g _t·a·cos(ω·t)+b·sin(ω·t+φ _t)

＝(g _t·a+sin(φ _t)·b)·cos(ω·t)+b·cos(φ _t)·sin(ω·t)

The empty two paths of signals of the reality that receives is respectively:

A＝v ₁·(a·cos(ξ ₁)+b·sin(ξ ₁))

＝v ₁·(λ ₁·(-a·sin(ξ ₂)+b·cos(ξ ₂))+λ ₂·(a·cos(ξ ₂)+b·sin(ξ ₂)))

＝v ₁·(λ ₁·A′+λ ₂·B)

Wherein

λ ₁＝-cos(ξ ₁)·sin(ξ ₂)+sin(ξ ₁)·cos(ξ ₂)＝sin(ξ ₁-ξ ₂)

λ ₂＝cos(ξ ₁)·cos(ξ ₂)+sin(ξ ₁)·sin(ξ ₂)＝cos(ξ ₁-ξ ₂)

It is thus clear that the mixing of these factors is equivalent to an IQ imbalance.In other words, the unbalanced Overlay of dual IQ still is equivalent to single IQ imbalance.Therefore, the present invention can be applied to transmitter equally, and does not need the new algorithm of exploitation in addition.

Fig. 6 schematically shows the unbalanced strategic process figure of IQ that uses loop to return (Loopback) scheme compensation transmitter and receiver.Can utilize RF to produce desirable tone signal (tone) (step 601) earlier in the receiver porch, this tone signal does not comprise the imbalance of transmitter, uses above-mentioned algorithm to correct the IQ uneven (step 602) of receiver.And then from transmitter base band generation tx-rx loop return signal, the IQ imbalance of this moment should be the imbalance (step 603) of transmitter.Reuse above-mentioned algorithm, just can estimate IQ uneven (step 604) with compensation transmitter.

Those skilled in the art can be obvious, can carry out various modifications and modification and without departing from the spirit and scope of the present invention to above-mentioned exemplary embodiment of the present invention.Therefore, be intended to that the present invention is covered and drop in appended claims and the come scope thereof to modification of the present invention and modification.

Claims (6)

1. subdue compensation method to the unbalanced numeral of IQ for one kind, comprising:

(a). the sampling number certificate is carried out preliminary treatment, to eliminate pilot tone;

(b). upgrade gain factor and phase factor; And

(c). carry out the IQ correction based on gain factor after upgrading and phase factor.

2. numeral as claimed in claim 1 is subdued compensation method, it is characterized in that, said pretreated step comprises:

The signal of 6 the OFDM symbols of being separated by in the said sampling number certificate is subtracted each other.

3. numeral as claimed in claim 1 is subdued compensation method, it is characterized in that, before said preliminary treatment, also comprises:

Initialization gain factor and phase factor; And

Carry out the IQ correction according to initialized gain factor and phase factor.

4. numeral as claimed in claim 3 is subdued compensation method, it is characterized in that,

Repeat said step (a)～(c),, estimate the uneven factor of gain and the unbalance in phase factor through recurrence.

5. numeral as claimed in claim 4 is subdued compensation method, it is characterized in that,

Said sampling number is according to being expressed as:

I+j*Q；

Said IQ checking list is shown:

I′＝I，

Q′＝Q*G _N+I*P _N

Wherein, G _NThe gain factor of expression after N renewal, P _NThe phase factor of expression after N renewal.

6. numeral as claimed in claim 1 is subdued compensation method, it is characterized in that, said renewal gain factor and phase factor are based on that real part energy, imaginary part energy and the real imaginary part product of the certain symbol that adds up carry out.

Images