CN105847198A - IQ imbalance estimation and compensation method of OFDM-WLAN radio frequency testing system - Google Patents

Abstract

The invention discloses an IQ imbalance estimation and compensation method applicable to an OFDM-WLAN radio frequency testing system. The IQ imbalance estimation and compensation method comprises the steps of firstly obtaining coarse estimation of an IQ imbalance parameter by utilizing a long training sequence of a WLAN signal based on the smoothness characteristic of a channel impact response; then carrying out iterative operation by using pilot frequency information in a symbol based on a minimum mean square error criterion so as to obtain more precise amplitude and phase error estimation; and carrying out united balancing of the IQ imbalance and a channel for a signal based on the estimation result. The IQ imbalance estimation and compensation method is applicable to the condition that a channel has frequency selectivity; and compared with the traditional LMS algorithm, the IQ imbalance estimation and compensation method speeds up the convergence speed of the algorithm greatly as an initial value approximate to a steady-state solution can be obtained only by adding a small amount of computation, also can obtain a good estimation and compensation effect in the presence of a relatively small number of the symbols, and has very strong practicability.

Description

The IQ imbalance of OFDM-WLAN radio frequency test system is estimated and compensation method

Technical field

The present invention relates to the transmitter IQ imbalance in OFDM-WLAN RF consistency testing system estimate and compensation side Method, belongs to signal processing, wireless communication field.

Background technology

IQ imbalance refer to transmitter and receiver in-phase branch and orthogonal between amplitude and not the mating of phase place.? Ideally, the gain on I road and Q road is equal, phase place strict orthogonal, but the Circuits System of reality is difficult to accomplish above-mentioned ideal Situation, the uneven wave filter between the operation of transmitter and receiver end nonideal Up/Down Conversion, I road and Q road all can be in system Middle introducing IQ is uneven.

In traditional communication system, transceiver uses classical super-heterodyne architecture, and this structure has good suppression The performance of Image interference, but in the terminal, component size is even more important with the factor of cost, and Direct Conversion structure is just The super-heterodyne architecture based on above need for improved, eliminates zero intermediate frequency part, and radiofrequency signal is direct by quadrature frequency conversion Change into baseband signal.Compared to super-heterodyne architecture, Direct Conversion structure remove intermediate-frequency section and image-rejection filters from And substantially reducing volume and the power consumption of equipment, this structure use simultaneously can single chip integrated low-pass filtering and baseband signal power amplifier Realize, simplify the design of terminal, and this structure realize multiband, multi-standard launch with receive have bigger from By spending, these advantages make Direct Conversion transmitter receiver enjoy huge commercial value and become the main flow of mobile terminal design, Become the solution that OFDM terminal is good.But, present in super-heterodyne architecture, IQ imbalance problem is still present in zero Frequently in structure, and comparing super-heterodyne architecture, the IQ imbalance of Direct Conversion structure is generally more serious and is difficult to eliminate.

In an ofdm system, the unbalanced existence of IQ can introduce Image interference thus cause the bit error rate of system to rise.Newly Generation communication system tends to use the modulation system of high-order to promote the handling capacity of system, and the modulation methods of high-order makes The Image interference that IQ imbalance is introduced by system is more sensitive, and the unbalanced disturbance of small IQ i.e. can make systematic function serious Decline.So, the estimation of IQ imbalance and compensation problem cause at academia and discuss widely and study.Existing algorithm is the biggest What part was studied is the IQ imbalance problem of receiver end, rarely has grinding of the IQ imbalance problem caused merely towards transmitter Study carefully, and the IQ imbalance that actually transmitter causes seriously governs the performance of system, be the important finger of RF consistency test Mark.

Summary of the invention

Goal of the invention: for the most deep enough uneven to transmitter IQ in existing research problem, the present invention proposes one The unbalanced estimation of IQ and compensation method be applicable to the transmitter of OFDM-WLAN radio frequency test system.

Technical scheme: a kind of IQ imbalance being applicable to OFDM-WLAN radio frequency test system is estimated and compensation method, including Below step:

(1) signal receiving VSA carries out serioparallel exchange, and carry out FFT translate the signals into Frequency domain；

(2) use the training sequence of signal, utilize the smoothness properties of channel that IQ imbalance parameter is carried out rough estimate, and disappear Except channel estimate in the unbalanced impact of IQ thus be equalized the initial value of sequence；

(3) pilot frequency information in symbol utilize least mean-square error (LMS) criterion be iterated computing to obtain more Accurate equalized sequence；

(4) to carry out united channel Jun Heng with IQ imbalance for sub-carrier and image component thereof.

Channel estimation results is expressed as in (2) by described step:

$\hat{h}=\mathrm{\α}diag\left\{\mathrm{\λ}\right\}+\mathrm{\β}diag\left\{\mathrm{\λ}\right\}\·LTS2$

Wherein, diag{ λ } it is real channel impulse response, α=cos (Δ φ)+j ε_TSin (Δ φ), β=ε_T cos (Δ φ)+j sin (Δ φ), with amplitude and unbalance in phase parameter ε_TRelevant with Δ φ, its estimated value is respectively as follows:

$\widehat{\mathrm{\β}}=\frac{{\hat{h}}_{k+1}-{\hat{h}}_k}{LTS{2}_{N-(k+1)+2}({\hat{h}}_{k+1}+{\hat{h}}_k)}$

Wherein, LTS2=LTS^#/ LTS, LTS are the frequency domain representation of training sequence, and subscript # represents what sequence of complex numbers was conjugated FFT result, k is subcarrier index value,Estimate for channel,Represent the real part of the number of winning the confidence,Represent the void of the number of winning the confidence Portion.

As preferably, eliminating the unbalanced impact of IQ during channel is estimated in described step (2), the frequency domain response of channel represents For:

$\widehat{\mathrm{\Λ}}=\frac{\hat{h}}{\mathrm{\α}+\mathrm{\β}\·LTS2}.$

In described step (2) after obtaining the IQ imbalance parameter rough estimate result with channel, arrange at the beginning of equalized sequence Value is:

$w_k^{\left(0\right)}=\[\begin{array}{cc}\widehat{\mathrm{\α}}\widehat{\mathrm{\Λ}}\left(k\right)& \widehat{\mathrm{\β}}\widehat{\mathrm{\Λ}}\left(k\right)\end{array}\]$

$w_{N-k+2}^{\left(0\right)}=\[\begin{array}{cc}{\left(\widehat{\mathrm{\α}}\widehat{\mathrm{\Λ}}\right(N-k+2\left)\right)}^{*}& {\left(\widehat{\mathrm{\β}}\widehat{\mathrm{\Λ}}\right(N-k+2\left)\right)}^{*}\end{array}\].$

Described step (3) utilizes the ART network iterative formula of LMS criterion to be for the pilot frequency information in symbol:

$\hat{s}\left(k\right)=w_k{z}_k$

${\hat{s}}^{*}(N-k+2)=w_{N-k+2}{z}_k$

WhereinZ (k) represents the kth subcarrier of transmitting sequence, s K () represents the kth subcarrier of receiving sequence, w_kAnd w_N-k+2For the equalized sequence updated according to LMS criterion.

Concrete, the iterative process of equalizing coefficient is:

$w_k^{(i+1)}=w_k^{\left(i\right)}+u_{LMS}{\left(z_k^{\left(i\right)}\right)}^{*}{e}_k^{\left(i\right)}$

$w_{N-k+2}^{(i+1)}=w_{N-k+2}^{\left(i\right)}+u_{LMS}{\left(z_k^{\left(i\right)}\right)}^{*}{e}_{N-k+2}^{\left(i\right)}$

Wherein, k={2 ..., N/2},WithRepresent equalized sequence and the transmitting sequence in i moment respectively,For using training sequenceKth pilot sub-carrier error signal when carrying out ith iteration,For using training sequenceThe N-k+2 pilot sub-carrier carry out ith iteration Time error signal, u_LMSThe step-length used by iterative process.

As preferably, use Normalized LMS Algorithm, iteration step length be designed as:

${\mathrm{\μ}}_{LMS}\left(i\right)=\frac{{\mathrm{\μ}}_{step}}{\left|\right|z\left(i\right)|{|}^2}$

Wherein μ_stepFor step size normalization, span is 0 ＜ μ_step＜ 2, | | z (i) | |²For receiving the energy of signal.

In described step (4), channel and IQ unbalanced joint equalization method are:

Sub-carrier and image component thereof carry out associating consideration, are split as multiple by unbalanced with IQ for channel joint equalization Solving of 2 × 2 decoupling equations, to k={2 ..., N/2} defines:

${\mathrm{\Γ}}_k=\left[\begin{array}{cc}\mathrm{\α}\mathrm{\λ}\left(k\right)& \mathrm{\β}\mathrm{\λ}\left(k\right)\\ {\mathrm{\β}}^{*}{\mathrm{\λ}}^{*}(N-k+2)& {\mathrm{\α}}^{*}{\mathrm{\λ}}^{*}(N-k+2)\end{array}\right]$

The mode then using zero forcing equalization can obtain being estimated as of frequency domain data:

${\hat{s}}_k={(\mathrm{\δ}I+{\mathrm{\Γ}}_k^{*}{\mathrm{\Γ}}_k)}^{-1}{\mathrm{\Γ}}_k^{*}{z}_k$

Wherein δ ＞ 0 is that normalization factors is to resist Г_kFor the situation of ill-condition matrix, I is unit battle array.

Beneficial effect: a kind of IQ imbalance being applicable to OFDM-WLAN radio frequency test system disclosed by the invention estimate with Compensation method, the smoothness properties being first depending on channel impulse response utilizes the long training sequence of WLAN signal to obtain IQ imbalance ginseng The rough estimate of number, then according to minimum mean square error criterion, uses the pilot frequency information in symbol to be iterated computing to obtain more For accurate amplitude and phase error estimation and phase error, and according to estimated result, signal is carried out IQ is uneven and the joint equalization of channel. The method is applicable to channel and there is the situation of frequency selectivity, and is compared to traditional LMS algorithm, only increases a small amount of fortune Calculation can obtain the initial value approaching steady state solution, accelerates convergence of algorithm speed greatly, in the less situation of number of symbols Good estimation and compensation effect can also be obtained down, there is the strongest practicality.

Accompanying drawing explanation

Fig. 1 is transmitter IQ unbalanced system model schematic of the present invention；

Fig. 2 is the equivalent block diagram of transmitter IQ unbalanced system model of the present invention；

Fig. 3 is that IQ imbalance of the present invention is estimated and Jun Heng method flow diagram；

Fig. 4 is (a) (b) planisphere comparing result figure afterwards before the signal IQ imbalance compensation using 16QAM modulation；

Fig. 5 is (a) (b) planisphere comparing result figure afterwards before the signal IQ imbalance compensation using 64QAM modulation；

Fig. 6 is the BER performance verification result figure of the inventive method.

Detailed description of the invention

The invention discloses a kind of IQ imbalance that is applied in WLAN to estimate and Jun Heng method.In order to carry out must The channel wanted is estimated and follows the trail of, and WLAN standard IEEE 802.11a/g/n/ac provides long training sequence, and at symbol Middle insertion pilot frequency information, the present invention is directed to above-mentioned known array and provides a kind of channel and IQ imbalance Combined estimator with equilibrium Method.

Below as a example by IEEE 802.11ac signal, in conjunction with accompanying drawing, the method proposed in invention is carried out the most in detail Describe in detail bright.

Accompanying drawing 1 is transmitter system when there is amplitude and unbalance in phase, within the system IQ uneven mainly by The clock of system causes, and all shows as constant in its impact of each frequency range launching signal, is referred to as the IQ unrelated with frequency Uneven.Assuming that the amplitude that caused by clock of transmitting terminal and unbalance in phase parameter are respectively ε_TWith Δ φ, then preferable time domain Signal x_L(t)=x_I(t)+jx_QT (), by after frequency up-conversion operation, the radiofrequency signal of its transmission is represented by:

x_RF(t)=(1+ ε_T)cos(ω₀T+ Δ φ) x_I(t)-(1-ε_T)sin(ω₀t-Δφ)x_Q(t) (1)

Preferably receiver is by this radiofrequency signal and x_LO(t)=exp (-j ω₀T) it is mixed, then passes through low pass filter Filter the signal after high fdrequency component to be represented by:

$\begin{array}{c}{x}_{LP}\left(t\right)=(1+{\mathrm{\ϵ}}_T)\left(\cos \right(\mathrm{\Δ}\mathrm{\φ})+j\sin (\mathrm{\Δ}\mathrm{\φ}\left)\right)x_I\left(t\right)+(1-{\mathrm{\ϵ}}_T)\left(\sin \right(\mathrm{\Δ}\mathrm{\φ})+j\cos (\mathrm{\Δ}\mathrm{\φ}\left)\right)x_Q\left(t\right)\\ ={\mathrm{\αx}}_L\left(t\right)+{\mathrm{\βx}}_L^{*}\left(t\right)\end{array}$

Wherein:

α=cos (Δ φ)+j ε_TSin (Δ φ), β=ε_T cos(Δφ)+jsin(Δφ) (2)

The equivalent block diagram of the unbalanced Model in Time Domain of transmitter IQ as shown in Figure 2, OFDM symbol s=[s (1) s (2) … s(N)]^TThrough IFFT operational transformation to time domain, and add Cyclic Prefix to the formation of symbol headGrasp through parallel-serial conversion Carrying out frequency up-conversion operation by antenna after work, by the unbalanced impact of transmitter IQ, the signal of actual transmission is represented by:The length of channel finite impulse response is less than the length of Cyclic Prefix, before therefore receiving terminal removes circulation After sewing, receiving sequence is represented by:

$\begin{array}{c}\stackrel{\‾}{z}=H^c\stackrel{\‾}{y}+\stackrel{\‾}{v}\\ =H^c(\mathrm{\α}\stackrel{\‾}{s}+\mathrm{\β}conj(\stackrel{\‾}{s}\left)\right)+\stackrel{\‾}{v}\end{array}---\left(3\right)$

Wherein, H^cFor the cyclic shift matrices of N × N size,White Gaussian noise for receiving terminal.Matrix H^cCirculation move Position characteristic makes the result of its Fourier transformation be diagonal matrix, above formula two ends carries out Fourier transform operation simultaneously, and defines Subscript # represents the FFT result that sequence of complex numbers is conjugated, and the relation between itself and former sequence FFT transformation results is represented by:

$X=\left[\begin{array}{c}X\left(1\right)\\ X\left(2\right)\\ .\\ .\\ .\\ X(N/2)\\ X(N/2+1)\\ X(N/2+2)\\ .\\ .\\ .\\ X\left(N\right)\end{array}\right]\⇒X^{\#}=\left[\begin{array}{c}{X}^{*}\left(1\right)\\ X^{*}\left(N\right)\\ .\\ .\\ .\\ X^{*}(N/2+2)\\ X^{*}(N/2+1)\\ X^{*}(N/2)\\ .\\ .\\ .\\ X^{*}\left(2\right)\end{array}\right]---\left(4\right)$

The relation between frequency domain receiving sequence z and former sequence s that can obtain is:

Z=diag (λ) (α s+ β s^#)+v (5)

Wherein, diag{ λ } it is real channel impulse response, v is white Gaussian noise.

As it is shown on figure 3, a kind of IQ imbalance being applicable to OFDM-WLAN radio frequency test system disclosed in the embodiment of the present invention Estimate and compensation method, mainly comprise the following steps:

S1: the signal receiving VSA carries out serioparallel exchange, and carry out FFT translate the signals into Frequency domain；

S2:IQ imbalance rough estimate: use the training sequence of signal, utilize the smoothness properties of channel to IQ imbalance parameter Carry out rough estimate, and eliminate the unbalanced impact of IQ during channel is estimated thus be equalized the initial value of sequence；

The uneven thin estimation of S3:IQ: the pilot frequency information in symbol utilizes LMS criterion be iterated computing to obtain more Accurate equalized sequence, and by iteration step length normalization to ensure algorithmic statement；

S4: joint equalization: it is Jun Heng with IQ imbalance that sub-carrier and image component thereof carry out united channel.

Utilize the structure in accompanying drawing 3 that channel and IQ imbalance are carried out Combined estimator and compensation, by subcarrier and its mirror image Pairing, then sub-carrier index value k={2 ..., relation below N/2} definable:

z_k=Γ_ks_k+v_k (6)

Wherein, v_kThe white Gaussian noise introduced for channel, remaining parameters is defined as follows:

$\begin{array}{c}{z}_k=\left[\begin{array}{c}z\left(k\right)\\ z^{*}(N-k+2)\end{array}\right],s_k=\left[\begin{array}{c}s\left(k\right)\\ s^{*}(N-k+2)\end{array}\right]\\ {\mathrm{\Γ}}_k=\left[\begin{array}{cc}\mathrm{\α}\mathrm{\λ}\left(k\right)& \mathrm{\β}\mathrm{\λ}\left(k\right)\\ {\mathrm{\β}}^{*}{\mathrm{\λ}}^{*}(N-k+2)& {\mathrm{\α}}^{*}{\mathrm{\λ}}^{*}(N-k+2)\end{array}\right]\end{array}---\left(7\right)$

The iterative formula utilizing LMS algorithm to define ART network is:

$\hat{s}\left(k\right)=w_k{z}_k---\left(8\right)$

${\hat{s}}^{*}(N-k+2)=w_{N-k+2}{z}_k---\left(9\right)$

Wherein w_kAnd w_N-k+2For the equalized sequence updated according to LMS criterion, its initialization sequence is obtained by following operation:

IEEE 802.11ac adds VHT-LTF field in frame head and estimates with channel with the frequency deviation carrying out necessity, its The VHT-LTF of 20MHz signal is defined as:

VHTLTF_28,28={ 1,1, LTF_left,0,LTF_right,-1,-1} (10)

Wherein,

LTF_left=1,1 ,-1 ,-1,1,1 ,-1,1 ,-1,1,1,1,1,1,1 ,-1 ,-1,1,1 ,-1,1 ,-1,1,1,1,1}

LTF_right=1 ,-1 ,-1,1,1 ,-1,1 ,-1,1 ,-1 ,-1 ,-1 ,-1 ,-1,1,1 ,-1 ,-1,1 ,-1,1 ,-1,1, 1,1,1}

Ignore effect of noise, utilize the long training sequence field receiving signal to carry out LS channel estimation, its estimated result It is represented by:

$\hat{h}=\mathrm{\α}diag\left\{\mathrm{\λ}\right\}+\mathrm{\β}diag\left\{\mathrm{\λ}\right\}\·LTS2---\left(11\right)$

Wherein, LTS2=LTS^#/ LTS, wherein LTS is the frequency domain representation of training sequence.For subcarrier index value k, (k is The forward position of LTS2 transition), have:

$\begin{array}{c}{\hat{h}}_k={\mathrm{\α\λ}}_k+{\mathrm{\β\λ}}_k\·LTS{2}_{N-k+2}\\ {\hat{h}}_{k+1}={\mathrm{\α\λ}}_{k+1}+{\mathrm{\β\λ}}_{k+1}\·LTS{2}_{N-(k+1)+2}\end{array}---\left(12\right)$

Owing to the coherence bandwidth of channel is spaced much larger than between subcarrier, therefore the frequency domain response on adjacent sub-carrier can be near Like for equal, additionally according to LTS2_Nαk+2=LTS2_N-(k+1)+2Two formulas are made difference can obtain:

$\begin{array}{c}{\hat{h}}_{k+1}-{\hat{h}}_k=\mathrm{\α}({\mathrm{\λ}}_{k+1}-{\mathrm{\λ}}_k)+\mathrm{\β}LTS{2}_{N-(k+1)+2}({\mathrm{\λ}}_{k+1}-{\mathrm{\λ}}_k)\\ \⇒\mathrm{\β}=\frac{{\hat{h}}_{k+1}-{\hat{h}}_k}{LTS{2}_{N-(k+1)+2}({\mathrm{\λ}}_{k+1}+{\mathrm{\λ}}_k)}\end{array}---\left(13\right)$

It is less that the actual amplitude in ofdm system and phase place do not mate value, therefore being estimated as of IQ imbalance parameter beta:

$\widehat{\mathrm{\β}}=\frac{{\hat{h}}_{k+1}-{\hat{h}}_k}{LTS{2}_{N-(k+1)+2}({\hat{h}}_{k+1}+{\hat{h}}_k)}---\left(14\right)$

Being estimated as of parameter alpha can be obtained according to triangle inequality relation:

Compensate there is the unbalanced channel estimation results of IQ, it is possible to obtain:

$\widehat{\mathrm{\Λ}}=\frac{\hat{h}}{\mathrm{\α}+\mathrm{\β}\·LTS2}---\left(16\right)$

Therefore, after obtaining the IQ imbalance parameter rough estimate result with channel, the initial value of equalized sequence could be arranged to:

$\begin{array}{c}{w}_k^{\left(0\right)}=\[\begin{array}{cc}\widehat{\mathrm{\α}}\widehat{\mathrm{\Λ}}\left(k\right)& \widehat{\mathrm{\β}}\widehat{\mathrm{\Λ}}\left(k\right)\end{array}\]\\ w_{N-k+2}^{\left(0\right)}=\[\begin{array}{cc}{\left(\widehat{\mathrm{\α}}\widehat{\mathrm{\Λ}}\right(N-k+2\left)\right)}^{*}& {\left(\widehat{\mathrm{\β}}\widehat{\mathrm{\Λ}}\right(N-k+2\left)\right)}^{*}\end{array}\]\end{array}---\left(17\right)$

Then, the pilot signal in IEEE 802.11ac is utilized to be iterated, IEEE 802.11ac 20MHz signal Pilot sub-carrier index value is K_Pilot=± 7, ± 21}, the value in each pilot tone is:

p_n{-21 ,-7,7,21}={ Ψ_nmod4,Ψ_(n+1)mod4,ψ_(n+2)mod4,ψ_(n+3)mod4} (18)

The linear displacement characteristic of pilot frequency information ensure that the effectiveness of equalized sequence solution, introduces time index i in the sequence, OrderWithRepresent equalized sequence and the transmitting sequence in i moment respectively, then for k={2 ..., N/2} equalizes system The iterative process of number is:

$w_k^{(i+1)}=w_k^{\left(i\right)}+u_{LMS}{\left(z_k^{\left(i\right)}\right)}^{*}{e}_k^{\left(i\right)}---\left(19\right)$

$w_{N-k+2}^{(i+1)}=w_{N-k+2}^{\left(i\right)}+u_{LMS}{\left(z_k^{\left(i\right)}\right)}^{*}{e}_{N-k+2}^{\left(i\right)}---\left(20\right)$

WhereinFor using training sequenceKth pilot sub-carrier when carrying out ith iteration Error signal,For using training sequenceThe N-k+2 pilot sub-carrier carry out Error signal during ith iteration, u_LMSThe step-length used by iterative process, uses Normalized LMS Algorithm, is defined as:

${\mathrm{\μ}}_{LMS}\left(i\right)=\frac{{\mathrm{\μ}}_{step}}{\left|\right|z\left(i\right)|{|}^2}---\left(21\right)$

Wherein μ_stepFor step size normalization, span is 0 ＜ μ_step＜ 2, | | z (i) | |²For receiving the energy of signal.Logical Cross above-mentioned iterative estimate, can obtain the accurate estimation of IQ imbalance parameter, and and then obtain channel shock response thus Carry out the frequency domain compensation of signal.It is specially and subcarrier and image component thereof are carried out associating consideration, channel is unbalanced with IQ Joint equalization is split as solving of multiple 2 × 2 decoupling equations, uses the mode of zero forcing equalization to try to achieve frequency domain data according to the following formula Estimate, thus operate at channel and the unbalanced joint equalization of IQ of the complete pair signals of frequency domain.

${\hat{s}}_k={(\mathrm{\δ}I+{\mathrm{\Γ}}_k^{*}{\mathrm{\Γ}}_k)}^{-1}{\mathrm{\Γ}}_k^{*}{z}_k---\left(22\right)$

The unbalanced compensating system of IQ is there is in the amplitude utilizing this method to estimate with unbalance in phase factor pair, Verify its systematic function, choose the MCS4 data using 16QAM modulation in IEEE 802.11ac agreement and use 64QAM modulation MCS7 data test, accompanying drawing 3 and accompanying drawing 4 are the planisphere contrast situation before and after two groups of sample datas compensate.Permissible It is evident that affected by IQ imbalance, the point on planisphere is that discrete shape is distributed in around standard planisphere point, and high-order Modulation system is affected bigger by IQ imbalance, and the data interlacing on various constellations point together thus causes erroneous judgement.Through mending Data after repaying inhibit image component, add the accuracy of channel estimation and equalization, effectively improve the abnormal of planisphere Become so that the point of planisphere is all collected near standard point, it is seen that the IQ imbalance frequency domain compensating method proposed in literary composition has non- The best compensation effect.

Fig. 6 shows that method that the present invention proposes to bit error rate performance, effectively improves system through the system of overcompensation Bit error rate performance, eliminate the floor effect that IQ imbalance causes, and when only using 5 symbols to be iterated training Good performance can be obtained.

Claims (8)

1. the IQ imbalance being applicable to OFDM-WLAN radio frequency test system is estimated and compensation method, it is characterised in that: include Below step:

(1) signal receiving VSA carries out serioparallel exchange, and carries out FFT and translate the signals into frequency Territory；

(2) use the training sequence of signal, utilize the smoothness properties of channel that IQ imbalance parameter is carried out rough estimate, and eliminate letter Road estimate in the unbalanced impact of IQ thus be equalized the initial value of sequence；

(3) pilot frequency information in symbol utilize least mean-square error (LMS) criterion be iterated computing the most accurate to obtain Equalized sequence；

(4) to carry out united channel Jun Heng with IQ imbalance for sub-carrier and image component thereof.

IQ imbalance the most according to claim 1 is estimated and compensation method, it is characterised in that: channel in described step (2) Estimated result is expressed as:

$\hat{h}=\mathrm{\α}diag\left\{\mathrm{\λ}\right\}+\mathrm{\β}diag\left\{\mathrm{\λ}\right\}\·LTS2$

Wherein, diag{ λ } it is real channel impulse response, α=cos (Δ φ)+j ε_TSin (Δ φ), β=ε_T cos(Δ φ)+j sin (Δ φ), with amplitude and unbalance in phase parameter ε_TRelevant with Δ φ, its estimated value is respectively as follows:

$\widehat{\mathrm{\β}}=\frac{{\hat{h}}_{k+1}-{\hat{h}}_k}{LTS{2}_{N-(k+1)+2}({\hat{h}}_{k+1}+{\hat{h}}_k)}$

Wherein, LTS2=LTS^#/ LTS, LTS are the frequency domain representation of training sequence, and subscript # represents the FFT that sequence of complex numbers is conjugated As a result, k is subcarrier index value,Estimate for channel,Represent the real part of the number of winning the confidence,Represent the imaginary part of the number of winning the confidence.

IQ imbalance the most according to claim 2 is estimated and compensation method, it is characterised in that: described step (2) eliminates The unbalanced impact of IQ in channel estimation, the frequency domain response of channel is expressed as:

$\widehat{\mathrm{\Λ}}=\frac{\hat{h}}{\mathrm{\α}+\mathrm{\β}\·LTS2}.$

IQ imbalance the most according to claim 3 is estimated and compensation method, it is characterised in that: described step is obtaining in (2) After taking the rough estimate result of IQ imbalance parameter and channel, the initial value arranging equalized sequence is:

$\begin{array}{c}{w}_k^{\left(0\right)}=\left[\begin{array}{cc}\widehat{\mathrm{\α}}\widehat{\mathrm{\Λ}}\left(k\right)& \widehat{\mathrm{\β}}\widehat{\mathrm{\Λ}}\left(k\right)\end{array}\right]\\ w_{N-k+2}^{\left(0\right)}=\left[\begin{array}{cc}{\left(\widehat{\mathrm{\α}}\widehat{\mathrm{\Λ}}\right(N-k+2\left)\right)}^{*}& {\left(\widehat{\mathrm{\β}}\widehat{\mathrm{\Λ}}\right(N-k+2\left)\right)}^{*}\end{array}\right]\end{array}.$

IQ imbalance the most according to claim 1 estimate and compensation method, it is characterised in that: in described step (3) for Pilot frequency information in symbol utilizes the ART network iterative formula of LMS criterion to be:

$\begin{array}{c}\hat{s}\left(k\right)=w_k{z}_k\\ {\hat{s}}^{*}(N-k+2)=w_{N-k+2}{z}_k\end{array}$

WhereinZ (k) represents the kth subcarrier of transmitting sequence, s (k) table Show the kth subcarrier of receiving sequence, w_kAnd w_N-k+2For the equalized sequence updated according to LMS criterion.

IQ imbalance the most according to claim 5 is estimated and compensation method, it is characterised in that: the iterative process of equalizing coefficient For:

$\begin{array}{c}{w}_k^{\left(i+1\right)}=w_k^{\left(i\right)}+u_{LMS}{\left(z_k^{\left(i\right)}\right)}^{*}{e}_k^{\left(i\right)}\\ w_{N-k+2}^{(i+1)}=w_{N-k+2}^{\left(i\right)}+u_{LMS}{\left(z_k^{\left(i\right)}\right)}^{*}{e}_{N-k+2}^{\left(i\right)}\end{array}$

Wherein, k={2 ..., N/2},WithRepresent equalized sequence and the transmitting sequence in i moment respectively,For using training sequenceKth pilot sub-carrier error signal when carrying out ith iteration,For using training sequenceThe N-k+2 pilot sub-carrier carry out ith iteration Time error signal, u_LMSThe step-length used by iterative process.

IQ imbalance the most according to claim 6 is estimated and compensation method, it is characterised in that: use Normalized LMS Algorithm, Iteration step length is designed as:

${\mathrm{\μ}}_{LMS}\left(i\right)=\frac{{\mathrm{\μ}}_{step}}{\left|\right|z\left(i\right)|{|}^2}$

Wherein μ_stepFor step size normalization, span is 0 ＜ μ_step＜ 2, | | z (i) | |²For receiving the energy of signal.

IQ imbalance the most according to claim 2 is estimated and compensation method, it is characterised in that: channel in described step (4) Joint equalization method unbalanced with IQ is:

Sub-carrier and image component thereof carry out associating consideration, and channel and the unbalanced joint equalization of IQ are split as multiple 2 × 2 Solving of decoupling equation, to k={2 ..., N/2} defines:

${\mathrm{\Γ}}_k=\left[\begin{array}{cc}\mathrm{\α}\mathrm{\λ}\left(k\right)& \mathrm{\β}\mathrm{\λ}\left(k\right)\\ {\mathrm{\β}}^{*}{\mathrm{\λ}}^{*}(N-k+2)& {\mathrm{\α}}^{*}{\mathrm{\λ}}^{*}(N-k+2)\end{array}\right]$

The mode then using zero forcing equalization can obtain being estimated as of frequency domain data:

${\hat{s}}_k={(\mathrm{\δ}I+{\mathrm{\Γ}}_k^{*}{\mathrm{\Γ}}_k)}^{-1}{\mathrm{\Γ}}_k^{*}{z}_k$

Wherein δ ＞ 0 is that normalization factors is to resist Γ_kFor the situation of ill-condition matrix, I is unit battle array.