CN104980376A - Transmitter-receiver joint frequency selective IQ imbalance estimation and compensation method for self-loopback structures - Google Patents

Abstract

The invention discloses a transmitter-receiver joint frequency selective IQ imbalance estimation and compensation method for self-loopback structures, comprising the following steps: aiming at amplitude distortion and phase deviation of in-phase (I) and quadrature (Q) signals at a radio-frequency front end and different frequency responses of the two signals, designing two self-loopback structures to cascade an up-conversion unit and a down-conversion unit of the signals to obtain two IQ imbalance models; using a frequency domain training sequence to obtain two cascaded equation sets through the two self-loopback structures, solving the cascaded equation sets, and respectively estimating the parameters needed for compensating the frequency points of frequency selective IQ imbalance of a transmitter and a receiver; and compensating frequency selective IQ imbalance of the transmitter and the receiver in a frequency domain according to the compensating coefficients of the frequency points. The method of the invention is low in computing complexity, and can improve the system performance obviously.

Description

A kind of unbalance estimation of transmit-receive combination frequency selectivity IQ from loopback configuration and compensation method

Technical field

The invention belongs to broadband wireless communication technique field, specifically estimate and compensation method from the frequency domain that the transmit-receive combination frequency selectivity IQ of loopback configuration is unbalance.

Background technology

Increasing equipment is provided with radio communication function, causes wireless system to turn to staple commodities market.This means that the price competition of wireless product is large, be devoted to create and pursue low cost solution.This is in the especially individual problem of multiaerial system, because they need multiple radio frequency (RF) front end.In addition, increasing wireless standard needs solution flexibly, to support multiple standard agreement simultaneously.

Direct Conversion is up-and-coming concept meeting in these demands.Because it does not need outside intermediate frequency (IF) filter and image-rejection filters, make transceiver easily integrated, miniaturized.Important using value is had in the communication system as miniaturized in mobile phone etc. at mobile terminal, integrated requirement is higher.

Common transceiver scheme can be divided into two kinds: Direct Conversion scheme and superhet scheme, and the former is also called zero intermediate frequency scheme, realizes by analog/digital quadrature demodulation.But due to the defect of physical device, in the process of Up/Down Conversion, easily cause amplitude distortion and the phase deviation of I and Q two paths of signals, namely IQ is unbalance, introduce Image interference.Meeting that IQ is unbalance causes signal constellation (in digital modulation) figure distortion, and serious restriction receiver obtains synchronous ability, makes output signal produce adjacent-channel power and leaks interference, EVM index is worsened, brings the raising of the error rate thus.In order to overcome these problems, generally there are two kinds of solutions:

1) the better RF device of performance is adopted.This method can make receiver price rise.

2) adopt digital compensation technique in base band, overcome radio frequency defect.This method can make the cost of transmitter, receiver decline along with the raising of chip technology always.Therefore be widely used.

Along with the development of Digital Signal Processing, estimate and eliminate the unbalance impact on systematic function of IQ to realize on IQ is unbalance in the digital domain.The present invention proposes a kind of unbalance estimation of transmit-receive combination frequency selectivity IQ from loopback configuration and backoff algorithm.

Technical background: the unbalance factor of definition transmitter is γ _t(n) and β _tn (), can be expressed as:

${\mathrm{\γ}}_T\left(n\right)=\frac{h_{I,T}\left(n\right)+g_T{e}^{j{\mathrm{\φ}}_T}{h}_{Q,T}\left(n\right)}{2}$ (formula 1)

${\mathrm{\β}}_T\left(n\right)=\frac{h_{I,T}\left(n\right)-g_T{e}^{j{\mathrm{\φ}}_T}{h}_{Q,T}\left(n\right)}{2}$ (formula 2)

In formula, g _tand φ _trepresent amplitude and the phase imbalance of transmitter respectively, h _{i, T}(n) and h _{q, T}n () represents the frequency response that I with Q two branch road of transmitter is different to signal respectively.In like manner, we define the unbalance factor gamma of receiver _r(n) and β _rn (), is expressed as:

${\mathrm{\γ}}_R\left(n\right)=\frac{h_{I,R}\left(n\right)+g_R{e}^{j{\mathrm{\φ}}_R}{h}_{Q,R}\left(n\right)}{2}$ (formula 3)

${\mathrm{\β}}_R\left(n\right)=\frac{h_{I,R}\left(n\right)-g_R{e}^{j{\mathrm{\φ}}_R}{h}_{Q,R}\left(n\right)}{2}$ (formula 4)

In formula, g _rand φ _rrepresent amplitude and the phase imbalance of transmitter respectively, h _{i, R}(n) and h _{q, R}n () represents the frequency response that in receiver, I with Q two branch road is different to signal respectively.

By deriving, can show that the unbalance model of the IQ of transmitter and receiver all can be expressed as:

$y\left(n\right)=\mathrm{\γ}\left(n\right)\⊗x\left(n\right)+\mathrm{\β}\left(n\right)\⊗x^{*}\left(n\right)$ (formula 5)

Wherein γ (n) and β (n) represent the unbalance factor, and x (n) is primary signal, and y (n) is the signal after unbalance, symbol represent convolution, (.) * represents conjugation.

Technical scheme: in order to realize foregoing invention object, the present invention proposes a kind of unbalance estimation of transmit-receive combination frequency selectivity IQ from loopback configuration and compensation method, estimates and compensates, comprise the steps: the frequency selectivity IQ of transmitter and receiver is unbalance

1) upconverting unit of signal and down-converter unit are directly cascaded up, formed from loopback configuration, obtain the first unbalance model; By 90 ° of phase-shifters, upconverting unit and down-converter unit are cascaded up again, form another from loopback configuration, obtain the second unbalance model.

The forms of time and space of the first unbalance model can be expressed as:

$\begin{array}{c}y\left(n\right)={\mathrm{\γ}}_R\left(n\right)\⊗s\left(n\right)+{\mathrm{\β}}_R\left(n\right)\⊗s^{*}\left(n\right)\\ ={\mathrm{\γ}}_R\left(n\right)\⊗[{\mathrm{\γ}}_T\left(n\right)\⊗x\left(n\right)+{\mathrm{\β}}_T\left(n\right)\⊗x^{*}\left(n\right)]\\ +{\mathrm{\β}}_R\left(n\right)\⊗[{\mathrm{\γ}}_T\left(n\right)\⊗x\left(n\right)+{\mathrm{\β}}_T\left(n\right)\⊗x^{*}\left(n\right){]}^{*}\\ =[{\mathrm{\γ}}_T\left(n\right)\⊗{\mathrm{\γ}}_R\left(n\right)+{{\mathrm{\β}}^{*}}_T\left(n\right)\⊗{\mathrm{\β}}_R\left(n\right)]\⊗x\left(n\right)\\ +[{\mathrm{\γ}}_R\left(n\right)\⊗{\mathrm{\β}}_T\left(n\right)+{{\mathrm{\γ}}^{*}}_T\left(n\right)\⊗{\mathrm{\β}}_R\left(n\right)]\⊗x^{*}\left(n\right)\\ ={\mathrm{\γ}}_1\left(n\right)\⊗x\left(n\right)+{\mathrm{\β}}_1\left(n\right)\⊗x^{*}\left(n\right)\end{array}$ (formula 6)

In formula, x (n) is for transmitting, and y (n) is Received signal strength, and s (n) is the Equivalent Base-Band signal after up-conversion.

The forms of time and space of the second unbalance model can be expressed as:

$\begin{array}{c}y\left(n\right)={\mathrm{\γ}}_R\left(n\right)\⊗s_1\left(n\right)+{\mathrm{\β}}_R\left(n\right)\⊗{s_1}^{*}\left(n\right)\\ ={\mathrm{\γ}}_R\left(n\right)\⊗[{\mathrm{j\γ}}_T\left(n\right)\⊗x\left(n\right)+{\mathrm{j\β}}_T\left(n\right)\⊗x^{*}\left(n\right)]\\ +{\mathrm{\β}}_R\left(n\right)\⊗[j{\mathrm{\γ}}_T\left(n\right)\⊗x\left(n\right)+{\mathrm{j\β}}_T\left(n\right)\⊗x^{*}\left(n\right){]}^{*}\\ =j[{\mathrm{\γ}}_T\left(n\right)\⊗{\mathrm{\γ}}_R\left(n\right)-{{\mathrm{\β}}^{*}}_T\left(n\right)\⊗{\mathrm{\β}}_R\left(n\right)]\⊗x\left(n\right)\\ +j[{\mathrm{\γ}}_R\left(n\right)\⊗{\mathrm{\β}}_T\left(n\right)-{{\mathrm{\γ}}^{*}}_T\left(n\right)\⊗{\mathrm{\β}}_R\left(n\right)]\⊗x^{*}\left(n\right)\\ ={\mathrm{\γ}}_2\left(n\right)\⊗x\left(n\right)+{\mathrm{\β}}_2\left(n\right)\⊗x^{*}\left(n\right)\end{array}$ (formula 7)

In formula, x (n) is for transmitting, and y (n) is Received signal strength, s ₁(n)=s (n) e ^{j pi/2}=js (n) is the output signal of up-conversion Equivalent Base-Band signal after 90 ° of phase-shifters.

Formula 6 and 7 is the expression formulas in time domain, does discrete Fourier transform (DFT), be transformed into frequency domain to it, then formula 6,7 changes into as shown in the formula form:

$\begin{array}{cc}\begin{array}{c}Y\left(k\right)=\mathrm{\γ}\left(k\right)X\left(k\right)+\mathrm{\β}\left(k\right)X^{*}(-k)\\ =\mathrm{\γ}\left(k\right)X\left(k\right)+\mathrm{\β}\left(k\right)X^{*}(N-k)\end{array}& k=\mathrm{1,2}...N\end{array}$ (formula 8)

In formula, X (k) and Y (k) is respectively corresponding frequency-region signal, and γ (k) and β (k) is corresponding frequency domain unbalance parameter, and N represents total number of frequency.

2) use frequency domain training sequence, obtain two cascade equation group through two from loopback configuration, cascade equation group is solved, estimate the parameter needed for the unbalance each frequency compensation of transmitter and receiver IQ respectively; The method for solving of the penalty coefficient of each frequency of described transmitter and receiver is:

2.1) the frequency domain unbalance parameter of transmitter computes and receiver;

If launch two sections of different domain complex sequence X ₁(k) and X ₂k (), after IQ is unbalance, the frequency-region signal received is Y ₁(k) and Y ₂(k), two equations of model frequency domain form that simultaneous is unbalance, can obtain:

$\left[\begin{array}{c}{Y}_1\left(k\right)\\ Y_2\left(k\right)\end{array}\right]=\left[\begin{array}{c}{X}_1\left(k\right),{X^{*}}_1(N-k)\\ X_2\left(k\right),{X^{*}}_2(N-k)\end{array}\right]\left[\begin{array}{c}\mathrm{\γ}\left(k\right)\\ \mathrm{\β}\left(k\right)\end{array}\right]$ (formula 9)

Formula 9 is solved, γ (k) and β (k) can be tried to achieve, be expressed as:

$\left[\begin{array}{c}\mathrm{\γ}\left(k\right)\\ \mathrm{\β}\left(k\right)\end{array}\right]={\left[\begin{array}{c}{X}_1\left(k\right),{X^{*}}_1(N-k)\\ X_2\left(k\right),{X^{*}}_2(N-k)\end{array}\right]}^{-1}\left[\begin{array}{c}{Y}_1\left(k\right)\\ Y_2\left(k\right)\end{array}\right]$ (formula 10)

For two kinds from loopback configuration, use above-mentioned frequency domain training sequence by two kinds of unbalance models, obtain corresponding frequency domain unbalance parameter γ respectively ₁(k), β ₁(k) and γ ₂(k), β ₂(k);

2.2) calculate the penalty coefficient of each frequency of Receiver And Transmitter, concrete formula is:

$\left\{\begin{array}{c}{W}_R\left(k\right)=\frac{{\mathrm{\β}}_1\left(k\right)-{\mathrm{\β}}_2\left(k\right)/j}{{{\mathrm{\γ}}_1}^{*}(N-k)+{{\mathrm{\γ}}_2}^{*}(N-k)/j}\\ w_T\left(k\right)=\frac{{\mathrm{\β}}_1\left(k\right)+{\mathrm{\β}}_2\left(k\right)/j}{{\mathrm{\γ}}_1\left(k\right)+{\mathrm{\γ}}_2\left(k\right)/j}\end{array}\right.$ (formula 11)

In formula, W _tk () is the penalty coefficient of each frequency of transmitter, W _rk () is the penalty coefficient of each frequency of receiver.

3) according to the penalty coefficient of each frequency, compensate the frequency selectivity IQ of transmitter and receiver is unbalance respectively at frequency domain, concrete grammar is:

Before channel estimation/equalization, carry out precompensation to transmitting at frequency domain, concrete formula is:

Y _pre(k)=X (k)-W _t(k) X ^*(N-k) (formula 12)

Wherein, X (k) for transmitting, Y _prek () is the signal after precompensation; If we estimate W _t(k), and carry out precompensation to transmitting before channel estimation/equalization, the Image interference of the signal that up-conversion causes can be it often fully compensated, and remaining part by channel estimation/equalization time compensated.

Before channel estimation/equalization, compensate to received signal, concrete formula is:

Y _comp(k)=Y (k)-W _r(k) Y ^*(N-k) (formula 13)

Wherein, Y (k) is Received signal strength, Y _compk () is the signal after compensation.If we estimate W _r(k), and before channel estimation/equalization, signal is compensated, the Image interference that down-conversion causes can be it often fully compensated, and remaining part by channel estimation/equalization time compensated.Thus the IQ of transmitter and receiver is unbalance is compensated under the scheme of the present invention.

Beneficial effect: the present invention propose the unbalance estimation of transmit-receive combination frequency selectivity IQ from loopback configuration and and compensation method, compensate the frequency selectivity IQ of transmitter and receiver is unbalance before formal wireless communication procedure starts, under the expense of low complex degree, eliminate the unbalance performance impact brought to systematic function, thus realize more effective communication.Find out from simulation result, contrast through the unbalance and uncompensated situation of frequency selectivity IQ, the solution of the present invention makes the performance of whole system significantly enhance, and close to the situation not having IQ unbalance.

Summary of the invention

Goal of the invention: the amplitude distortion of IQ two paths of signals existed for radio-frequency front-end and phase deviation problem and different frequency response problems, the invention provides a kind of unbalance estimation of transmit-receive combination frequency selectivity IQ from loopback configuration and compensation method, by design from loopback configuration, the unbalance parameter of transceiver is estimated at frequency domain simultaneously, then according to the parameter that estimates, to transmitting terminal and the IQ of receiving terminal is unbalance compensates respectively.

Accompanying drawing explanation

Fig. 1 be the embodiment of the present invention adopt from loopback configuration block diagram;

Fig. 2 is method flow diagram of the present invention;

Fig. 3 is that signal of the present invention is at the unbalance block diagram of the IQ of frequency domain;

Fig. 4 is the IQ imbalance compensation block diagram of signal of the present invention at frequency domain;

Fig. 5 is the performance comparison figure in the specific embodiment of the invention, and simulation parameter is 4 transmitting antennas, 4 reception antennas, single carrier, and 16QAM (Quadrature Amplitude Modulation, QAM) modulates;

Fig. 6 is another the performance comparison figure in the specific embodiment of the invention, and simulation parameter is 4 transmitting antennas, 4 reception antennas, single carrier, and 64QAM (Quadrature Amplitude Modulation, QAM) modulates.

Embodiment

Below for millimeter wave WLAN (wireless local area network) (IEEE 802.11aj), be described in further detail from loopback configuration estimation and compensation method transmit-receive combination frequency selectivity IQ of the present invention is unbalance by reference to the accompanying drawings.Should understand these embodiments to be only not used in for illustration of the present invention and to limit the scope of the invention, after having read the present invention, the amendment of those skilled in the art to the various equivalent form of value of the present invention has all fallen within right appended by the application.

In the embodiment of the present invention, IEEE 802.11aj supports two kinds of bandwidth 540MHz and 1080MHz, and we select 540MHz.In order to simulation comparison, the present embodiment defines simulation parameter in table 1.

Table 1 simulation parameter is arranged

As shown in Figure 2, for millimeter wave WLAN (wireless local area network) (IEEE 802.11aj), the unbalance estimation of a kind of IQ from loopback configuration transmit-receive combination and compensation method, comprise the following steps:

(1) step 1: it is unbalance that transmitter and receiver all exists IQ, according to list of references, the unbalance factor of definition transmitter is γ _t(n) and β _tn (), is expressed as:

${\mathrm{\γ}}_T\left(n\right)=\frac{h_{I,T}\left(n\right)+g_T{e}^{j{\mathrm{\φ}}_T}{h}_{Q,T}\left(n\right)}{2}$ (formula 1)

${\mathrm{\β}}_T\left(n\right)=\frac{h_{I,T}\left(n\right)-g_T{e}^{j{\mathrm{\φ}}_T}{h}_{Q,T}\left(n\right)}{2}$ (formula 2)

Wherein, g _tand φ _trepresent amplitude and the phase imbalance of transmitter respectively, h _{i, T}(n) and h _{q, T}n () represents the frequency response that in transmitter, I with Q two branch road is different to signal respectively.In this example, g _tfor 1dB, φ _tbe 3 °, h _{i, T}(n)=[1,0.04 ,-0.03], h _{q, T}(n)=[1 ,-0.04 ,-0.03].

In like manner, we define the unbalance factor gamma of receiver _r(n) and β _rn (), is expressed as:

${\mathrm{\γ}}_R\left(n\right)=\frac{h_{I,R}\left(n\right)+g_R{e}^{j{\mathrm{\φ}}_R}{h}_{Q,R}\left(n\right)}{2}$ (formula 3)

${\mathrm{\β}}_R\left(n\right)=\frac{h_{I,R}\left(n\right)-g_R{e}^{j{\mathrm{\φ}}_R}{h}_{Q,R}\left(n\right)}{2}$ (formula 4)

Wherein, g _rand φ _rrepresent amplitude and the phase imbalance of transmitter respectively, h _{i, R}(n) and h _{q, R}n () represents the frequency response that in receiver, I with Q two branch road is different to signal respectively.In this example, g _rfor 1dB, φ _rbe 3 °, h _{i, R}(n)=[1,0.05], h _{q, R}(n)=[1 ,-0.05].

By deriving, we show that the unbalance model of the IQ of transmitter and receiver all can be expressed as:

$y\left(n\right)=\mathrm{\γ}\left(n\right)\⊗x\left(n\right)+\mathrm{\β}\left(n\right)\⊗x^{*}\left(n\right)$ (formula 5)

Wherein γ (n) and β (n) represent the unbalance factor, and x (n) is primary signal, and y (n) is the signal after unbalance, symbol represent convolution, (.) * represents conjugation.

(2) step 2: in order to compensate the IQ of receive-transmit system is unbalance, design two kinds from loopback configuration, as shown in Figure 1, switch is changeable from loopback configuration.

The upconverting unit of signal and down-converter unit are directly cascaded up, obtain the first unbalance model, can be expressed as:

$\begin{array}{c}y\left(n\right)={\mathrm{\γ}}_R\left(n\right)\⊗s\left(n\right)+{\mathrm{\β}}_R\left(n\right)\⊗s^{*}\left(n\right)\\ ={\mathrm{\γ}}_R\left(n\right)\⊗[{\mathrm{\γ}}_T\left(n\right)\⊗x\left(n\right)+{\mathrm{\β}}_T\left(n\right)\⊗x^{*}\left(n\right)]\\ +{\mathrm{\β}}_R\left(n\right)\⊗{[{\mathrm{\γ}}_T\left(n\right)\⊗x\left(n\right)+{\mathrm{\β}}_T\left(n\right)\⊗x^{*}\left(n\right)]}^{*}\\ =[{\mathrm{\γ}}_T\left(n\right)\⊗{\mathrm{\γ}}_R\left(n\right)+{{\mathrm{\β}}^{*}}_T\left(n\right)\⊗{\mathrm{\β}}_R\left(n\right)]\⊗x\left(n\right)\\ +[{\mathrm{\γ}}_R\left(n\right)\⊗{\mathrm{\β}}_T\left(n\right)+{{\mathrm{\γ}}^{*}}_T\left(n\right)\⊗{\mathrm{\β}}_R\left(n\right)]\⊗x^{*}\left(n\right)\end{array}$ (formula 14)

In formula, x (n) is for transmitting, s (n) for the Equivalent Base-Band signal after up-conversion, y (n) be Received signal strength.Here, we suppose:

$\left\{\begin{array}{c}{\mathrm{\γ}}_T\left(n\right)\⊗{\mathrm{\γ}}_R\left(n\right)+{{\mathrm{\β}}^{*}}_T\left(n\right)\⊗{\mathrm{\β}}_R\left(n\right)={\mathrm{\γ}}_1\left(n\right)\\ {\mathrm{\γ}}_R\left(n\right)\⊗{\mathrm{\β}}_T\left(n\right)+{{\mathrm{\γ}}^{*}}_T\left(n\right)\⊗{\mathrm{\β}}_R\left(n\right)={\mathrm{\β}}_1\left(n\right)\end{array}\right.$ (formula 15)

Then formula 14 can be reduced to:

$y\left(n\right)={\mathrm{\γ}}_1\left(n\right)\⊗x\left(n\right)+{\mathrm{\β}}_1\left(n\right)\⊗x^{*}\left(n\right)$ (formula 16)

The upconverting unit of signal and down-converter unit are cascaded up by 90 ° of phase-shifters, form another from loopback configuration, obtain the second unbalance model, can be expressed as:

$\begin{array}{c}y\left(n\right)={\mathrm{\γ}}_R\left(n\right)\⊗s_1\left(n\right)+{\mathrm{\β}}_R\left(n\right)\⊗{s_1}^{*}\left(n\right)\\ ={\mathrm{\γ}}_R\left(n\right)\⊗[{\mathrm{j\γ}}_T\left(n\right)\⊗x\left(n\right)+{\mathrm{j\β}}_T\left(n\right)\⊗x^{*}\left(n\right)]\\ +{\mathrm{\β}}_R\left(n\right)\⊗{[j{\mathrm{\γ}}_T\left(n\right)\⊗x\left(n\right)+{\mathrm{j\β}}_T\left(n\right)\⊗x^{*}\left(n\right)]}^{*}\\ =j[{\mathrm{\γ}}_T\left(n\right)\⊗{\mathrm{\γ}}_R\left(n\right)-{{\mathrm{\β}}^{*}}_T\left(n\right)\⊗{\mathrm{\β}}_R\left(n\right)]\⊗x\left(n\right)\\ +j[{\mathrm{\γ}}_R\left(n\right)\⊗{\mathrm{\β}}_T\left(n\right)-{{\mathrm{\γ}}^{*}}_T\left(n\right)\⊗{\mathrm{\β}}_R\left(n\right)]\⊗x^{*}\left(n\right)\end{array}$ (formula 17)

In formula, s ₁(n)=s (n) e ^{j pi/2}=js (n) is the output signal of up-conversion Equivalent Base-Band signal after 90 ° of phase-shifters, and x (n) is for transmitting, and y (n) is final Received signal strength.Suppose:

$\left\{\begin{array}{c}j[{\mathrm{\γ}}_T\left(n\right)\⊗{\mathrm{\γ}}_R\left(n\right)-{{\mathrm{\β}}^{*}}_T\left(n\right)\⊗{\mathrm{\β}}_R\left(n\right)]={\mathrm{\γ}}_2\left(n\right)\\ {j[\mathrm{\γ}}_R\left(n\right)\⊗{\mathrm{\β}}_T\left(n\right)+{{\mathrm{\γ}}^{*}}_T\left(n\right)\⊗{\mathrm{\β}}_R\left(n\right)]={\mathrm{\β}}_2\left(n\right)\end{array}\right.$ (formula 18)

Then formula 17 can be reduced to:

$y\left(n\right)={\mathrm{\γ}}_2\left(n\right)\⊗x\left(n\right)+{\mathrm{\β}}_2\left(n\right)\⊗x^{*}\left(n\right)$ (formula 19)

Formula 16 and 19 is the expression formulas in time domain, does discrete Fourier transform (DFT), be transformed into frequency domain to it, then formula 16,19 changes formula form into:

$\begin{array}{cc}\begin{array}{c}Y\left(k\right)=\mathrm{\γ}\left(k\right)X\left(k\right)+\mathrm{\β}\left(k\right)X^{*}(-k)\\ =\mathrm{\γ}\left(k\right)X\left(k\right)+\mathrm{\β}\left(k\right)X^{*}(N-k)\end{array}& k=\mathrm{1,2}...N\end{array}$ (formula 8)

In formula, X (k) and Y (k) is respectively corresponding frequency-region signal, and γ (k) and β (k) is corresponding frequency domain unbalance parameter, and N represents total number of frequency, and N equals 256 in this example.The corresponding unbalance block diagram of frequency domain is as Fig. 3.

(3) step 3: in IEEE 802.11aj system, we are for choosing two groups of different domain complex sequences in channel estimating and balanced targeting sequencing, sequence length is all 256.Suppose that the different frequency domain training sequence of transmitting two sections is X ₁(k) and X ₂k (), after IQ is unbalance, the signal received is Y ₁(k) and Y ₂(k), simultaneous two equations, can obtain:

$\left[\begin{array}{c}{Y}_1\left(k\right)\\ Y_2\left(k\right)\end{array}\right]=\left[\begin{array}{c}{X}_1\left(k\right),{X^{*}}_1(N-k)\\ X_2\left(k\right),{X^{*}}_2(N-k)\end{array}\right]\left[\begin{array}{c}\mathrm{\γ}\left(k\right)\\ \mathrm{\β}\left(k\right)\end{array}\right]$ (formula 9)

Formula 9 is solved, γ (k) and β (k) can be tried to achieve, be expressed as:

$\left[\begin{array}{c}\mathrm{\γ}\left(k\right)\\ \mathrm{\β}\left(k\right)\end{array}\right]={\left[\begin{array}{c}{X}_1\left(k\right),{X^{*}}_1(N-k)\\ X_2\left(k\right),{X^{*}}_2(N-k)\end{array}\right]}^{-1}\left[\begin{array}{c}{Y}_1\left(k\right)\\ Y_2\left(k\right)\end{array}\right]$ (formula 10)

For two kinds from loopback configuration, use above-mentioned frequency domain training sequence by two kinds of unbalance models, obtain γ respectively ₁(k), β ₁(k) and γ ₂(k), β ₂(k), discrete Fourier transform (DFT) is done and simultaneous to formula 15 and formula 18, can try to achieve:

$\left\{\begin{array}{c}{W}_R\left(k\right)={\mathrm{\β}}_R\left(k\right)/{{\mathrm{\γ}}^{*}}_R(N-k)=\frac{{\mathrm{\β}}_1\left(k\right)-{\mathrm{\β}}_2\left(k\right)/j}{{{\mathrm{\γ}}_1}^{*}(N-k)+{{\mathrm{\γ}}_2}^{*}(N-k)/j}\\ w_T\left(k\right){=\mathrm{\β}}_T\left(k\right)/{\mathrm{\γ}}_T\left(k\right)=\frac{{\mathrm{\β}}_1\left(k\right)+{\mathrm{\β}}_2\left(k\right)/j}{{\mathrm{\γ}}_1\left(k\right)+{\mathrm{\γ}}_2\left(k\right)/j}\end{array}\right.$ (formula 20)

(4) step 4: after estimating parameter, compensates the frequency selectivity IQ of transmitter and receiver is unbalance respectively at frequency domain, and at the signal IQ imbalance compensation block diagram of frequency domain as Fig. 4, the value of each frequency that formula 20 is tried to achieve is the penalty coefficient needed.

For the unbalance compensation of transmitter, we adopt frequency domain precompensation scheme, namely before by up-conversion, preliminary treatment is carried out to transmitting of frequency domain, the IQ of transmitting terminal can be avoided so unbalance after the radio-frequency front-end of mimo channel and receiving terminal, introduce more interference and error.

We to the precompensation transmitted of frequency domain as shown in the formula:

Y _pre(k)=X (k)-W _t(k) X ^*(N-k) (formula 12)

Wherein, X (k) for transmitting, Y _prek () is the signal after precompensation, if we estimate W _tk () be i.e. β _t(k)/γ _tk the value of (), the Image interference of the signal that up-conversion causes can be it often fully compensated, and remaining part will be compensated when the channel estimation/equalization of 802.11aj.

For the unbalance compensation of transmitter, compensation is to received signal shown below:

Y _comp(k)=Y (k)-W _r(k) Y ^*(N-k) (formula 13)

If we estimate W _rk () be i.e. β _r(k)/γ _r ^*(N-k) value, the Image interference that down-conversion causes can be it often fully compensated, and remaining part will be compensated when the channel estimation/equalization of 802.11aj.

Thus the frequency selectivity IQ of transmitter and receiver is unbalance is all compensated under loopback configuration scheme of the present invention, wireless communication system returns to operating state.Before and after compensating, the EVM value of transmitter and receiver is as shown in table 2.

EVM value before and after table 2 compensates

According to the parameter shown in table 1, under 802.11aj system, the present invention is respectively at two kinds of modulation system: 16QAM, performance simulation is carried out under 64QAM, obtain the unbalance uncompensation of IQ, have IQ imbalance compensation and without the error rate (BER) performance map (Fig. 5 in the unbalance three kinds of situations of IQ, Fig. 6), validity of the present invention is demonstrated.Find out from simulation result, contrast is not through the situation of IQ imbalance compensation, and the solution of the present invention makes the performance of whole system significantly enhance, and close to the situation not having IQ unbalance.

Claims (6)

1., from the unbalance estimation of transmit-receive combination frequency selectivity IQ and the compensation method of loopback configuration, it is characterized in that, comprise the following steps:

1) upconverting unit of signal and down-converter unit are directly cascaded up, formed from loopback configuration, obtain the first unbalance model; By 90 ° of phase-shifters, upconverting unit and down-converter unit are cascaded up again, form another from loopback configuration, obtain the second unbalance model;

2) use frequency domain training sequence, obtain two cascade equation group through two from loopback configuration, cascade equation group is solved, estimate the parameter needed for the unbalance each frequency compensation of transmitter and receiver frequency selectivity IQ respectively;

3) according to the penalty coefficient of each frequency, compensate the frequency selectivity IQ of transmitter and receiver is unbalance respectively at frequency domain.

2., as claimed in claim 1 from the unbalance estimation of transmit-receive combination frequency selectivity IQ and the compensation method of loopback configuration, two unbalance models in described step 1 construct from loopback configuration by switch is changeable.

3., as claimed in claim 1 from the unbalance estimation of transmit-receive combination frequency selectivity IQ and the compensation method of loopback configuration, it is characterized in that, the forms of time and space of described first unbalance model is:

$\begin{array}{c}y\left(n\right)={\mathrm{\γ}}_R\left(n\right)\⊗s\left(n\right)+{\mathrm{\β}}_R\left(n\right)\⊗s^{*}\left(n\right)\\ ={\mathrm{\γ}}_R\left(n\right)\⊗\[{\mathrm{\γ}}_T\left(n\right)\⊗x\left(n\right)+{\mathrm{\β}}_T\left(n\right)\⊗x^{*}\left(n\right)\]\\ +{\mathrm{\β}}_R\left(n\right)\⊗{\[{\mathrm{\γ}}_T\left(n\right)\⊗x\left(n\right)+{\mathrm{\β}}_T\left(n\right)\⊗x^{*}\left(n\right)\]}^{*}\\ =\[{\mathrm{\γ}}_T\left(n\right)\⊗{\mathrm{\γ}}_R\left(n\right)+{{\mathrm{\β}}^{*}}_T\left(n\right)\⊗{\mathrm{\β}}_R\left(n\right)\]\⊗x\left(n\right)\\ +\[{\mathrm{\γ}}_R\left(n\right)\⊗{\mathrm{\β}}_T\left(n\right)+{{\mathrm{\γ}}^{*}}_T\left(n\right)\⊗{\mathrm{\β}}_R\left(n\right)\]\⊗x^{*}\left(n\right)\\ ={\mathrm{\γ}}_1\left(n\right)\⊗x\left(n\right)+{\mathrm{\β}}_1\left(n\right)\⊗x^{*}\left(n\right)\end{array}$ (formula 6)

In formula, x (n) is for transmitting, and y (n) is Received signal strength, and s (n) is the Equivalent Base-Band signal after up-conversion, ${\mathrm{\γ}}_T\left(n\right)=\frac{h_{I,T}\left(n\right)+g_T{e}^{{\mathrm{j\φ}}_T}{h}_{Q,T}\left(n\right)}{2},{\mathrm{\β}}_T\left(n\right)=\frac{h_{I,T}\left(n\right)-g_T{e}^{{\mathrm{j\φ}}_T}{h}_{Q,T}\left(n\right)}{2},$ G _tand φ _trepresent amplitude and the phase imbalance of transmitter respectively, h _i,T(n) and h _q,Tn () represents the frequency response that I with Q two branch road of transmitter is different to signal respectively, ${\mathrm{\γ}}_R\left(n\right)=\frac{h_{I,R}\left(n\right)+g_R{e}^{{\mathrm{j\φ}}_R}{h}_{Q,R}\left(n\right)}{2},{\mathrm{\β}}_R\left(n\right)=\frac{h_{I,R}\left(n\right)-g_R{e}^{{\mathrm{j\φ}}_R}{h}_{Q,R}\left(n\right)}{2},$ G _rand φ _rrepresent amplitude and the phase imbalance of receiver respectively, h _i,R(n) and h _q,Rn () represents the frequency response that in receiver, I with Q two branch road is different to signal respectively.

4., as claimed in claim 1 from the unbalance estimation of transmit-receive combination frequency selectivity IQ and the compensation method of loopback configuration, it is characterized in that, the forms of time and space of described second unbalance model is:

$\begin{array}{c}y\left(n\right)={\mathrm{\γ}}_R\left(n\right)\⊗s_1\left(n\right)+{\mathrm{\β}}_R\left(n\right)\⊗{s_1}^{*}\left(n\right)\\ ={\mathrm{\γ}}_R\left(n\right)\⊗\[{\mathrm{j\γ}}_T\left(n\right)\⊗x\left(n\right)+{\mathrm{j\β}}_T\left(n\right)\⊗x^{*}\left(n\right)\]\\ +{\mathrm{\β}}_R\left(n\right)\⊗{\[{\mathrm{j\γ}}_T\left(n\right)\⊗x\left(n\right)+{\mathrm{j\β}}_T\left(n\right)\⊗x^{*}\left(n\right)\]}^{*}\\ =j\[{\mathrm{\γ}}_T\left(n\right)\⊗{\mathrm{\γ}}_R\left(n\right)-{{\mathrm{\β}}^{*}}_T\left(n\right)\⊗{\mathrm{\β}}_R\left(n\right)\]\⊗x\left(n\right)\\ +j\[{\mathrm{\γ}}_R\left(n\right)\⊗{\mathrm{\β}}_T\left(n\right)-{{\mathrm{\γ}}^{*}}_T\left(n\right)\⊗{\mathrm{\β}}_R\left(n\right)\]\⊗x^{*}\left(n\right)\\ ={\mathrm{\γ}}_2\left(n\right)\⊗x\left(n\right)+{\mathrm{\β}}_2\left(n\right)\⊗x^{*}\left(n\right)\end{array}$ (formula 7)

In formula, x (n) is for transmitting, and y (n) is Received signal strength, s ₁(n)=s (n) e ^{j pi/2}=js (n) is the output signal of up-conversion Equivalent Base-Band signal s (n) after 90 ° of phase-shifters, g _tand φ _trepresent amplitude and the phase imbalance of transmitter respectively, h _i,T(n) and h _q,Tn () represents the frequency response that I with Q two branch road of transmitter is different to signal respectively, ${\mathrm{\γ}}_R\left(n\right)=\frac{h_{I,R}\left(n\right)+g_R{e}^{{\mathrm{j\φ}}_R}{h}_{Q,R}\left(n\right)}{2},{\mathrm{\β}}_R\left(n\right)=\frac{h_{I,R}\left(n\right)-g_R{e}^{{\mathrm{j\φ}}_R}{h}_{Q,R}\left(n\right)}{2},$ G _rand φ _rrepresent amplitude and the phase imbalance of receiver respectively, h _i,R(n) and h _q,Rn () represents the frequency response that in receiver, I with Q two branch road is different to signal respectively.

5., as claimed in claim 1 from the unbalance estimation of transmit-receive combination frequency selectivity IQ and the compensation method of loopback configuration, it is characterized in that, in described step 2, the concrete method for solving of the penalty coefficient of each frequency of transmitter and receiver is:

2.1) the frequency domain unbalance parameter of transmitter computes and receiver;

If launch two sections of different domain complex sequence X ₁(k) and X ₂k (), after IQ is unbalance, the frequency-region signal received is Y ₁(k) and Y ₂(k), two equations of model frequency domain form that simultaneous is unbalance, can obtain:

$\left[\begin{array}{c}{Y}_1\left(k\right)\\ Y_2\left(k\right)\end{array}\right]=\left[\begin{array}{c}{X}_1\left(k\right){X^{*}}_1(N-k)\\ X_2\left(k\right){X^{*}}_2(N-k)\end{array}\right]\left[\begin{array}{c}\mathrm{\γ}\left(k\right)\\ \mathrm{\β}\left(k\right)\end{array}\right]$ (formula 9)

In formula, γ (k), β (k) represent frequency domain unbalance parameter, and N represents total number of frequency;

Formula 9 is solved, tries to achieve γ (k) and β (k), be expressed as:

$\left[\begin{array}{c}\mathrm{\γ}\left(k\right)\\ \mathrm{\β}\left(k\right)\end{array}\right]={\left[\begin{array}{c}{X}_1\left(k\right),{X^{*}}_1(N-k)\\ X_2\left(k\right),{X^{*}}_2(N-k)\end{array}\right]}^{-1}\left[\begin{array}{c}{Y}_1\left(k\right)\\ Y_2\left(k\right)\end{array}\right]$ (formula 10)

For two kinds from loopback configuration, use above-mentioned frequency domain training sequence by two kinds of unbalance models, obtain corresponding frequency domain unbalance parameter γ respectively ₁(k), β ₁(k) and γ ₂(k), β ₂(k);

2.2) calculate the penalty coefficient of each frequency of Receiver And Transmitter, concrete formula is:

$\left\{\begin{array}{c}{W}_R\left(k\right)=\frac{{\mathrm{\β}}_1\left(k\right)-{\mathrm{\β}}_2\left(k\right)/j}{{{\mathrm{\γ}}_1}^{*}(N-k)+{{\mathrm{\γ}}_2}^{*}(N-k)/j}\\ W_T\left(k\right)=\frac{{\mathrm{\β}}_1\left(k\right)+{\mathrm{\β}}_2\left(k\right)/j}{{\mathrm{\γ}}_1\left(k\right)+{\mathrm{\γ}}_2\left(k\right)/j}\end{array}\right.$ (formula 11)

In formula, W _tk () is the penalty coefficient of each frequency of transmitter, W _rk () is the penalty coefficient of each frequency of receiver.

6. as claimed in claim 1 from the unbalance estimation of transmit-receive combination frequency selectivity IQ and the compensation method of loopback configuration, it is characterized in that, described step 3 compensates the frequency selectivity IQ of transmitter and receiver is unbalance before channel estimation/equalization, and concrete grammar is:

Carry out precompensation to transmitting at frequency domain, concrete formula is:

Y _pre(k)=X (k)-W _t(k) X ^*(N-k) (formula 12)

In formula, X (k) is frequency domain transmission signal, Y _prek () is the frequency-region signal after precompensation, W _tk () is the penalty coefficient of each frequency of transmitter;

Compensate at frequency domain to received signal, concrete formula is:

Y _comp(k)=Y (k)-W _r(k) Y ^*(N-k) (formula 13)

In formula, Y (k) is frequency-domain received signal, Y _compk () is the frequency-region signal after compensation, W _rk () is the penalty coefficient of each frequency of receiver.