CN105847198B - The IQ imbalances of OFDM-WLAN radio frequency test systems estimate and compensation method - Google Patents

Abstract

The invention discloses a kind of estimations of IQ imbalances and compensation method suitable for OFDM-WLAN radio frequency test systems, the smoothness properties for being first depending on channel impulse response obtain the rough estimate of IQ imbalance parameters using the long training sequence of WLAN signal, then according to minimum mean square error criterion, it is iterated operation using the pilot frequency information in symbol to obtain more accurate amplitude and phase error estimation and phase error, and carries out the joint equalization of IQ imbalance and channel to signal according to estimated result.The case where IQ imbalances proposed by the present invention estimation and compensation method are suitable for channel there are frequency selectivities, and it is compared to traditional LMS algorithm, only increase a small amount of operation and can be obtained the initial value for approaching steady state solution, greatly accelerate convergence speed of the algorithm, good estimation and compensation effect can also be obtained in the case where number of symbols is less, and there is very strong practicability.

Description

IQ Imbalance Estimation and Compensation Method for OFDM-WLAN RF Test System

technical field

The invention relates to a method for estimating and compensating transmitter IQ imbalance in an OFDM-WLAN radio frequency consistency testing system, and belongs to the fields of signal processing and wireless communication.

Background technique

IQ imbalance refers to the mismatch in magnitude and phase between the in-phase branch and quadrature of the transmitter and receiver. In an ideal situation, the gains of the I and Q channels are equal, and the phases are strictly orthogonal, but the actual circuit system is difficult to achieve the above ideal situation, the non-ideal up-and-down conversion operation of the transmitter and receiver, the Any unbalanced filter in between will introduce IQ imbalance in the system.

Transceivers in traditional communication systems mostly adopt the classic superheterodyne architecture, which has a good performance in suppressing image interference, but in mobile terminals, the factors of component size and cost are more important, and the direct conversion structure is exactly Based on the above requirements, the superheterodyne structure is improved, the zero-IF part is removed, and the radio frequency signal is directly converted into a baseband signal through quadrature down-conversion. Compared with the superheterodyne structure, the direct conversion structure removes the intermediate frequency part and the image interference filter, thereby greatly reducing the size and power consumption of the device. At the same time, the structure is realized by monolithically integrated low-pass filtering and baseband signal power amplifier , which simplifies the design of the terminal, and this structure has a greater degree of freedom in realizing multi-band, multi-standard transmission and reception. These advantages make the direct conversion frequency conversion transmitter and receiver enjoy great commercial value and become the mainstream of mobile terminal design. Become a good solution for OFDM terminal. However, the IQ imbalance problem in the superheterodyne structure still exists in the zero-IF structure, and compared with the superheterodyne structure, the IQ imbalance of the direct conversion structure is usually more serious and difficult to eliminate.

In an OFDM system, the existence of IQ imbalance will introduce image interference, which will increase the bit error rate of the system. In order to improve the throughput of the system, the new-generation communication system tends to adopt high-order modulation methods, and the high-order modulation method makes the system more sensitive to the image interference introduced by IQ imbalance, and the slight disturbance of IQ imbalance will make the system Performance is severely degraded. Therefore, the problem of IQ imbalance estimation and compensation has aroused extensive discussion and research in academia. Most of the existing algorithms study the IQ imbalance problem at the receiver side, and there are few studies on the IQ imbalance problem caused by the transmitter. In fact, the IQ imbalance caused by the transmitter seriously restricts the performance of the system. It is an important index for RF conformance testing.

Contents of the invention

Purpose of the invention: Aiming at the problem that the IQ imbalance of the transmitter is not deep enough in the existing research, the present invention proposes a method for estimating and compensating for the IQ imbalance in the transmitter of the OFDM-WLAN radio frequency test system.

Technical solution: a method for estimating and compensating for IQ imbalance suitable for an OFDM-WLAN radio frequency test system, comprising the following steps:

(1) Carry out serial-to-parallel conversion to the signal received by the vector signal analyzer, and perform FFT transformation to transform the signal into the frequency domain;

(2) Use the training sequence of the signal, use the smoothness of the channel to roughly estimate the IQ imbalance parameter, and eliminate the influence of the IQ imbalance in the channel estimation to obtain the initial value of the equalization sequence;

(3) Use the minimum mean square error (LMS) criterion to iteratively calculate the pilot information in the symbol to obtain a more accurate equalization sequence;

(4) Perform joint channel and IQ imbalance equalization on subcarriers and their image components.

In the described step (2), the channel estimation result is expressed as:

Among them, diag{λ} is the real channel impulse response, α=cos(Δφ)+jε _T sin(Δφ), β=ε _T cos(Δφ)+j sin(Δφ), and the amplitude and phase imbalance parameters ε _T is related to Δφ, and its estimated values are:

Among them, LTS2=LTS ^# /LTS, LTS is the frequency domain representation of the training sequence, the superscript # represents the FFT transformation result of the complex sequence conjugate, k is the subcarrier index value, For channel estimation, Indicates to take the real part of the signal, Indicates to take the imaginary part of the signal.

As preferably, in described step (2), eliminate the influence of IQ imbalance in channel estimation, the frequency domain response of channel is expressed as:

In described step (2), after obtaining the rough estimation result of IQ unbalanced parameter and channel, the initial value of setting equalization sequence is:

In the described step (3), the adaptive estimation iteration formula utilizing the LMS criterion for the pilot information in the symbol is:

in z(k) represents the kth subcarrier of the transmitted sequence, s(k) represents the kth subcarrier of the received sequence, w _k and w _N-k+2 are equalized sequences updated according to the LMS criterion.

Specifically, the iterative process of equalizing coefficients is:

where k={2,...,N/2}, and represent the equalization sequence and emission sequence at time i, respectively, for the training sequence The error signal of the k-th pilot subcarrier of the i-th iteration, for the training sequence The error signal of the N-k+2th pilot subcarrier of the i-th iteration, u _LMS is the step size used in the iterative process.

As an optimization, the normalized LMS algorithm is adopted, and the iteration step size is designed as:

Wherein μ _step is a normalized step size, the value range is 0<μ _step <2, and ||z(i)|| ² is the energy of the received signal.

The joint equalization method of channel and IQ imbalance in described step (4) is:

Jointly consider subcarriers and their image components, split the joint equalization of channel and IQ imbalance into multiple 2×2 decoupling equations, and define k={2,…,N/2} as follows:

Then the estimation of the frequency domain data can be obtained by adopting zero-forcing equalization method:

Among them, δ>0 is a normalization factor to resist the situation that Г _k is an ill-conditioned matrix, and I is an identity matrix.

Beneficial effects: the present invention discloses a method for estimating and compensating IQ imbalance suitable for OFDM-WLAN radio frequency test system. Firstly, the rough estimation of IQ imbalance parameter is obtained by using the long training sequence of WLAN signal according to the smooth characteristic of channel impulse response. Then according to the minimum mean square error criterion, the pilot information in the symbol is used for iterative calculation to obtain more accurate amplitude and phase error estimates, and the IQ imbalance and channel joint equalization are performed on the signal according to the estimated results. This method is suitable for the case where the channel has frequency selectivity, and compared with the traditional LMS algorithm, only a small amount of operations can be added to obtain the initial value close to the steady-state solution, which greatly speeds up the convergence speed of the algorithm. In less cases, good estimation and compensation effects can also be obtained, which has strong practicability.

Description of drawings

Fig. 1 is the transmitter IQ unbalanced system model schematic diagram that the present invention adopts;

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

Fig. 3 is the flow chart of the method for IQ imbalance estimation and equalization of the present invention;

Figure 4 is a comparison result diagram of the constellation diagram before (a) and after (b) the signal IQ imbalance compensation of the 16QAM modulation;

Fig. 5 is the comparison result diagram of the constellation diagram before (a) and after (b) the signal IQ imbalance compensation of the 64QAM modulation;

Fig. 6 is a graph showing the BER performance verification results of the method of the present invention.

Detailed ways

The invention discloses a method for estimating and equalizing IQ imbalance in a wireless local area network. In order to carry out necessary channel estimation and tracking, the wireless local area network standard IEEE 802.11a/g/n/ac provides a long training sequence, and inserts pilot information in the symbol, and the present invention provides a channel and IQ A method for joint estimation and equalization of imbalances.

Taking the IEEE 802.11ac signal as an example, the method proposed in the invention will be further described in detail in conjunction with the accompanying drawings.

Attached Figure 1 is the transmitter system when there is amplitude and phase imbalance. In this system, the IQ imbalance is mainly caused by the system clock, and its influence is constant in each frequency band of the transmitted signal, which is called frequency-dependent Unrelated IQ imbalance. Assuming that the amplitude and phase imbalance parameters caused by the clock at the transmitting end are ε _T and Δφ respectively, then the ideal time domain signal x _L (t) = x _I (t) + jx _Q (t) after the up-conversion operation, its transmission The radio frequency signal can be expressed as:

x _RF (t)＝(1＋ε _T )cos(ω ₀ t＋Δφ)x _I (t)-(1-ε _T )sin(ω ₀ t-Δφ)x _Q (t) (1)

An ideal receiver mixes the RF signal with x _LO (t)=exp(-jω ₀ t), then the signal after the high-frequency component is filtered by the low-pass filter can be expressed as:

in:

α＝cos(Δφ)+ _jεT sin(Δφ), β＝ _εT cos(Δφ)+jsin(Δφ) (2)

The equivalent block diagram of the time-domain model of transmitter IQ imbalance is shown in Figure 2, OFDM symbol s=[s(1) s(2)...s(N)] ^T is transformed into the time domain through IFFT operation, and added Cyclic prefix to symbol head formation After the parallel-to-serial conversion operation, the up-conversion operation is performed through the antenna. Due to the influence of the IQ imbalance of the transmitter, the actual transmitted signal can be expressed as: The length of the channel finite impulse response is less than the length of the cyclic prefix, so after the receiving end removes the cyclic prefix, the received sequence can be expressed as:

Among them, H ^c is a cyclic shift matrix of N×N size, is Gaussian white noise at the receiver. The cyclic shift characteristic of the matrix H ^c makes the result of its Fourier transform a diagonal matrix. The Fourier transform operation is performed on both sides of the above formula at the same time, and the superscript # is defined to represent the FFT transform result of the complex sequence conjugate. The relationship with the original sequence FFT transformation result can be expressed as:

The relationship between the frequency-domain received sequence z and the original sequence s can be obtained as follows:

z=diag(λ)(αs+βs ^# )+v (5)

Among them, diag{λ} is the real channel impulse response, and v is Gaussian white noise.

As shown in Figure 3, a method for estimating and compensating an IQ imbalance applicable to an OFDM-WLAN radio frequency test system disclosed by an embodiment of the present invention mainly includes the following steps:

S1: Perform serial-to-parallel conversion on the signal received by the vector signal analyzer, and perform FFT transformation to transform the signal into the frequency domain;

S2: Rough estimation of IQ imbalance: use the training sequence of the signal, use the smooth characteristics of the channel to roughly estimate the IQ imbalance parameter, and eliminate the influence of IQ imbalance in channel estimation to obtain the initial value of the equalization sequence;

S3: Fine estimation of IQ imbalance: use the LMS criterion to iteratively calculate the pilot information in the symbol to obtain a more accurate equalization sequence, and normalize the iteration step size to ensure the convergence of the algorithm;

S4: joint equalization: performing joint channel and IQ imbalance equalization on subcarriers and their image components.

Using the structure in Figure 3 to jointly estimate and compensate the channel and IQ imbalance, and pair the subcarrier with its mirror image, the following relationship can be defined for the subcarrier index value k={2,...,N/2}:

z _k ＝Γ _k s _k +v _k (6)

Among them, v _k is the Gaussian white noise introduced by the channel, and the definitions of other parameters are as follows:

The iterative formula for defining adaptive estimation using the LMS algorithm is:

Among them, w _k and w _N-k+2 are equalization sequences updated according to the LMS criterion, and their initialization sequences are obtained by the following operations:

IEEE 802.11ac adds a VHT-LTF field to the frame header for necessary frequency offset and channel estimation. The VHT-LTF of its 20MHz signal is defined as:

VHTLTF _28,28 ={1,1,LTF _left ,0,LTF _right ,-1,-1} (10)

in,

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}

Neglecting the influence of noise, the long training sequence field of the received signal is used for LS channel estimation, and the estimation result can be expressed as:

Wherein, LTS2=LTS ^# /LTS, where LTS is the frequency domain representation of the training sequence. For the subcarrier index value k (k is the leading edge of the LTS2 transition), there are:

Since the coherent bandwidth of the channel is much larger than the interval between subcarriers, the frequency domain responses on adjacent subcarriers can be approximately equal. In addition, according to LTS2 _Nαk+2 = LTS2 _N-(k+1)+2 , the difference between the two formulas can be have to:

In the actual OFDM system, the magnitude and phase mismatch values are small, so the estimation of the IQ imbalance parameter β is:

According to the triangle inequality relationship, the parameter α can be estimated as:

Compensating the channel estimation results with IQ imbalance can obtain:

Therefore, after obtaining the rough estimation results of the IQ imbalance parameters and the channel, the initial value of the equalization sequence can be set as:

Then, use the pilot signal in IEEE 802.11ac to iterate, the pilot subcarrier index value of the IEEE 802.11ac 20MHz signal is K _Pilot ={±7,±21}, and the values on each pilot are:

p _n {-21,-7,7,21}={Ψ _nmod4 ,Ψ _(n+1)mod4 ,ψ _(n+2)mod4 ,ψ _(n+3)mod4 } (18)

The linear shift characteristic of the pilot information ensures the validity of the equalization sequence solution, and the time subscript i is introduced in the sequence, so that and represent the equalization sequence and transmission sequence at time i respectively, then the iterative process for k={2,...,N/2} equalization coefficients is:

in for the training sequence The error signal of the k-th pilot subcarrier of the i-th iteration, for the training sequence The error signal of the N-k+2th pilot subcarrier of the i-th iteration, u _LMS is the step size used in the iterative process, using the normalized LMS algorithm, it is defined as:

Wherein μ _step is a normalized step size, the value range is 0<μ _step <2, and ||z(i)|| ² is the energy of the received signal. Through the above iterative estimation, an accurate estimation of the IQ imbalance parameter can be obtained, and then the impulse response of the channel can be obtained to perform frequency domain compensation of the signal. Specifically, in order to jointly consider subcarriers and their image components, the joint equalization of channel and IQ imbalance is split into multiple 2×2 decoupling equations, and the frequency domain data is obtained by using zero-forcing equalization according to the following formula In order to complete the joint equalization operation of the channel and IQ imbalance of the signal in the frequency domain.

Use the amplitude and phase imbalance factors estimated by this method to compensate the system with IQ imbalance and verify its system performance. The MCS4 data using 16QAM modulation and the MCS7 data using 64QAM modulation in the IEEE 802.11ac protocol are selected for testing. Attached Figure 3 and Figure 4 are the comparison of constellation diagrams before and after compensation of two sets of sample data. It can be clearly seen that due to the influence of IQ imbalance, the points on the constellation diagram are discretely distributed around the standard constellation diagram points, and the high-order modulation mode is more affected by the IQ imbalance, and the data on different constellation points are intertwined so that cause misjudgment. The compensated data suppresses the image component, increases the accuracy of channel estimation and equalization, and effectively improves the distortion of the constellation diagram, so that the points of the constellation diagram are all gathered near the standard point. It can be seen that the IQ imbalance frequency domain proposed in the paper The compensation method has a very good compensation effect.

Figure 6 shows the performance of the method proposed by the present invention on the bit error rate performance, the compensated system effectively improves the bit error rate performance of the system, eliminates the floor effect caused by IQ imbalance, and performs iterative training using only 5 symbols good performance can be obtained.

Claims (5)

1. An IQ imbalance estimation and compensation method suitable for an OFDM-WLAN radio frequency test system is characterized in that: the method comprises the following steps:

(1) performing serial-parallel conversion on the signals received by the vector signal analyzer, and performing FFT (fast Fourier transform) conversion to convert the signals to a frequency domain;

(2) using a training sequence of a signal, carrying out rough estimation on IQ imbalance parameters by using the smooth characteristic of a channel, and eliminating the influence of IQ imbalance in channel estimation so as to obtain an initial value of an equalization sequence;

(3) carrying out iterative operation on pilot frequency information in the symbol by using a least mean square error (LMS) criterion to obtain an updated equalization sequence; the adaptive estimation iterative formula using the LMS criterion for the pilot information in the symbol is:

whereinz (k) denotes the k-th subcarrier of the transmitted sequence, s (k) denotes the k-th subcarrier of the received sequence, w_kAnd w_N-k+2Is an equalization sequence updated according to the LMS criterion; the iterative process of the equalization coefficients is:

where k is {2, …, N/2},andrespectively representing the equalization sequence and the transmit sequence at time i,to adopt a training sequenceThe error signal at the i-th iteration is performed on the k-th pilot subcarrier,to adopt a training sequenceOf the (N-k + 2) th pilot sub-carrier, u_LMSThe step size used for the iterative process;

(4) performing combined channel and IQ imbalance equalization on the subcarriers and the mirror image components thereof; the specific combined equalization method comprises the following steps:

jointly considering the subcarriers and the mirror image components thereof, splitting the joint equalization of the channel and IQ imbalance into the solution of a plurality of 2 × 2 decoupling equations, and defining k as {2, …, N/2 }:

then the frequency domain data can be estimated by using zero-forcing equalization as follows:

where δ > 0 is a normalization factor to combat Γ_kIn the case of a pathologic matrix, I is a unit matrix.

2. The IQ imbalance estimation and compensation method according to claim 1, wherein: the result of channel estimation in step (2) is represented as:

where diag { λ } is the true channel impulse response, α ═ cos (Δ φ) + j ε_Tsin(Δφ)，β＝ε_Tcos (Δ φ) + jsin (Δ φ), and the amplitude and phase imbalance parameters ε_TAnd Δ φ, the estimated values are:

wherein LTS2 ═ LTS^#LTS, LTS is the frequency domain representation of the training sequence, superscript # represents the FFT result of the conjugate of the complex sequence, k is the subcarrier index value,in order to estimate the channel, it is,the representation is taken of the real part of the signal,the representation takes the imaginary part of the signal.

3. The IQ imbalance estimation and compensation method according to claim 2, wherein: in the step (2), the influence of IQ imbalance in channel estimation is eliminated, and the frequency domain response of the channel is represented as:

4. the IQ imbalance estimation and compensation method according to claim 3, wherein: after acquiring the coarse estimation result of the IQ imbalance parameters and the channel in the step (2), setting the initial value of the equalization sequence as follows:

5. the IQ imbalance estimation and compensation method according to claim 1, wherein: the iteration step length is designed as follows by adopting a normalized LMS algorithm:

wherein mu_stepThe value range is more than 0 mu for normalizing the step length_step＜2，||z(i)||²Is the energy of the received signal.