KR20030088813A - Estimation and compensaion method of I/Q mismatch and DC offset - Google Patents

Abstract

PURPOSE: A device for estimating and compensating I/Q mismatch is provided to estimate channels, I/Q mismatch, and DC offset in a time region by using LSs(Least Squares), and to compensate the I/Q mismatch and the DC offset by using the estimated value, thereby minimizing training data groups. CONSTITUTION: An estimator(10) estimates I/Q mismatch, DC offset, and a channel coefficient by using LSs from a receiving signal corresponding to training data. A compensator(70) sequentially removes the DC offset and the I/Q mismatch from the receiving signal by using an estimated value, and consists of adders(20,50), a multiplier(40), and a complex conjugate generator(30). The adder(20) removes the DC offset by using the DC offset estimated value. The complex conjugate generator(30) generates a complex conjugate signal of the receiving signal whose DC offset is removed. The multiplier(40) multiplies an I/Q mismatch estimated value by the generated signal. The adder(50) removes the I/Q mismatch.

Description

Estimation and Compensation Method of I / M Mismatch and Apparatus, Apparatus and Estimation and Compensation Method of I / M Mismatch and DC Offset and Apparatus {Estimation and compensaion method of I / Q mismatch and DC offset}

The present invention relates to a method and apparatus for estimating and compensating I / Q mismatch and DC offset, and more particularly, to a channel, I / Q mismatch, and DC offset using data aid in a digital signal receiver. The present invention relates to a method and apparatus for estimating I / Q mismatch and DC offset using an estimated value in the time domain.

In order to detect a received signal in a general digital communication system, the passband signal is down-converted and converted into a complex baseband signal.

The complex baseband signal is composed of an in-phase (I) channel signal and a quadrature-phase (Q) channel signal.

In-phase (I) channel signals and quadrature (Q) channel signals pass through amplifiers, lowpass filters, and analog-to-digital converters (ADCs).

In this process, the transmission characteristics of the I channel signal and the Q channel signal are not identical to each other.

Also, during this process, unwanted DC signal components can be added to the I channel signal and the Q channel signal. This is called DC offset.

I / Q mismatch and DC offset distort the received baseband signal and degrade the receiver's performance.

There are several methods of prior art regarding the estimation and compensation of I / Q mismatch and DC offset.

Among them, FE Churchill, GW Ogar, and BJ Thompson, "The correction of I and Q errors in a coherent processor", IEEE Trans.Aerosp.Electro.Syst., Vol.AES-17, no. 1, pp 131-137, Jan. 1981. and RA Green, R. Anderson-Sprecher, and JW Pierre, "Quadrature receiver mismatch calibration", IEEE Trans. Sig. Proc., Vol. 47, no. 11, pp. 3130-133, Nov. 1999.) discloses a method of applying a sinusoidal test signal to the mixer input of a receiver to convert it to baseband and then estimating I / Q mismatch and DC offset from the baseband test signal. The disadvantage is that it can only be used in situations where a sinusoidal signal can be applied to the receiver.

L. Yu and WM Snelgrove, "A novel adaptive mismatch cancellation system for quadrature IF radio receivers", IEEE Trans. Circuits and Sys. II, vol. 46, no. 6, pp. 789-801, Jun. 1999 . And M. Valkama, M. Renfors, and V. Koivunen, "Advanced methods for I / Q imbalance compensation in communication receivers", IEEE Trans. Sig. Proc., Vol. 49, no. 10, pp. 2335-2344, Oct. 2001., discloses a method for eliminating I / Q mismatch using a blind adaptive filter, but has a disadvantage of slow convergence speed of the blind adaptive filter.

JK Cavers and MW Liao, "Adaptive compensation for imbalance and offset losses in direct conversion tranceivers", IEEE Trans. Veh. Technol., Vol. 42, no. 4, pp. 581-588, Nov. 1993. I / Q mismatch and DC offset cancellation using data-assisted adaptive filter is described, but it is proposed assuming a channel environment that does not cause inter-symbol interference (ISI) between transmitted data. It can be used only in additive white Gaussian noise or flat fading channels, not in frequency selective channels, and requires a large number of training data to adapt the filter.

In addition, Korean Patent Application No. 10-2000-0004053 (name of the invention: a channel estimating apparatus considering an I / Q mismatch error and a digital signal receiver having the same) is an I / Q mismatch estimating method using data assistance. A method of compensating I / Q mismatch by using an equalizer by estimating a signal channel and an image signal channel after performing a Fast Fourier Transform (FFT) operation in the frequency domain It is.

However, the prior patent application method was developed assuming an orthogonal frequqncy division multiplexing (OFDM) system, and has a disadvantage in that it cannot be applied to a single carrier system.

SUMMARY OF THE INVENTION The present invention has been made to solve the above-described problems, and an object of the present invention is to be applicable to both a single carrier system and a multi-carrier system in a general channel environment including not only an AWGN channel and a flat fading channel but also a frequency selective channel. The present invention provides a method for estimating and compensating I / Q mismatch and DC offset using data assistance, and an apparatus for estimating and compensating I / Q mismatch.

In order to achieve the above object, the present invention comprises the steps of estimating the I / Q mismatch, DC offset and channel from the received signal corresponding to the training data using the least square method;

Sequentially removing a DC offset, an I / Q mismatch, and a channel from a received signal using the estimate obtained in the estimating step;

To provide an estimation and compensation method for including I / Q mismatch and DC offset.

In order to achieve the above object, the present invention includes an estimator for estimating the I / Q mismatch from the received signal corresponding to the training data using the least square method;

A complex conjugate signal generator for generating a complex conjugate signal of the received signal, a multiplier for multiplying the complex conjugate signal with an I / Q mismatch estimate obtained by an estimator, and an I / Q mismatch with a signal that has passed the multiplier in the received signal; A compensator composed of an adder for removing;

It is intended to provide an I / Q mismatch estimating and compensating device that includes.

1 is a block diagram showing a receiving apparatus of a general digital communication system.

2 is a block diagram of a device for compensating I / Q mismatch and DC offset to which an estimation method according to the present invention is applied.

3 is a diagram illustrating an average square error of channel estimates according to the present invention.

4 is a diagram showing an average square error of an I / Q imbalance coefficient estimate according to the present invention.

5 is a diagram illustrating an average square error of a DC offset estimate according to the present invention.

6 is a view comparing the image removal rate when the I / Q mismatch and the DC offset is compensated by the conventional method and the compensation according to the present invention.

<Description of the symbols for the main parts of the drawings>

10: estimator 20, 50: adder

30: plural pair generator 40: multiplier

60: equalizer 70: compensator

Hereinafter, the configuration and operation of the present invention will be described in detail with reference to the accompanying drawings.

1 is a block diagram of a receiving apparatus of a digital communication system.

In Figure 1 Is the signal transmitted from the transmitter, Is the angular frequency of the carrier, The equivalent lowpass impulse response of the channel, Is the additive noise in the passband, Is the DC offset.

At this time, the passband signal entering the mixer 2 Is expressed as in Equation 1.

From here Is the equivalent low-pass signal of the received signal, Wow It is expressed as convolution of.

Output signal of local oscillator (6) Is the same as Equation 2.

From here and Are mismatches in magnitude and phase, respectively, , to be.

Output signal of mixer (2) Is Is indicated by Is passed through the low pass filter (4). Is expressed as in Equation 3.

From here Is the band-limited equivalent lowpass noise.

remind Sampling at a symbol period yields a signal as shown in Equation (4).

From here , Is Impulse response of equivalent channel by impulse applied before time unit, Is the length of the channel, Is an information symbol.

When the dog training data is transmitted, the received data corresponding to the training data is written in vector form as shown in Equation 5.

From here , , Is the length of all elements equal to 1 ( ) Is a column vector.

A matrix of dimensions Is the same as Equation 6.

To simplify the equation , If Can be written as in Equation 7.

From here For convenience of discussion Mismatch coefficient, Let be the channel coefficient.

Channel coefficients to detect information symbols transmitted from signals received in the above signal model , Mismatch coefficient , DC offset factor Need information about

Reference numerals 1 and 3 of FIG. 1 indicate an adder 5 and an A / D converter.

The method of estimating these coefficients will now be described.

Since the above equation is not a linear model, first, the same vector as Equation 8 is defined to linearize Equation 7.

Then, Equation 7 is converted into the linear model of Equation 9 as follows.

From here ego to be.

By using the model of Equation 9 Deriving the least squares (LS) estimation method for Eq.

In Equation 10 Is , , And Since this is an estimate for, the right side of If you compare Wow You can get an estimate for.

In other words, When I say Wow Is the same as Equation (11) and (12).

First Just estimate Estimate of Can be obtained in several ways.

For convenience Let's say.

Under Three ways to obtain

first, of From the standpoint of fitting It is expressed by Equation 13 as a way to obtain.

second, Estimate of of Estimate of It is represented by Equation 14 by dividing the corresponding elements by the conjugate of and taking the average one by one.

third, Wow From equation (7) To obtain the equation, in equation (7) Wow From the LS perspective Is obtained as shown in Equation 15.

Since the method described above is derived by generalizing the channel, it can be applied to all AWGN channels, flat fading channels, and frequency selective channels.

Also Necessary for estimation of Since the matrix is the transmission signal data seen in the time domain, this matrix can be obtained by obtaining the value of the training data in the time domain regardless of the modulation scheme or the number of carriers.

The following describes how to remove I / Q mismatch and DC offset.

2 is a block diagram of an I / Q mismatch and a DC offset compensation device to which the estimation method proposed in the present invention is applied.

The apparatus consists of an estimator 10 and a compensator 70.

In the estimator 10, the received signal corresponding to the training data ( ) Using I / Q mismatch (Equation 11 to 15) using least-squares ), DC offset ( ) And channel coefficients ( Estimate).

The corrector 70 sequentially removes the DC offset and the I / Q mismatch from the received signal using the estimate obtained from the estimator 10, and adds the adders 20 and 50 to the multiplier 40 and the complex conjugate generator 30. It is composed.

The adder 20 receives the received signal ( DC offset estimate obtained by the estimator 10 from To remove the DC offset.

The complex conjugate generator 30 generates a complex conjugate signal of the received signal from which the DC offset is removed.

The multiplier 40 generates an I / Q mismatch estimate for the signal generated by the complex conjugate generator 30. Multiply by

The adder 50 eliminates I / Q mismatch by the following equation (16).

In the above configuration, when there is no DC offset and only IQ mismatch, Equation 8 To And replace with of The channel can be obtained as shown in Equation 11 and the I / Q mismatch can be obtained as shown in Equations 13 to 15.

In this case, the complex conjugate generator 30 generates a complex conjugate signal of the received signal immediately without configuring the front adder 20, and the adder 50 of the rear stage removes the I / Q mismatch from the received signal.

Equalizer 60 is a channel coefficient estimate obtained by using the present invention ( ) To compensate for the channel.

In this case, the equalizer 60 compensates for gain and phase in a flat fading channel and compensates for inter-symbol interference (ISI) in a frequency selective channel. do.

For convenience of explanation Wow We will explain how to eliminate I / Q mismatch and DC offset when we know.

Using this device, it is possible to remove the I / Q mismatch and the DC offset by calculating the equation using Equation 4 as follows.

In Equation 16 It can be seen that I / Q mismatch and DC offset are removed signals.

In relationship with Because of If you know Wow Is available.

Accordingly, the right side of Equation 16 is a form in which a known constant is multiplied by a desired signal, and it can be seen that I / Q mismatch and DC offset can be removed by the above method.

I / Q mismatch coefficient And DC offset Is estimated using the method described above.

Next, the channel data, the I / Q mismatch coefficient, and the training data for optimizing the estimation performance of the DC offset will be described.

Assuming that the training data is a phase shift keying (PSK) signal, the training data satisfying the following three equations Minimize the mean square error (MSE) of.

From here Is a constant.

In order to prove the above equations (17) to (19), a covariance is used. If MSE is given by Equation 20.

From here Is a positive semi-definite matrix Eigenvalue of Are all It is a mistake.

The sum of eigenvalues is the trace of this square matrix. Same as

By the way Is composed of phase modulation (PSK) symbols from column 1 to column 2L, and columns 2L + 1 are all 1s. All diagonal elements in are constant constants.

Therefore, the relationship of equation (21) holds.

From here Is a constant.

Eigenvalue of Because of The trace of is as shown in Equation 22.

therefore MSE To find a condition that minimizes Have constraints It is equivalent to finding the condition that minimizes.

To solve this problem, the method of Lagrangemultipliers is introduced to define a new cost function such as Equation 23.

From here Is a Lagrangian multiplier.

To minimize To find the equation (23) Differentiating with respect to Eq.

To satisfy Is as follows.

Because of Must satisfy the condition of Equation 26 below.

By the way Is Have independent eigenvectors Can be eigenvalue decomposition as

From here Each column Is a matrix of orthogonal eigenvectors, to be.

Since is a Hermitian matrix, an orthogonal eigenvector set can always be obtained.

From Equation 21 and Equation 26, the relation of Equation 28 can be obtained.

From here to be.

In Equation 28 Substituting can obtain the properties of equations (17) to (19).

Equation 17-19 is In addition to minimizing the MSE of the estimation, there is an advantage that the channel estimation can be performed with little influence of I / Q mismatch and DC offset even when the channel is estimated by the conventional channel estimation method.

That is, the signal model when there is no I / Q mismatch and DC offset in the conventional channel estimation method (Equation 7 ego In the case of), the following equation (22) is obtained.

In addition, substituting the signal model (7) with the I / Q mismatch and the DC offset produces the same result as (30).

Equation 20 is noise in Equation 29 end The same is true except that replaced by.

Assuming that to be.

That is, even in the existing channel estimation method, if the training data satisfying the conditions described in the present invention are used, channel estimation can be performed almost unaffected by I / Q mismatch and DC offset.

As an example of training data that satisfies the condition of Equation 17-19, An example of two training data is shown in Table 1.

Item Preamble 1 Preamble 2 Training data {-1, -1,1,1,1,1,1, -1, j, -j, j, -1, -j, -1, -1,1, -1, -1} {-j, 1,1, j, -j, 1, -1, j, -j, -1, -1, j, j, 1, -1, -j, -j, 1} Property , , , ,

Table 1 shows the analysis of the properties of these two training data.

In Table 1, preamble 1 is training data that satisfies the conditions of Equations 17 and 19, but not Equation 18, and preamble 2 is training data that satisfies all of the conditions of Equations 17-19.

That is, preamble 1 is optimal training data only in the conventional channel estimation method that does not consider I / Q mismatch, while preamble 2 is optimal training data for estimation in consideration of I / Q mismatch and does not consider I / Q mismatch. It is also the best training data for channel estimation in the case of no use.

Using the two training data from Table 1 , , We compare the estimated performance of the results with computer simulations.

The simulation environment used for the following computer simulations is based on channel length L = 3 and size mismatch. Phase mismatch , DC offset to be.

in this case to be.

3, 4 and 5 are respectively , , The MSE is shown.

Wow In estimating, it can be seen that the optimal training data preamble 2 shows less MSE than the preamble 1 as shown in FIGS. 3 and 5.

Therefore, the training data with the conditions presented in the present invention shows better performance than the training data with only the optimal conditions in the existing channel estimation without considering the I / Q mismatch.

As in Figure 5, In the estimation of, preamble 1 and preamble 2 show the same MSE.

This is because both preamble 1 and preamble 2 have the condition of Equation 19 necessary for DC offset estimation.

In Figure 3 ego In this case, that is, the channel estimation by the conventional channel estimation method for the case where there is no I / Q mismatch and DC offset is also shown.

In the absence of I / Q mismatch and DC offset, preamble 1 and preamble 2 show the same performance.

In the case of using the estimation method and the optimal training data preamble 2 proposed in the present invention, the channel estimation performance is almost the same as in the case where there is no I / Q mismatch and no DC offset.

FIG. 6 compares the image rejection ratio (IRR) in the case of compensating I / Q mismatch and DC offset with the method of Document 5 described in the related art and the case according to the present invention in an AWGN environment.

and When I denote signal power and image power, IRR is defined as follows.

The method proposed in the present invention uses 18 training data and compares 18, 50 and 100 training data of the prior art document 5.

The method presented in the present invention shows better performance than the case of using 18 and 50 training data by the method of the conventional document 5, and shows similar results to those using 100 training data by the method of the conventional document 5. Can be.

That is, the method proposed in the present invention in order to obtain the same IRR performance requires a smaller number of training data than the method of the prior art 5.

6 is the same as the one used in the simulations of FIGS. The IRR is shown when using the method of the present invention in the in-frequency selective channel.

The method presented in the present invention shows that the frequency selective channel can operate as well as the AWGN channel.

As described above, in the present invention, the channel, I / Q mismatch, and DC offset are estimated using the least-squares method in the time domain using the help of data in a digital receiver, and the I / Q mismatch and DC are estimated using the estimated values. By compensating the offset, the training data group is minimized and the training data of short length is different from the existing method because it is based on the estimation technique unlike the conventional adaptive filter based technique.

In addition, the present invention can be applied to both a single carrier system and a multi-carrier system in a general channel environment including not only AWGN channel, flat fading channel but also frequency selective channel.

Claims (26)

Estimating I / Q mismatch from the received signal corresponding to the training data using a least square method; Removing the I / Q mismatch from the received signal using the I / Q mismatch estimate obtained in the estimating step; Including and estimating I / Q mismatch. The method according to claim 1, wherein in the estimating step, Channel coefficient , Channel coefficient estimate , Mismatch coefficient , Vector , Estimate of When we say If, the mismatch coefficient ( Estimate of To, of A method for estimating and compensating for I / Q mismatch, characterized in that it is obtained from the viewpoint of fitting to the N-Q. The method of claim 2, wherein the mismatch coefficient ( Estimate of Equation 32 is the estimation and compensation method of the I / Q mismatch, characterized in that. The method of claim 2 or 3, wherein the estimate of the channel coefficients Equation 33 is the estimation and compensation method of the I / Q mismatch, characterized in that. The method according to claim 1, wherein in the estimating step, Channel coefficient , Mismatch coefficient , Vector , Estimate of When we say If, the mismatch coefficient ( Estimate of To, Estimate of of Estimate of A method for estimating and compensating I / Q mismatch, characterized by dividing the corresponding elements by conjugate one by one and taking the average. The method of claim 5, wherein the mismatch coefficient ( Estimate of Equation 34 is the estimation and compensation method of the I / Q mismatch, characterized in that. The method according to claim 1, wherein in the estimating step, Channel coefficient , Mismatch coefficient , Vector , Estimate of When we say If, the mismatch coefficient ( Estimate of To, Channel coefficient estimate in Equation 35 below Method of estimating and compensating I / Q mismatch, characterized in that obtained from the least square method (LS) perspective by substituting. here , , Is the length of all elements equal to 1 ( Column vector, Is Matrix of dimensions, , to be. The method of claim 7, wherein the mismatch coefficient ( Estimate of Equation 36 is the estimation and compensation method of I / Q mismatch, characterized in that. 9. The channel coefficient estimate of claim 7 or 8 Equation 37 is the estimation and compensation method of the I / Q mismatch, characterized in that. The method of claim 1, wherein the training data satisfies the condition of Equation 38. Estimating the I / Q mismatch and the DC offset from the received signal corresponding to the training data using the least square method; Sequentially removing the DC offset and the I / Q mismatch from the received signal using the estimate obtained in the estimating step; A method for estimating and compensating for including I / Q mismatch and DC offset. The method according to claim 11, wherein in the estimating step, Channel coefficient , Channel coefficient estimate , Mismatch coefficient , DC offset , Estimate of , Vector , Estimate of When we say If, the mismatch coefficient ( Estimate of To, of A method for estimating and compensating for I / Q mismatch and DC offset, characterized in that it is obtained from the viewpoint of fitting to the P. The method according to claim 12, wherein the mismatch coefficient ( Estimate of Is an equation for estimating and compensating I / Q mismatch and DC offset. The method of claim 12 or 13, wherein the estimate of the channel coefficients Equation 40 is an estimate of the DC offset. Is an equation for estimating and compensating I / Q mismatch and DC offset. The method according to claim 11, wherein in the estimating step, Channel coefficient , Mismatch coefficient , DC offset , Estimate of , Vector , Estimate of When we say If, the mismatch coefficient ( Estimate of To, Estimate of of Estimate of A method for estimating and compensating I / Q mismatches and DC offsets, characterized by dividing the corresponding elements by conjugate one by one and taking the average. The method of claim 15, wherein the mismatch coefficient ( Estimate of Equation 42 is the estimation and compensation method of I / Q mismatch and DC offset. The method according to claim 11, wherein in the estimating step, Channel coefficient , Channel coefficient estimate , Mismatch coefficient , DC offset , Estimate of , Vector , Estimate of When we say If, the mismatch coefficient ( Estimate of To, In equation 43 below Wow A method for estimating and compensating I / Q mismatch and DC offset, characterized in that it is obtained from the least square method (LS) perspective. here , , Is the length of all elements equal to 1 ( Column vector, Is Matrix of dimensions, , to be. The method of claim 17, wherein the mismatch coefficient ( Estimate of Equation 44 is the estimation and compensation method of I / Q mismatch and DC offset. 19. The method of claim 17 or 18, wherein an estimate of the channel coefficients Equation 45 is an estimate of the DC offset. Equation 46 is the estimation and compensation method of I / Q mismatch and DC offset. 12. The method of claim 11, wherein the training data satisfies the condition of Equation 47. An estimator for estimating I / Q mismatch from a received signal corresponding to training data using a least square method; A complex conjugate signal generator for generating a complex conjugate signal of the received signal, a multiplier for multiplying the complex conjugate signal with an I / Q mismatch estimate obtained by an estimator, and an I / Q mismatch with a signal that has passed the multiplier in the received signal; A compensator composed of an adder for removing; I / Q mismatch estimating and compensation device comprising. 22. The method of claim 21, wherein the estimator further includes a function of estimating a channel using a least square method from a received signal, and an equalizer further configured at the rear of the compensator to compensate a channel of the received signal passing through the compensator. / Q Mismatch estimator and compensation device. 23. The method of claim 22, wherein the equalizer compensates for gain and phase when the received signal is a flat fading channel, and compensates for intersymbol interference in the equalizer when the channel is a frequency selective channel. Device. An estimator for estimating I / Q mismatch and DC offset using a least square method from the received signal corresponding to the training data; An adder for removing a DC offset from the received signal to generate a signal from which the DC offset has been removed, a complex conjugate signal generator for generating a complex conjugate signal of the signal from which the DC offset has been removed, and a complex conjugate to remove the DC offset; A compensator comprising a multiplier for multiplying the signal by the I / Q mismatch estimate obtained by the estimator and an adder for removing the I / Q mismatch from the signal from which the DC offset has been removed by the multiplier; Apparatus for estimating and compensating I / Q mismatch and DC offset for inclusion. 25. The method of claim 24, wherein the estimator further includes a function of estimating a channel using a least square method from a received signal, and an equalizer further configured at the rear of the compensator to compensate a channel of the received signal passing through the compensator. / Q Mismatch and DC offset estimation and compensation device. 27. The method of claim 25, wherein the equalizer compensates for gain and phase when the received signal is a flat fading channel and compensates for intersymbol interference in the equalizer when the frequency selective channel is used. Estimation and Compensation Device.