TWI673960B - Tuning method for improving the quality of broadband RF signals - Google Patents

Abstract

The invention is a calibration method for improving the quality of a broadband radio frequency signal. First, a joint signal model with radio frequency imperfection is established. The ideal signal is input to the joint signal model to generate a received signal, and the front-end filter parameters of the joint signal model are estimated. Receive signals for compensation. Then, the DC offset of the received signal is estimated to estimate the parameters of the DC offset. Finally, the parameters of the rear-end filter of the joint signal model are estimated, and according to the parameters of the back-end filter and the parameters of the DC offset, Receive signals for compensation. The invention can be used for correcting the signals of the transmitting end and the receiving end. When the transmitting end and the receiving end are connected in series, the signal of the receiving end is adjusted first and then the transmitting end is adjusted to correct the signal offset and DC offset generated by the filter. To compensate for imperfect signals.

Description

Calibration method for improving broadband RF signal quality

The invention relates to a signal compensation technology, in particular to a calibration method for improving the quality of a broadband radio frequency signal.

Radio frequency (RF), also known as radio frequency, radio frequency, high frequency, is often used as a synonym for radio. As a wireless communication system, data has been transmitted.

The transmitter sends a radio frequency signal to the middle of the receiver. The radio frequency signal will have additional RF impairments, such as IQ imbalance (In-phase / Quadrature-phase imbalance), shaping filter imbalance (shaping filter imbalance), and DC offset and other issues. In order to limit the bandwidth, the transceiver must use a pulse-shaping filter to reduce the signal bandwidth to meet the system bandwidth limit and reduce inter-symbol interference (ISI). Currently, Nyquist filters are used. (Nyquist filter) and square-root-raise-cosine (SRRC) to shape the transmission signal at the transmitting end or the receiving end, but when the transmitting end and the receiving end use different shaping filters, it will cause the transmitting end and the receiving end There is a problem of imbalance of the shaping filter between the terminals. Furthermore, in order to reduce costs, many users will use a cheaper direct conversion architecture. During the conversion process, some of the local oscillator power will leak to the RF signal and mix with the transmission signal, resulting in IQ DC at the transmission end. The effect of offset defects.

In addition, due to the influence of factors such as refraction, diffraction, or scattering in the indoor or outdoor environment, causing the receiving end to receive signals of two or more different paths at different delay times, it will also cause inter-symbol interference and cause Reduced performance. In addition, when the RF module between the transmitting end and the receiving end performs up / down frequency conversion, the incomplete synchronization of the oscillator will cause a frequency offset, and the Doppler frequency shift due to high-speed movement will also cause a carrier frequency offset. In systems using single-carrier frequency division multiplexing access and orthogonal frequency division multiplexing access technology, the carrier frequency offset has a great impact, which will not only interfere with the transmission of wireless communication systems, but also cause inter-carrier interference (Inter -carrier Interference (ICI).

In view of this, the present invention proposes a calibration method for improving the quality of broadband radio frequency signals in order to effectively overcome the above-mentioned problems in view of the lack of the conventional techniques.

The main purpose of the present invention is to provide a calibration method for improving the quality of wideband radio frequency signals, which can correct imperfect signals to solve problems such as signal offset caused by a signal passing through a filter, and DC offset to compensate Imperfect signal.

Another object of the present invention is to provide a calibration method for improving the quality of a broadband radio frequency signal, which can simultaneously compensate for imperfect signals at the receiving end and the transmitting end to compensate for the imperfect signals.

In order to achieve the above-mentioned object, the present invention provides a calibration method for improving the quality of a broadband radio frequency signal. The steps include: establishing a joint signal model with imperfect radio frequency; inputting an ideal signal into the joint signal model to generate a receiving signal, The front-end filter of the joint signal model is estimated and the parameters of the front-end filter are estimated, and then the received signal is compensated according to the parameters of the front-end filter; the DC offset of the received signal is estimated to estimate the DC offset parameter ; And estimate the rear-end filter of the joint signal model to estimate the parameters of the back-end filter, and compensate the received signal based on the parameters of the back-end filter and the DC offset parameter.

The step of estimating the parameters of the front-end filter further includes the following steps: dividing the received signal into a phase channel (I channel) signal and an amplitude channel (Q channel) signal; defining the relationship between the front-end filter and the phase channel signal and the amplitude channel signal , Expressed in equations as follows: among them Is the phase channel (I channel) signal, Is the amplitude channel (Q) signal, , , , Is the parameter of the front-end filter, Is the phase channel (I channel) signal, Is the amplitude channel (Q) signal, , Is the DC offset parameter of the front end; then the phase channel signal and the amplitude channel signal are digitally sampled by analog rotation to form a matrix with the training code in the ideal signal; finally, the phase channel signal and the Parameters of each parallel filter corresponding to the complex front-end filter in the amplitude channel signal.

The step of estimating the parameters of the back-end filter further includes the following steps, so that the phase channel signal is equal to a pulse function and the amplitude channel signal is equal to zero, which is expressed by the equation as follows: among them , Is the parameter of the back-end filter sampled by analog to digital, , , , Is the parameter of the front-end filter sampled by analog to digital conversion, Is a pulse function; the back-end filter is estimated using a convolution matrix algorithm and a least squares method , Parameters; let the amplitude channel signal be equal to the pulse function and the phase channel signal be equal to zero, which is expressed by the equation as follows: among them , Is the parameter of the back-end filter sampled by analog to digital, , , , Is the parameter of the front-end filter sampled by analog to digital conversion, Is a pulse function; the back-end filter is estimated using the convolution matrix algorithm and the least square method , Parameters.

After the step of establishing a joint signal model with radio frequency imperfection, it further includes a joint signal model in which a transmitting end transmits an ideal signal to a receiving end, and a reception signal generated by the joint signal model generates a frequency offset, and After the analog signal of the received signal is digitally sampled, it enters the step of estimating the front-end filter of the joint signal model to estimate the parameters of the front-end filter.

Detailed descriptions will be provided below through specific embodiments to make it easier to understand the purpose, technical content, features and effects of the present invention.

First, please refer to the first figure, which is a system architecture diagram of this embodiment. The method of this embodiment can be used to compensate for imperfect reception signals at the receiving end or the transmitting end. The compensation methods at the transmitting end and the receiving end are the same. Therefore, this embodiment only uses the structure of the receiving end as an example.

As shown in the first figure, it is a system architecture applied to the method for correcting a signal at the receiving end of the present invention. A receiving device 10 may be a transceiver of model AD9371 produced by Analog Devices (ADI), and the receiving device 10 Includes a receiver 12 which can divide the signal into phase channel (I channel) and amplitude channel (Q channel), and the I channel and Q channel are connected to four front-end filters 14 respectively. , , , ), The front-end filter 14 is connected to two front-end downconverters 16 respectively, and the front-end downconverter 16 will generate a front-end frequency offset. , . Two front-end downconverters 16 are further connected to four rear-end filters 18 ( , , , ), The rear-end filter 18 is then connected to two rear-end downconverters 19, and the rear-end downconverter 19 will generate a rear-end frequency offset. , . In this embodiment, the front-end filter 14 and the back-end filter 18 may be parallel filters. When the signal passes through the front-end filter 14 and the back-end filter 18, a phase amplitude (IQ) imbalance will be generated. Therefore, if the front-end can be estimated, The parameters of the filter 14 and the back-end filter 18 can be used to compensate the broadband radio frequency imperfection factor of the receiving device 10 to compensate for the IQ imbalance in the radio frequency signal.

After explaining the structure of the receiving device 10 in this embodiment, please refer to the second figure at the same time. As shown in the figure, first enter step S10. The receiving device 10 establishes a joint signal model with radio frequency imperfection, in which the joint signal The model can be expressed as the following equation (1): (1) (2) (3) of which To receive the signal, , Is the imperfection factor for wideband RF, Is the ideal signal, Is the DC offset.

Then proceed to step S12, input an ideal signal Generate a received signal to the joint signal model The parameters of the front-end filter 14 are estimated and compensated. In this embodiment, the ideal signal is a binary phase shift shifting (BPSK) signal, and there is no phase amplitude (IQ) at the transmitting end. ) Unbalanced effect, so in this embodiment, the parameters of the front-end filter 14, the back-end filter 18, and the DC offset can be simply considered, so that the front-end filter 14, the back-end filter 18, and the The shifted parameters are estimated, and the received signal is compensated according to the parameters. The detailed steps for estimating the parameters of the front-end filter 14 are shown below. In order to make the received signal be a real number signal, the above equations (2) and (3) are substituted into equation (1), and Expand to get the following equation (4): (4) Continue to expand the above equation (4), and reduce the received signal into phase channel (I channel) signal and amplitude channel (Q channel) signal. The two signals can be obtained from the following two equations (5) (6): (5) (6) After obtaining the two signals of the I channel signal and the Q channel signal, define the relationship between the four front-end filters 14 and the I-channel signal and the Q-channel signal, and define the front-end filter 14 as , , , , And transform equation (5) and equation (6) into the following equation (7): (7) of which I channel signal, Is the Q channel signal, , , , Is the parameter of the front-end filter, Is the I channel signal, Is the Q channel signal, , Is the DC offset parameter of the front end. The ideal value of the four front-end filters 14 can be expressed by the following equation (8) (8) The received signal with the IQ imbalance effect is shown in equation (7). , And the training code known to be transmitted in the ideal signal Arranged into a matrix, the relationship is as follows equations (9) (10): (9) (10) of which Are The convolution matrix term, and then let the common terms of equations (9) and (10) As shown in the following formula (11): (11) Finally, use the least square method (LS) to find , , , The best solution is as follows (12): (12) Estimate the parameters of the front-end filter 14 of the receiving device 10 , , , Then, the imperfect signal generated by passing through the front-end filter 14 is compensated.

Next, step S14 is performed to estimate the parameters of the DC offset. Please refer to the first figure for the DC offset value of the front-end downconverter 16 of the receiving device 10. After four back-end filters 18 ( , , ), The DC offset value from the rear end of the downconverter 19 Cancellation. In this way, the DC offset at the receiving end can be eliminated. Its mathematical formula is the following equations (13) (14): (13) (14) Know the DC offset of the front end And back-end DC offset values After the relationship, you can use equation (13) to arrange into a matrix, and you can find the DC offset of the back end. , As shown in the following equation (15): (15)

Continuing, the process proceeds to step S16, in order to obtain four back-end filters 18 of the receiving device 10 ( , , The parameters of the ideal signal after analog to digital sampling are: Since the receiving device 10 does not pass the signal of the IQ imbalance effect, the ideal signal passes the four front-end filters 14 of the receiving device ( , , , After that, four back-end filters 18 ( , , ) Compensate the effect of the backend on the received signal, and divide the I channel signal and the Q channel signal into two channels. That is, the signal after correction by the post-processing of the receiving device 10, the overall mathematical formula can be arranged as the following equations (16) (17): (16) (17) In the above equations (16) and (17) It is proposed to rearrange the mathematical formula as follows (18) (19): (18) (19) Take I channel signal as an example, use equation (18) and The relationship between the two is that only the part of the I channel signal is kept equal to a pulse function, and the part of the Q channel signal is made equal to zero. The mathematical formula is the following equation (20): (20) The above formula can be rearranged as the following equation (21): (21) of which for Convolution matrix entry , Is a pulse function. Then arrange the above equation (21) into a matrix and use the least square method (LS) to find with , Its equation (22) is as follows:

Continue the same concept, and use the relationship between equation (19) y ^Q ( n ) and z ^Q ( n ) for the Q channel signal. Keep the Q channel signal part equal to a pulse function, and the I channel signal part is Zero, the relationship is as follows equation (23): Rearrange the above formula into the following equation (24): Where E is also equation (21), and finally the above equations are arranged into a matrix, as shown in equation (25): And by using the least square method (LS) to obtain f ₃ and f ₄ , four back-end filters 18 ( f ₁ , f ₂ , f ₃ , f ₄ ) parameters can be obtained. The resulting imperfect signals are compensated.

In addition to the above embodiments, the present invention performs correction of a wideband signal only for a single transmitting end or a receiving end, and can perform correction of a two-way wideband signal. That is, the present invention can perform correction for a receiving end and a transmitting end simultaneously. Details are as follows.

First, please refer to the third figure, which is a system architecture diagram of the receiving device and the transmitting device in this embodiment. As shown in the figure, it includes a receiving device 10 and a transmitting device 20. The structure of the receiving device 10 is the same as that of the foregoing embodiment, so the description will not be repeated. The transmitting device 20 is a device for transmitting a signal. The transmitting device 20 includes a receiver 22, which can divide the signal into an I channel and a Q channel, and the I channel and the Q channel are respectively connected to four back-end filters 24 (g ₁ , g ₂ , G ₃ , g ₄ ), the back-end filter 24 is connected to two back-end downconverters 26 respectively, and the back-end downconverter 26 will generate back-end frequency offsets ( a ^I , a ^Q ). The two rear-end downconverters 26 are further connected to four front-end filters 28 ( h ₁ , h ₂ , h ₃ , h ₄ ), and the front-end filters 28 are respectively connected to the two front-end downconverters 29, and the front-end downconverters 29 are also There will be front-end frequency offsets ( b ^I , b ^Q ). In this embodiment, the rear-end filter 24 and the front-end filter 28 may be parallel filters. When the signal passes through the rear-end filter 24 and the front-end filter 28, a phase amplitude (IQ) imbalance will be generated. Therefore, if it can be estimated The parameters of the end filter 24 and the front filter 28 can be used to compensate the broadband radio frequency imperfection factor of the transmitting device 20 to compensate the IQ imbalance in the radio frequency signal.

Next, please refer to the fourth figure to explain the steps and procedures of this embodiment. This embodiment proposes that the calibration process is to first adjust the receiving device 10 and then adjust the transmitting device 20. First, it proceeds to step S20 to establish a joint signal model with radio frequency imperfection in the receiving device 10, and the mathematical model is expressed as the following equation (26):

b = b ^I + j. b ^Q (28) where x ( t ) is the received signal, h _{R, +} ( t ), h _{R, -} ( t ) represent the overall broadband RF imperfection factor in the receiving device 10, and s ( t ) is the ideal signal as well as Then the I / Q low-pass filter impulse responses are respectively expressed, α _T is the I / Q amplitude imbalance and θ _T is the I / Q phase imbalance, and b is the DC offset of the transmitting device 10.

Then proceed to step S22, and with reference to the third figure, the front end of the transmitting device 20 generates an ideal signal s ( t ), and the ideal signal expression is as follows: Equation (29): s ( t ) = s ^I ( t ) + j . 0 (29) The ideal signal s ( t ) passes through the rear-end filter 24, rear-end down-converter 26, front-end filter 28, and front-end down-converter 29 of the transmitting device 20 in this order to generate a transmission signal x ( t ), We can get the signal of x ( t ). Its mathematical model is the following equation (30) (31) (32):

b = b ^I + j. b ^Q (32) where h _{R, +} ( t ), h _{R, -} ( t ) represent the overall broadband RF imperfection factor of the transmitting device, as well as Then it represents the impulse response of the two low-pass filters of the I channel and the Q channel, respectively, α _T is the amplitude imbalance of the I and Q channels and θ _T is the phase imbalance of the I and Q channels, and b represents the DC offset of the transmitting device 20. In order for the signal to be expressed as a real signal, substituting equation (31) into equation (30) and expanding the common terms, the derivation result is the following equation (33): In which It is a filter constant term, which has nothing to do with IQ imbalance (Imbalance), so this value is omitted, and the signal at the transmitting end can be obtained by the following equation (34): x ( t ) = s ^I ( t ) + b (34) This signal is only retained The filter effect of the I channel has no filter effect affected by the Q channel, so it can achieve the purpose of isolating the imperfection factor of the broadband RF at the transmitting end.

Continuing, after completing the signal model transmitted by the transmitting device 20, a frequency offset of the center frequency is intentionally generated between the transmitting device 20 and the receiving device 10, and then the receiving device 10 receives the signal, and the optimization of the frequency offset is used here The correction principle makes the signal have a deviation value e ^{j 2π △ ft} between the transmitting device 20 and the receiving device 10, where the frequency offset term represented by μ is the center frequency f _T of the transmitting device 20 and the center of the receiving device 10 The frequency f _R is subtracted. This frequency deviation value e ^{j 2π △ ft} can optimize the correction performance of the receiving device 10, and then the signal received by the receiving device 10 can be expressed by the following equation (35) after analog rotation digital sampling: y ( t ) = e ^{j 2π △ ft} . x ( t ) (35)

Continuing, after sampling by analog to digital, let μ = △ f. T _S , substituting equation (34) into equation (35), that is, substituting the ideal signal of the transmitting device 20 into the receiving signal, and expanding the e ^{j 2π μn} term in y ( n ) to obtain the following formula (36): Expanding the above equation (36), we get with For subsequent derivation, the following formula (37) is developed as follows: The cos and sin items in the above formula are replaced as follows for the convenience of subsequent derivation. The program after replacement (38) is shown below:

Then, based on the imperfect joint signal model of the receiving device 10, the equations (39), (40), and (41) of the received signal can be obtained as follows:

d = d ^I + j. d ^Q (41) where h _{R , +} ( n ), h _{R ,-} ( n ) represents the 10-band radio frequency imperfection factor of the receiving device, as well as The low-pass filter impulse response of the I channel and Q channel of the receiving device 10 respectively, and then pass through an I channel and Q channel amplitude imbalance α _R and I channel and Q channel phase imbalance θ _R of the I channel and Q The channel downconverter, d represents the DC offset of the receiving device 10.

In order to represent the signal as a real number, substituting equations (41) and (40) into equation (39), and expanding the two channels of I channel and Q channel, the following formula (42) is obtained: Expand the exp term in equation (42) as the following equation (43): Then, the above formula is simplified and divided into two signals of I channel and Q channel to obtain two equations (44) and (45):

After obtaining the two signals of I channel and Q channel, four front-end filters 14 can be defined as e ₁ ( n ), e ₂ ( n ), e ₃ ( n ), e ₄ ( n ), and equation (44) And equation (45) is transformed into equation (46) as follows: The ideal values of the four front-end filters 14 can be expressed by the following equation (47): Using the relationship between equation (37) and equation (38), the equation (46) can be replaced by the following equations (48) (49):

Arrange equation (48) and equation (49) in a matrix, and the relationship is as follows equations (50) and (51):

among them , , C, S matrix terms are each expressed by the following equation (52): Let the common matrix term of equation (50) and equation (51) be the following equation (53): At the same time, the equation (50) and the equation (51) are calculated by the least square method (LS) as the following equations (54) and (55):

The receiving end can have four front-end filter _{14 (e 1, e 2,} e 3, e 4) of the optimal solution, and after imperfect signal generated by front-end filter 14 to compensate.

Then, step S24 to step S26 are performed. The methods of correcting the DC offset and the parameters of the back-end filter 18 in steps S24 to S26 are the same as those in the above-mentioned implementation, so they will not be repeated here.

Then proceed to step S28, and continue to perform compensation for the transmitting device 20. Please continue to refer to the third and fourth figures. First, a joint signal model with radio frequency imperfection is established in the transmitting device 20, which is as shown in the following equation (56)

b = b ^I + j. b ^Q (59) where x ( t ) is the received signal, S _p ( t ) is the pre-compensated complex signal, h _{T, +} ( t ), h _{T, -} ( t ) are the wideband RF Perfection factor, as well as Then it represents the impulse response of the two low-pass filters of the I channel and the Q channel, respectively, α _T is the amplitude imbalance of the I and Q channels and θ _T is the phase imbalance of the I and Q channels, and b represents the DC offset at the transmitting end.

Then proceed to step S30, input an ideal signal S _p ( t ) to the joint signal model to generate a received signal x ( t ), and estimate and compensate for the parameters of the front-end filter 28. The ideal signal is BPSK signal in this embodiment. , There is no effect of the phase amplitude (IQ) imbalance at the transmitting end, so this embodiment can simply consider the parameters of the front-end filter 28, the back-end filter 24 and the DC offset to simultaneously filter the front-end of the joint signal model The parameters of the converter 28, the back-end filter 24, and the DC offset are estimated, and the received signal is compensated according to the parameters. The detailed steps for estimating the parameters of the front-end filter 28 are shown below. In order to make the received signal be a real number signal, substituting the above equations (57) and (58) into equation (56), and expanding y ( t ) to obtain the following equation (60): Then, the equation (60) is expanded and simplified into two signals of the I channel signal and the Q channel signal. The following formulas (61) and (62) can be obtained:

After obtaining two signals of the I channel and the Q channel, four front-end filters 28 can be defined as h ₁ ( t ), h ₂ ( t ), h ₃ ( t ), h ₄ ( t ) and equation (61) And equation (62) is transformed into equation (63) as follows: The ideal values of the four front-end filters 28 can be expressed by the following equation (64): After analog-to-digital sampling, the code becomes x ^I ( n ), x ^Q ( n ), and the training codes s ^I ( n ) and s ^Q ( n ) known to be transmitted in the ideal signal are arranged into a matrix, and the relationship is as follows: 65) (66):

Wherein ^{S I (n), S Q} (n) , respectively ^{S I (n), S Q} (n) of the swirling key matrix, and then using a least squares method (LS) to obtain _{h 1, h 2, h 3} , h The best solution of ₄ . Next, let the common term S _{LS of} equation (65) and equation (66) be the following equation (67): S _LS = [ 1 S ^I ( n ) S ^Q ( n )] (67) Continue the equation (65) and the equation (66) ) After the least square method (LS) is calculated as the following equations (68) (69) (70):

The best solutions of the four front-end filters 28 ( h ₁ , h ₂ , h ₃ , h ₄ ) can be obtained, and the imperfect signals generated by passing through the front-end filters 28 can be compensated.

After the parameters h ₁ , h ₂ , h ₃ , and h ₄ of the four front-end filters 28 in the transmitting device 20 are estimated, the process proceeds to step S32 to estimate the parameters of the DC offset, please refer to the third In the figure, after the DC offset value a of the downconverter 26 at the rear of the transmitting device 20 has passed the four front-end filters 28 ( h ₁ , h ₂ , h ₃ , h ₄ ), the DC offset from the front-end downconverter 28 is required. cancellation values b, to remove the DC offset, which equation the following equation (71) (72): a I. 1 ^T. h ₁ + a ^Q. 1 ^{T. _{h 2 + b I = 0 a}} I. 1 ^T. h ₃ + a ^Q. 1 ^T. h ₄ + b ^Q = 0 (71)

a = d + j. a ^Q (72) After knowing the relationship between the back-end DC offset value a and the front-end DC offset value b, h ₁ , h ₂ , h ₃ , h ₄ estimated by equation (68) and equation (69) can be used By arranging into a matrix, the back-end DC offset value a can be obtained. The matrix can be arranged as the following equation (73), and then using the inverse matrix of h ₁ , h ₂ , h ₃ , h ₄ to obtain the back-end DC offset. The offset value a is the following equation (74):

And finally into the step S34, for the four front-end filter _{28 (h 1, h 2,} h 3, h 4) as compensation, after sampling by the analog digital ^{revolution, s I (n), s} Q (n) of the emitter 20 The signals of the fundamental frequency I-channel and Q-channel pass four back-end filters 24 (g ₁ , g ₂ , g ₃ , g ₄ ) and then four front-end filters 28 ( h ₁ , h ₂ , h ₃ , h ₄ ), and divide into two channels of I channel and Q channel, x ^I ( n ), x ^Q ( n ) are the signals corrected by the rear end of the transmitting device 20, and the overall mathematical formula can be arranged as the following equation (75) (76):

To find the parameters of the four back-end filters 24 (g ₁ , g ₂ , g ₃ , g ₄ ), put s ^I ( n ) and s ^Q ( n ) in the above formula, and rearrange the mathematical formula as follows ( 77) (78):

Taking the I channel as an example, using the relationship between equation (77) s ^I ( n ) and x ^I ( n ), keeping only the part of the I channel to make it equal to a pulse function, then the part of the Q channel makes it equal to zero. Mathematical formula Such as the following formula (79): The above formula can be rearranged as the following equation (80):

δ = [1 0… 0] ^T (81) where H _k is the convolution matrix term of h _k k = 1,2,3,4, and δ is a pulse function. Then arrange equation (80) into a matrix, and use the least square method (LS) to find g ₁ and g _3. Equation (82) is expressed as follows:

Continuing the same concept, the Q branch uses the relationship between equations (78) s ^Q ( n ) and x ^Q ( n ), keeping the part of the Q channel equal to a pulse function, and the part of the I channel making it Zero, the relationship is as follows equation (85): Rearrange the above equation (85) into the following equation (86): H is the same as equation (84), and finally the equation (86) is arranged into a matrix, as shown in the following equation (87):

Finally, the least square method (LS) can be used to obtain g ₂ and g ₄ to obtain the parameters of the four back-end filters 24 (g ₁ , g ₂ , g ₃ , and g ₄ ). 24 to compensate for imperfect signals.

In the present invention, the signal correction using the method of the above embodiment can obviously correct the signal into a more perfect signal. Next, various experimental data are exemplified to prove that the signal can be corrected into a more perfect signal using the method of this embodiment. Please refer to FIG. 5, FIG. 6a, FIG. 6b, FIG. 7a, and FIG. 7b. Since the method and result of correcting the signal by the transmitting device and the receiving device are the same, this embodiment only uses the receiving device 10 is explained as an example. First please refer to the fifth figure, which is a system architecture diagram of the experimental data of the detection signal, which includes a transmitter 30 which can be a model AD9371 transceiver produced by Analog Devices (ADI), and the transmitter 30 signal connection The receiving device 10 transmits a signal to the receiving device 10, and the frequency of the signal transmitted by the transmitter 30 is set to a single-frequency signal of 10 MHz. And a computer host 32 signal is connected to the receiving device 10, so that the corrected signal of the receiving device 10 can be transmitted to the computer host 32 to use the analysis software (MATLAB) in the computer host 32 to analyze and calculate the single frequency signal via the transmitting device 10 After the corrected signal, various experimental data of the corrected signal are displayed on the display screen 34 on the computer host 32.

The analysis of the computer host 32 (MATLAB) is as follows. First, please refer to Figure 6a. The original image rejection ratio (IMRR) before correction was 44.6274dBm, but after the signal is corrected by the receiving device 10, please refer to section 6. Figure 6b, IMRR can be increased to about 61dB. For the multi-carrier part, please refer to Figure 7a. The uncorrected Error Vector Magnitude (EVM) is maintained at about 33dB. After the receiving device 10 corrects the signal, please refer to Figure 7b. EVM can Increase to about 41dB.

In summary, the present invention can correct imperfect signals to solve the problems of signal offset and DC offset caused by the signal passing through the filter, and can simultaneously target the receiving end and the transmitting end. Compensate for imperfect signals to compensate for imperfect signals.

The foregoing are merely preferred embodiments of the present invention, and are not intended to limit the scope of implementation of the present invention. Therefore, all equal changes or modifications made according to the features and spirit described in the scope of the application of the present invention shall be included in the scope of patent application of the present invention.

10‧‧‧ receiving device

12‧‧‧ Receiver

14‧‧‧ front-end filter

16‧‧‧Front-end downconverter

18‧‧‧ back-end filter

19‧‧‧back-end downconverter

20‧‧‧ launcher

22‧‧‧ Receiver

24‧‧‧Back-end filter

26‧‧‧Back-end downconverter

28‧‧‧ front-end filter

29‧‧‧Front-end downconverter

30‧‧‧ launcher

32‧‧‧Computer host

34‧‧‧display

The first figure is a block diagram of a receiving device system of the present invention. The second figure is a flowchart of the steps for the receiver to correct the imperfect signal. The third figure is a block diagram of a receiving device and a transmitting device system of the present invention. The fourth figure is a flowchart of the steps of the receiving device and the transmitting device of the present invention. The fifth figure is a block diagram of a system for generating experimental data according to the present invention. Figure 6a is the IMMR analysis diagram of the signal before compensation according to the present invention. Figure 6b is the IMMR analysis signal of the signal after compensation according to the present invention. Figure 7a is the EVM analysis chart of the signal before compensation according to the present invention. Figure 7b is an EVM analysis diagram of the signal after compensation according to the present invention.

Claims (11)

A calibration method for improving the quality of a broadband radio frequency signal includes the following steps: establishing a joint signal model with radio frequency imperfection; inputting an ideal signal into the joint signal model to generate a reception signal for a front-end filter of the joint signal model Make an estimate, estimate the parameters of the front-end filter, and then compensate the received signal based on the parameters of the front-end filter; estimate the DC offset of the received signal to estimate the DC offset parameter; and The rear-end filter of the joint signal model is estimated to estimate the parameters of the back-end filter, and the received signal is compensated based on the parameters of the back-end filter and the parameter of the DC offset. The adjustment method for improving the quality of a broadband radio frequency signal as described in claim 1, wherein the joint signal model can be expressed as the following equation: The r ( t ) is the received signal, the h _{R, +} ( t ), h _{R, -} ( t ) are broadband imperfection factors, the y ( t ) is the ideal signal, and d is the DC offset . The adjustment method for improving the quality of a broadband radio frequency signal according to claim 1, wherein the step of estimating the parameters of the front-end filter further includes the following steps: dividing the received signal into a phase channel (I channel) signal and an amplitude channel ( Q channel) signal; define the relationship between the front-end filter, the phase channel signal and the amplitude channel signal, expressed by the equation as follows: Where r ^I ( t ) is the phase channel signal, the r ^Q ( t ) is the amplitude channel signal, and e ₁ ( t ), e ₂ ( t ), e ₃ ( t ), and e ₄ ( t ) are The parameters of the front-end filter, the y ^I ( t ) is the phase channel signal, the y ^Q ( t ) is the amplitude channel signal, the d ^I , d ^Q are the DC offset parameters of the front-end; The amplitude channel signal is digitally sampled by analogy to form a matrix with the training code in the ideal signal; and the least square (LS) algorithm is used to obtain the phase channel signal and the amplitude channel signal. Corresponding parameters of each parallel filter. The adjustment method for improving the quality of a broadband radio frequency signal as described in claim 3, wherein the phase channel signal and the amplitude channel signal are digitally sampled by analog rotation, and arranged in a matrix with the training code in the ideal signal, which can be expressed as The following equation: Wherein the r ^I (n) for the analog switch of the phase channel signal after sampled digital, the r ^Q (n) of analog switch the amplitude of the channel signal after sampled digital, the Y ^I (n) for the y ^I (n) the matrix of swirl, the Y ^Q (n) for y ^Q (n) of the matrix of swirl, the _{e 1, e 2, e 3} , e 4 of the front-end filter parameters, the d ^I, d ^Q DC front end of Offset parameter. The adjustment method for improving the quality of a broadband RF signal as described in claim 3, wherein the step of estimating the parameters of the back-end filter further includes the following steps: making the phase channel signal equal to a pulse function and the amplitude channel signal equal to zero, It is represented by equation (4) as follows: Where f ₁ ( n ), f ₂ ( n ) are parameters of the back-end filter sampled by analog rotation, and e ₁ ( n ), e ₂ ( n ), e ₃ ( n ), e ₄ ( n ) is the parameter of the front-end filter sampled by analog rotation, and δ ( n ) is the pulse function; the back-end filter f ₁ ( n ) is estimated using a convolution matrix algorithm and the least square method. f ₂ ( n ); let the amplitude channel signal be equal to the pulse function, and the phase channel signal be equal to zero, which is expressed by the equation as follows: Where f ₃ ( n ), f ₄ ( n ) are parameters of the back-end filter sampled by analog to digital, the e ₁ ( n ), e ₂ ( n ), e ₃ ( n ), e ₄ ( n ) is the parameter of the front-end filter sampled by analog rotation, and δ ( n ) is the pulse function; and the back-end filter f ₃ ( n ) is estimated using the convolution matrix algorithm and the least square method. , F ₄ ( n ). The adjustment method for improving the quality of a broadband radio frequency signal as described in claim 5, wherein the convolution matrix algorithm and the least square method are used to estimate the parameters of the back-end filters f ₁ ( n ), f ₂ ( n ) In the step, the equation of the convolution matrix algorithm is as follows: The E ₁ , E ₂ , E ₃ , and E _{4 are} convolution matrices of the e ₁ , e ₂ , e ₃ , and e ₄ , and the f ₁ and f _{2 are} parameters of the back-end filter. The adjustment method for improving the quality of a broadband RF signal as described in claim 5, wherein the convolution matrix algorithm and the least square method are used to estimate the parameters of the back-end filters f ₃ ( n ) and f ₄ ( n ). In the step, the equation of the convolution matrix algorithm is as follows: Where E ₁ , E ₂ , E ₃ , and E _{4 are} the convolution matrix of e ₁ , e ₂ , e ₃ , and e ₄ , f ₃ and f _{4 are} parameters of the back-end filter, and δ is the pulse function. The method for improving the quality of a broadband radio frequency signal according to claim 1, wherein the step of estimating the parameters of the back-end filter further includes compensating the DC offset, wherein the DC offset includes a front-end DC offset and A back-end DC offset, and the step of compensating the DC offset includes: canceling a parameter of the front-end DC offset and a parameter of the back-end DC offset, which is expressed by an equation as follows: Where c ^I and c ^{Q are} parameters of the back-end DC offset, and f ₁ ( n ), f ₂ ( n ), f ₃ ( n ), and f ₄ ( n ) are parameters of the back-end filter. d ^I and d ^{Q are} parameters of the front-end DC offset; and parameters of the back-end DC offset are obtained to compensate the received signal by using the parameters of the front-end DC offset and the parameters of the back-end DC offset. The adjustment method for improving the quality of a broadband radio frequency signal according to claim 1, wherein in the step of inputting the ideal signal into the joint signal model to generate the reception signal, a transmitting end transmits the ideal signal to a receiving end The combined signal model, the received signal generated by the combined signal model generates a frequency offset, and the analog signal of the received signal is digitally rotated, and then continues to enter the estimation of the front-end filter of the combined signal model, Steps to estimate the parameters of the front-end filter. The method for adjusting the quality of a broadband radio frequency signal as described in claim 9, wherein the step of estimating the parameters of the front-end filter includes: dividing the received signal into a phase channel (I channel) signal and an amplitude channel (Q channel) signal ; Define the relationship between the front-end filter, the phase channel signal and the amplitude channel signal, which is expressed as an equation as follows: The r ^I ( n ) is the phase channel signal after analog to digital sampling, the r ^Q ( n ) is the amplitude channel signal after analog to digital sampling, e ₁ ( n ), e ₂ ( n ), e ₃ ( n ) and e ₄ ( n ) are the parameters of the front-end filter after analog-to-digital sampling. ( n ) is the phase channel signal that generates a frequency offset, the ( n ) is the amplitude channel signal that generates a frequency offset, and d ^I and d ^Q are the DC offset parameters at the front end; the phase channel signal and the amplitude channel signal are arranged in a matrix; and the least square (LS) is used The algorithm obtains the parameters of each parallel filter corresponding to the front-end filters in the phase channel signal and the amplitude channel signal. The adjustment method for improving the quality of a broadband radio frequency signal according to claim 10, wherein the phase channel signal and the amplitude channel signal are arranged in a matrix, which can be expressed as the following equation: Wherein r ^I for the phase channel signal, that the amplitude r ^Q channel signals, the , The received signal to produce a frequency shift of the swirling matrix, C is the toplitz (sin (2π μn)) , S is the toplitz (cos (2π μn)) , which _{e 1, e 2, e 3} , e 4 is The parameters of the front-end filter, the b ^I and b ^Q are the DC offset parameters of the front-end.