JP2008236704A - Wireless communication apparatus - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To estimate a frequency offset accurately while corresponding to both a DC offset and a frequency offset. <P>SOLUTION: After removing a DC offset using a high-path filter from a baseband signal corresponding to a short preamble portion, a frequency offset is estimated at high accuracy, and the frequency offset is removed from the baseband signal received after the completing estimation of the frequency offset. A differential filter with a simple circuit configuration is used as a high-pass filter for removing the DC offset. Even if the level of the DC offset changes due to the gain switching of a low-noise amplifier, the frequency offset can be estimated fully accurately. <P>COPYRIGHT: (C)2009,JPO&INPIT

Description

  The present invention relates to a radio communication apparatus that receives a radio signal modulated by OFDM (Orthogonal Frequency Division Multiplexing), and in particular, receives an OFDM signal by a direct conversion method that does not use an intermediate frequency (IF) stage. The present invention relates to a wireless communication device to be processed.

  More particularly, the present invention relates to a radio communication apparatus that performs demodulation processing of an OFDM symbol by removing a frequency offset using a training sequence added to the head of a packet, and more particularly, a DC offset that varies with time in a received OFDM symbol. The present invention relates to a radio communication apparatus that performs accurate estimation of a frequency offset in a situation where there is simultaneous IQ imbalance.

  A wireless network is attracting attention as a system free from wiring in the conventional wired communication system. As a standard for a wireless network, IEEE (The Institute of Electrical and Electronics Engineers) 802.11 or the like can be cited.

  For example, when a wireless network is constructed indoors, there is a problem in that a multipath environment is formed in which a receiving device receives a superposition of a direct wave and a plurality of reflected / delayed waves. Multipath causes delay distortion (or frequency selective fading), and causes an error in communication. Intersymbol interference resulting from delay distortion occurs. For this reason, in the wireless LAN standard such as IEEE802.11a / g, an OFDM modulation scheme which is one of the multicarrier schemes is adopted.

  An OFDM transmitter assigns multiple data to be output by serial / parallel conversion of information sent serially for each symbol period slower than the information transmission rate, and modulates amplitude and phase for each subcarrier. And performing inverse FFT on the plurality of subcarriers to convert each subcarrier on the frequency axis into a signal on the time axis for transmission. The OFDM receiver performs the reverse operation, that is, performs FFT to convert the time-axis signal to the frequency-axis signal, demodulates each subcarrier in accordance with the modulation method, and performs parallel / serial conversion. The information sent with the original serial signal is reproduced. Here, the frequency of each carrier is set so that each subcarrier is orthogonal to each other within the symbol interval. The fact that subcarriers are orthogonal to each other means that the peak point of the spectrum of an arbitrary subcarrier always coincides with the zero point of the spectrum of another subcarrier and there is no crosstalk with each other. Therefore, since transmission data is distributed and transmitted to a plurality of carriers whose frequencies are orthogonal to each other, each carrier has a narrow band, has very high frequency utilization efficiency, and is resistant to frequency selective fading interference. By using an FFT (Fast Fourier Transform) algorithm, an efficient OFDM modulator / demodulator can be implemented. In addition to the wireless LAN, the OFDM transmission method includes various digital broadband communications such as terrestrial digital broadcasting (for example, see Non-Patent Document 6), fourth-generation mobile communication, and power line communication (Power Line Communication). Applied to the system.

At the RF front end of a wireless communication device, an analog baseband signal is usually up-converted to an RF band by a frequency converter (orthogonal modulator) at the time of transmission, and then band-limited by a bandpass filter. In general, the transmission power is amplified by a variable gain amplifier circuit. At the time of reception, the signal received by the antenna is amplified by a low noise amplifier (LNA) and then down-converted to a baseband signal using the local frequency _fLO . An automatic gain control circuit (AGC: Auto Gain Control) is used to keep the current of its own signal at an appropriate constant level.

Further, in recent wireless communication devices, a “direct conversion” method is adopted in which direct frequency conversion is performed using the carrier frequency f _c for the local frequency f _LO as a frequency converter for up-converting or down-converting transmission / reception signals. ing. The direct conversion method does not use an external IF (intermediate frequency) filter (also referred to as an RF interstage filter), so it is more suitable for miniaturization and integration than a superheterodyne configuration and reduces power consumption. In addition to being able to generate spurious frequencies in principle, the design of the transceiver is excellent. On the other hand, in the direct conversion of the receiving system, since the reception frequency and the local frequency are equal, there is a problem that a direct current component, that is, a DC offset is generated in the downconverter output due to self-mixing of the local signal (for example, (See Non-Patent Document 1). Self-mixing is due to the limited insulation between the local signal and the low-noise amplifier or mixer radio port. As a definition of the term, the 0 Hz position (zero IF) of the baseband signal in the OFDM modulation scheme is called DC.

  By the way, in the OFDM communication system, due to a subtle error in the frequency of the oscillator mounted on each of the transceivers (for example, an oscillator with an accuracy of about 20 ppm is used in a wireless LAN), the frequency on the receiver side. There is a problem that an offset occurs. The subcarriers do not interfere with each other originally, but if there is a frequency offset, the frequency orthogonality between the subcarriers is hindered, and the demodulation characteristic is deteriorated, that is, the received data is erroneous.

  For this reason, in packet-switched wireless communication systems such as IEEE802.11, a symbol that is known between the transmitter and the receiver, that is, a training sequence is provided at the beginning of the packet, and the receiver side uses the received training sequence to reduce the frequency. Performs automatic gain control of noise amplifier, DC offset estimation and removal, frequency offset estimation and removal, packet detection and timing detection. When the frequency offset processing is performed by an analog circuit, the circuit configuration becomes complicated and the power consumption increases. Therefore, the present inventors consider that it is preferable to perform these estimation / compensation processing by digital processing. The frequency offset is corrected by reversely rotating the phase of the data corresponding to the observed frequency offset.

Here, the frequency offset problem will be considered using IEEE 802.11a / g as an example. FIG. 15 shows a preamble configuration defined by IEEE 802.11a / g (see, for example, Non-Patent Documents 7 and 8). As shown in the drawing, a short preamble section of 8.0 microseconds and a long preamble section of 8.0 microseconds are added to the head. In the short preamble section, short preambles t _{1 to} t ₁₀ composed of a short training sequence (STS) are repeatedly sent 10 times. In addition, in the long preamble interval, 1.8 after the microseconds of guard interval (Guard Interval) GI2, long training series: long preamble T ₁ ~T ₂ consisting of a (Long Training Sequence LTS) is repeated twice Sent. One short preamble symbol is composed of 12 subcarriers is 0.8 microsecond length corresponding to a quarter of the IFFT / FFT period T _FFT. On the other hand, one long preamble is composed of 52 subcarriers and has a length of 3.2 microseconds corresponding to the IFFT / FFT period T _FFT . Also, as shown in FIG. 30, the OFDM signal does not use a DC subcarrier at 0 Hz, and avoids interference with a DC offset.

  In IEEE802.11a / g, the use of the preamble is not particularly defined. However, a general receiver uses four STS symbols of 0.8 microseconds to perform receiver gain setting and DC offset correction, and further uses the remaining six STS symbols to perform frequency offset. Estimation and correction, packet detection, and coarse timing detection.

The frequency offset estimation can be derived according to the following formula (1) using the periodicity of 0.8 microseconds of STS. Where T _sts is 0.8 microseconds, S (i) is an STS signal sampled at 20 MHz, S ^* (i) is its complex conjugate, and M is the average number of samples.

  Further, packet detection and coarse timing detection can be detected from the correlation value shown in the following equation (2) normalized by the power level of the STS. N is the average number of samples.

  In packet detection, a threshold level is set in the correlation value of the above two formulas, and if the value exceeds that value, it is determined that the packet is a packet. In coarse timing detection, it is possible to derive the coarse timing by comparing with the correlation value obtained at the previous timing using the characteristic that the correlation value changes from increasing to decreasing from the end timing of STS. it can.

  As described above, the receiver side performs automatic gain control of the low noise amplifier, DC offset estimation and removal, frequency offset estimation and removal, packet detection and timing detection using the preamble portion of the packet.

  However, the accuracy of frequency offset estimation, packet detection, and timing detection is easily affected by the DC offset, and accurate estimation cannot be performed even if the frequency offset is estimated while the DC offset exists. is there. In particular, in the direct conversion system described above, the problem of DC offset is serious due to self-mixing, and the quality of the received signal is impaired not only by the frequency offset but also by the DC offset.

  For example, when there is a DC offset at the input of the I axis and the Q axis, the correlation value increases even in a no signal state. That is, since it always increases by 1, the correlation value continues to increase and exceeds a threshold value that is a reference for packet detection. For this reason, the receiver recognizes that a packet has been received even in a no-signal state, and malfunctions.

  Further, when a DC offset exists in the input of the I axis and the Q axis, the correlation value does not change from an increasing tendency to a decreasing tendency due to the influence of the DC offset even in a portion where the received signal has no correlation. For this reason, the rough timing detection characteristics deteriorate.

  Further, when there is a DC offset, the estimation accuracy of the frequency offset deteriorates, and the residual frequency offset further deteriorates the characteristics of the received signal. The frequency offset that could not be removed is the phase rotation of the entire subcarrier of the OFDM symbol following the training sequence, and an error floor that does not eliminate packet errors is likely to occur even if the SN ratio is increased. Conversely, even if the DC offset is estimated while the frequency offset exists, accurate estimation cannot be performed. For this reason, it is necessary to deal with both DC offset and frequency offset.

In front of the circuit module which performs a frequency offset correction, noted and it is necessary to perform highly accurate DC offset correction in the first 4 symbols t ₁ ~t ₄ short as within time STS (described above). In general, it is difficult to realize DC offset correction with high accuracy in a short time, and the power consumption and circuit scale are remarkably increased.

  Conventionally, a method of removing a DC offset using a high pass filter (HPF), a method of estimating a DC offset and a frequency offset simultaneously, a method of estimating a DC offset and a frequency offset in parallel, DC offset estimation and frequency offset compensation The method of performing repeatedly etc. is mentioned.

  FIG. 16 schematically shows a configuration of a receiver that removes a DC offset using HPF (see, for example, Non-Patent Document 2). According to the illustrated receiver, the DC offset component included in the received OFDM symbol is removed by HPF, the frequency offset is estimated by subsequent signal processing, and the frequency offset is removed from the OFDM symbol following the training sequence. However, in this method, since a signal in the vicinity of the DC of the OFDM symbol is attenuated by HPF, the demodulation characteristics may be deteriorated.

So as not to attenuate the near-DC signal, the cut-off frequency f _c of the HPF, (see FIG. 17A) sufficiently may be so reduced to the sub-carrier interval. However, when the gain of the low-noise amplifier is changed by automatic gain control, there is a problem that the DC offset fluctuates with time (see, for example, Non-Patent Document 9). Since the cut-off frequency f _c is the response of the HPF is slow and small, DC offset varying time will pass through the HPF.

In the preamble configuration shown in FIG. 15, for example, the low noise amplifier is switched from a high gain to a low gain in the vicinity of the center of the short preamble section. would include (see FIG. 31 (a)), where, since the response of the HPF and the cut-off frequency f _c is smaller becomes slow, high-frequency components of the DC offset varies time by passing through the HPF In this case, the frequency offset estimation unit in the subsequent stage is passed through. When such a residual DC offset overlaps the subsequent long preamble section (see FIG. 31 (b)), it also affects the precise frequency offset estimation performed in the long preamble section. Accuracy will be reduced.

For example, since IEEE802.11a / g has a very short preamble period, the residual DC offset must be converged at high speed with HPF. To this convergence time to a minimum, it must be very large cut-off frequency f _c of the HPF (e.g., see non-patent document 5).

However, when a large cut-off frequency f _c of the HPF, while the tracking of the switching DC offset due to the gain switching of the low-noise amplifier better, because thus scraped until a valid signal in the vicinity of DC (see FIG. 17B ), The OFDM demodulation characteristics are degraded.

Since it is required convergence speed of the transient response to the HPF, the cut-off frequency f _c must be increased. In the preamble configuration shown in FIG. 15, in the STS, the DC component and the closest subcarrier have an interval of 1.25 MHz, so that even if the cutoff frequency is increased, the signal of the subcarrier near DC is deleted. There is no. On the other hand, since the subcarrier closest to DC is only 312.5 kHz after LTS, insertion of HPF deletes the signal of the subcarrier in the vicinity of DC, resulting in deterioration of demodulation characteristics.

FIG. 18 schematically shows a configuration of a receiver that simultaneously estimates a DC offset and a frequency offset (for example, see Non-Patent Document 3). In the illustrated receiver, the DC offset and the frequency offset are simultaneously estimated and compensated using maximum likelihood estimation. However, since maximum likelihood estimation requires a large amount of calculation and takes a long calculation time, it is difficult to implement it in a system in which the time for performing offset compensation is limited. In the preamble configuration shown in FIG. 15, the DC offset estimation needs to be completed using the first STS of about t ₁ to t ₄ , that is, 3.2 microseconds.

  FIG. 19 schematically shows the configuration of a receiver that estimates a DC offset and a frequency offset in parallel (see, for example, Non-Patent Document 4 and Patent Documents 1 and 2). The DC offset estimator estimates the DC offset by averaging the entire preamble. The subsequent frequency offset estimation unit can calculate a correlation function of the preamble signal, subtract the estimated DC offset, and estimate an accurate frequency offset from the signal from which the DC offset is removed. However, if the level of the DC offset changes due to switching of the gain of the low noise amplifier during the preamble reception, the estimation of the DC offset becomes inaccurate.

  FIG. 20 schematically shows a configuration of a receiver that repeatedly performs DC offset estimation and frequency offset compensation (see, for example, Patent Documents 3 to 6). According to the illustrated receiver, the DC offset is removed by the DC offset removing unit, and then the frequency offset is estimated. After compensating for the frequency offset, the DC offset is estimated and the residual DC offset is further canceled. This method requires time for convergence of the feedback loop for removing the DC offset, and is difficult to apply to a short preamble. Further, if the DC offset varies due to the gain switching of the low noise amplifier or the like, an error occurs in the estimation of the frequency offset.

  It is known that the DC offset level fluctuates in response to switching the gain set in the low noise amplifier. However, in any of the receivers shown in FIGS. 18 to 20, it is not sufficiently considered that the DC offset fluctuates in time with the gain switching in the low noise amplifier.

  On the other hand, the direct conversion OFDM receiver has a problem of IQ imbalance as well as DC offset due to self-mixing of local signals. In the direct conversion method, since there is no IF signal in the digital domain, IQ quadrature demodulation must be performed in the analog domain instead of the digital domain. For this reason, IQ imbalance arises by the unbalanced component of in phase (I) and quadrature phase (Quadrature: Q). In particular, the IQ imbalance related to the phase is caused by the fact that the phase difference between the local signals input to the mixers of the I and Q channels is not exactly 90 degrees. This is caused by a gain difference between channel signals (see, for example, Non-Patent Document 5). IQ imbalance is a factor that degrades the estimation accuracy of the frequency offset together with the DC offset, and also affects the decoding characteristics.

  Thus, in the OFDM receiver configuration that employs the direct conversion method, the quality of the received signal is impaired due to the frequency offset, DC offset, and IQ imbalance. None of the receivers shown in FIGS. 18-20 considers DC offset temporal variation and IQ imbalance at the same time.

US Patent Publication No. 2003/0174790 US Patent Publication No. 2005/0078509 US Patent Publication No. 2005/0020226 US Patent Publication No. 2003/0133518 US Patent Publication No. 2004/0202102 US Patent Publication No. 2005/0276358 Annu Batra, “03267r1P802-15_TG3a-Multi-band-OFDM-CFP-Presentation.ppt”, pp. 17, July 2003. W. Namgoong, T .; H. Co-authored by Meng, “Direct-Conversion RF Receiver Design” (IEEE Trans. On Commun. Vol. 49, No. 3, March 2001). G. T.A. Gil, I.D. H. Sohn, J .; K. Park, Y. et al. H. Lee, “Joint ML Estimation of Carrier Frequency, Channel, I / Q Mismatch, and DC Offset in Communication Receivers,” IEEE Trans. On Ve. C. K. Ho, S.H. Sun, P.M. He co-authored “Low Complexity Frequency Offset Estimation in the Presence of DC Offset” (Proc. Of IEEE International Conferences on Communications 2003, Vol. 3, 205. pp. 205, pp. 205, pp. 205 T.A. Yuba, Y. et al. Sanada co-authored “Decision Directed Schema for IQ Imbalance Compensation on OFDCM Direct Conversion Receiver” (IEICE Trans. On Communications, vol. E89-E. P. J. et al. "WLAN / WCDMA Dual = moe Receiver Architecture Design Trade-Offs" by Olsson (Proc. Of IEEE 6th CAS Symp, vol. 2, pp. 725-728, 31 May-2 June). M.M. “OFDM Modulation Technique” by Itani (Tripes, 2000) IEEE802.11a-Part11: Wireless LAN Medium Access Control (MAC) and Physical Layer (PHY) specifications; Highspeed Physical Layer in the 5GHZ Band. IEEE802.11g-Part11: Wireless LAN Medium Access Control (MAC) and Physical Layer (PHY) specifications; Highspeed Physical Layer in the 2.4GHZ Band. S. Otaka, T .; Yamaji, R .; Fujimoto, H .; Tanimoto, “A Low Offset 1.9 GHz Direct Conversion Receiver IC with Spurous Free Dynamic Range of over 67, 51” (IEC Trans. On Fund. 5 Fund.

  An object of the present invention is to provide an excellent radio that employs a direct conversion method and can appropriately remove a frequency offset and perform OFDM demodulation with higher characteristics in a situation where a DC offset varies with time. It is to provide a communication device.

  A further object of the present invention is to remove the DC offset and perform an accurate estimation of the frequency offset when a time-varying DC offset, IQ imbalance, and frequency offset are simultaneously present in the received OFDM symbol. The object is to provide an excellent wireless communication device.

The present invention has been made in consideration of the above problems, and a first aspect thereof is a wireless communication apparatus that receives a packet composed of an OFDM modulated signal,
A bandpass filter for extracting an OFDM signal in a desired band;
A low-noise amplifier that amplifies the desired OFDM signal, the gain of which is controlled according to the intensity of the received signal;
A frequency converter that down-converts the amplified OFDM signal into a baseband signal;
An analog-to-digital converter that converts a baseband signal into a digital signal;
A first high pass filter for removing a DC offset for a baseband signal corresponding to a predetermined preamble portion of the packet;
A frequency offset estimator for estimating a frequency offset from the signal after removing the DC offset by the first high-pass filter;
A frequency offset correction unit that removes the estimated frequency offset from the received baseband signal used for estimating the frequency offset;
A demodulator that demodulates the subcarrier signals arranged on the frequency axis from the baseband signal after compensating for the frequency offset;
A wireless communication device comprising:

  The present invention relates to a wireless communication apparatus that receives an OFDM signal by a direct conversion method. According to the direct conversion method, since the IF filter is not used, it is easy to widen the receiver and the configuration of the receiver is increased. However, the DC offset is generated by the self-mixing of the local signal. There is a problem of affecting timing detection.

  Therefore, the wireless communication apparatus according to the present invention estimates the frequency offset with high accuracy after removing the DC offset from the baseband signal corresponding to the predetermined preamble portion of the packet using the first high-pass filter. Then, the estimated frequency offset is removed from the received baseband signal after the estimation of the frequency offset is completed.

  Here, a differential filter having a simple circuit configuration is used as the first high-pass filter for removing the DC offset. Thus, even if the DC offset level changes due to switching of the gain of the low noise amplifier during the preamble reception, it is possible to sufficiently follow.

Also, increasing the cut-off frequency f _c of the order of DC offset removal, a problem that the low-noise amplifier may become although the tracking of the DC offset switches due to gain switching of degrading the demodulation characteristic will shaved to near-DC signal there are (see FIG. 17B), since the structure for performing the DC offset is removed by splitting the preamble portion used for frequency offset estimation, very large cut-off frequency f _c, the adverse effect on the demodulation characteristics of the subsequent data portion There is no worry to give.

  In addition, when the DC offset changes abruptly due to the gain switching of the low noise amplifier, the DC offset may pass through the differential filter. Therefore, when the differential filter as the first high-pass filter detects a sudden change in the DC offset, the differential signal is input to the subsequent frequency offset estimation unit. The frequency offset estimator can maintain high estimation accuracy by excluding the sample output from the differential filter as a signal including a DC offset from the target of frequency offset estimation when this detection signal is input. it can.

  It is assumed that the OFDM signal input to the wireless communication apparatus according to the present invention does not use a DC subcarrier. The frequency offset estimation unit estimates a frequency offset using a preamble in which two symbols of the same OFDM signal are transmitted.

  Specifically, one OFDM symbol consists of n subcarriers, and the i-th sample of the time waveform of the OFDM symbol transmitted twice is s (i), and the OFDM symbol transmitted the first time is OFDM symbol. Samples are {s (0), s (1),..., S (n−1)}, and samples of OFDM symbols transmitted for the second time are {s (n), s (n + 1),. −1)}, where Δf is the frequency offset and D is the DC offset, the first high-pass filter is represented by the following equation (4) with respect to the received baseband signal represented by the following equation (3). Processing is performed to output a sample signal d (i).

  Then, the frequency offset correction unit estimates the frequency offset Δf included in the received baseband signal r (i) by performing the processing expressed by the following equation (5) using the sample signal d (i). be able to.

  In IEEE 802.11a / g, which is an example of a wireless communication system to which the present invention is applied, a short preamble or a long preamble in which the same symbol is repeatedly transmitted is included at the head of the packet. The frequency offset can be estimated more accurately from the signal after the removal.

  Moreover, although the sample signal d (i) output from the first high-pass filter configured by the differential filter is shown in the above formula (4), more precisely, as shown in the following formula (6), The DC offset change D (i + 1) -D (i) between the i-th and i + 1-th samples is added. For this reason, there is no problem if the change in the DC offset is small, but if the DC offset varies greatly between the sample signals due to gain switching of the low noise amplifier or the like, the DC offset affects the estimation of the frequency offset.

  Therefore, the first high-pass filter is configured such that when the absolute value of the sample output d (i) increases due to the DC offset change D (i + 1) −D (i) between the i-th and i + 1-th samples, The detection signal is output to the subsequent frequency offset estimation unit. Then, the frequency offset unit can maintain high estimation accuracy by excluding the estimation of the frequency offset using the above equation (5) in the i-th sample in response to the input of the detection signal.

  As described above, the frequency offset estimation unit estimates the frequency offset from the signal from which the DC offset is removed. On the other hand, the frequency offset correction unit removes the frequency offset estimated by the frequency offset estimation unit from the received baseband signal used for estimation of the frequency offset (that is, the DC offset is not removed). In other words, the DC offset is removed only for the preamble portion where the frequency offset is estimated, but the DC offset is not removed although the frequency offset is removed for the preamble portion and the payload portion after the frequency offset is used for estimation. In such a case, even if the frequency offset estimation itself is highly accurate, there is a problem that the influence of the DC offset is propagated to the demodulation circuit at the subsequent stage.

  Therefore, it is preferable to provide means for removing the DC offset for the portion after the estimation of the frequency offset of the received baseband signal is completed.

  For example, when the frequency conversion unit adopts the direct conversion method, the frequency of the local frequency oscillated in the local oscillator is reversely rotated based on the frequency offset estimated in the frequency offset estimation unit to The influence of the DC offset and the frequency offset can be simultaneously removed from the received baseband signal including the DC offset after the offset estimation is completed.

  Alternatively, a DC offset estimation unit that estimates a DC offset in the baseband signal after digital conversion by the analog-digital conversion unit may be further provided, and the estimated DC offset may be removed from the baseband signal after digital conversion. Good.

  The estimation of the DC offset is usually performed by averaging digitally converted baseband signals. For this reason, when the DC offset changes abruptly due to gain switching of the low noise amplifier or the like, the average value has no meaning, and the accuracy deteriorates if the DC offset estimation is continued. Therefore, the differential filter, when detecting a sudden change in the DC offset, inputs the detection signal to the DC offset estimation unit. Then, the DC offset estimation unit excludes estimation data estimated until the detection signal is input, and calculates again to prevent a decrease in estimation accuracy.

  Alternatively, the received baseband signal down-converted by the direct conversion method is filtered by the second high-pass filter to remove the DC offset generated by self-mixing of the local signal and then converted to a digital signal. You may do it. Since all signals pass through the second high-pass filter, it is preferable to set a relatively small cut-off frequency so as not to delete signals near the DC of the OFDM symbol.

  If the cut-off frequency is small, the response of the second high-pass filter becomes slow, so that the influence of the DC offset remains long, but after the DC offset is removed by the first high-pass filter having a sufficiently large cut-off frequency. Since the frequency offset is estimated, high estimation accuracy can be obtained. Further, filtering with a large cut-off frequency is performed only for the preamble portion where frequency offset estimation is performed, and the demodulation characteristics of the subsequent signals are not deteriorated.

  In general, not only frequency offset estimation but also signal processing such as packet detection and coarse timing detection has a high sensitivity to characteristic degradation with respect to DC offset.

  Therefore, the wireless communication device may further include a detection unit that detects packet detection and coarse timing using a signal from which the DC offset has been removed by the differential filter.

  The analog-digital conversion unit may further include a switch that exclusively connects the output terminal of the first high-pass filter or the DC offset correction unit. This switch connects the output terminal of the analog-to-digital converter with the DC offset from the path to the first high-pass filter at the timing at which the coarse timing detector detects the end of a predetermined preamble portion for performing frequency offset estimation. Operates to switch to the path to the corrector.

  Until the end of the predetermined preamble portion is detected, the output terminal of the analog-digital conversion unit is connected to the path to the first high-pass filter. The frequency offset is estimated with high accuracy using the reception baseband signal from which the DC offset has been removed using the first high-pass filter, using the period until the end of.

  On the other hand, the DC offset estimation unit performs DC offset estimation using a period until a predetermined preamble portion is completed. When the end of the predetermined preamble portion is detected, the output terminal of the analog-digital conversion unit is switched to the path to the DC offset correction unit. With respect to the received baseband signal after the completion of the predetermined preamble portion, the DC offset can be corrected using the DC offset estimated with high accuracy by the DC offset estimation unit over a long period of time. . Further, since the DC offset removal using the first high-pass filter is not performed on the received baseband signal after the completion of the predetermined preamble portion, it is not necessary to consider the degradation of the SNR characteristic.

  In addition, the frequency offset correction unit uses the frequency offset estimated by the frequency offset estimation unit until the predetermined preamble portion ends, and uses the frequency offset for the received baseband signal after the completion of the predetermined preamble portion. To correct.

  One application target of the present invention is a wireless communication system defined by IEEE802.11a. In IEEE 802.11a, a short preamble portion composed of a short training sequence having a relatively large subcarrier interval and a long preamble portion composed of a long training sequence having a relatively small subcarrier interval are added to the head of the packet.

  The radio communication apparatus according to the present invention uses the fact that the subcarrier interval is relatively large in the short preamble portion, and after removing the DC offset with a differential filter having a sufficiently large cutoff frequency, estimates the frequency offset. By doing this, high estimation accuracy is obtained. That is, since the frequency offset correction is performed using only the short preamble section, the output terminal of the analog-digital conversion section is connected to the switch section in accordance with the timing when the short preamble section ends and the long preamble section starts. Switch from the path to the first high-pass filter to the path to the DC offset correction unit.

  The frequency offset correcting unit removes the frequency offset estimated by the frequency offset estimating unit in the short preamble unit from the long preamble unit, and the DC offset correcting unit is estimated by the DC offset estimating unit in the short preamble unit. The DC offset is removed from the long preamble part.

  It is arbitrary how the long preamble portion transmitted after the short preamble portion is used on the receiving side, but generally, after the coarse frequency offset correction is performed using the short preamble portion, More precise frequency offset correction and channel estimation are performed using the long preamble part. Therefore, in order to realize more accurate channel estimation, it is necessary to appropriately estimate the frequency offset in the short preamble portion.

  For example, the wireless communication apparatus may further include a second frequency offset estimation unit and a second frequency offset correction unit that respectively perform frequency offset estimation and removal in the long preamble unit following the short preamble unit. The second frequency offset estimation unit receives the frequency offset estimated by the short preamble unit and the received baseband signal after the long preamble unit after the DC offset is removed, and estimates the frequency offset. . Then, the second frequency offset correction unit removes the frequency offset estimated by the second frequency offset estimation unit from the received baseband signal after the long preamble unit.

  Alternatively, the reception baseband signal after the long preamble portion after the frequency offset and the DC offset estimated by the short preamble portion are removed is fed back to the frequency offset estimation portion, and the reception after the long preamble portion is received. A frequency offset can be estimated from the baseband signal. The frequency offset correction unit can also remove the received baseband signal after the long preamble portion.

  In either case, the received baseband in which the frequency offset and the DC offset estimated in the short preamble portion are removed, and the residual frequency offset is corrected based on the frequency offset estimation in the long preamble portion and thereafter. By performing channel estimation from the signal, channel information can be acquired with higher accuracy.

  Furthermore, when IQ imbalance is included in the received baseband signal, there is a possibility that desired reception characteristics cannot be ensured even if frequency offset correction is performed. Assuming such a case, it is possible to adopt an apparatus configuration in which an IQ imbalance estimation unit and an IQ imbalance correction unit are provided. When this configuration is adopted, it is possible to remove IQ imbalance contained in the received baseband signal, so that it is possible to further improve the reception characteristics.

  On the other hand, in the direct conversion OFDM receiver, there is not only a DC offset, but also a problem of IQ imbalance caused by the phase difference of the local signal input to the mixer of each axis of IQ and the amplitude difference between the mixers. . IQ imbalance is a factor that degrades the estimation accuracy of the frequency offset together with the DC offset, and also affects the decoding characteristics.

  When the frequency offset correction unit estimates the frequency offset using the received baseband signal in which the DC offset can be removed but the IQ imbalance remains, the frequency offset information includes IQ unbalanced frequency in addition to the original frequency offset value Δf. The component generated due to the equilibrium, that is, the IQ unbalanced component is included.

  In a normal communication system, preamble symbols are transmitted from the transmitter a plurality of times, and on the receiver side, the frequency offset estimation unit can estimate the frequency offset for each preamble symbol. Here, the frequency offset can be expressed as a vector in a complex space, but the direction of the vector representing the IQ unbalanced component varies for each preamble symbol. Therefore, the frequency offset estimation unit can sequentially reduce the IQ unbalanced component included in the frequency offset estimated value by sequentially adding the frequency offset estimated for each preamble symbol, and finally, Can obtain a more accurate frequency offset.

  Here, the receiver generally includes a low noise amplifier that amplifies the received signal and a gain control unit that adjusts the gain of the low noise amplifier. When the gain of the low noise amplifier is switched during the period when the frequency offset estimation unit is estimating the frequency offset, the IQ imbalance component included in the estimated frequency offset increases when a large gain is set. Therefore, there is a concern that if the frequency offset estimated over a plurality of preamble symbols is simply added, the proportion of the IQ unbalanced component cannot be made sufficiently small.

  Therefore, the frequency offset estimation unit weights and adds the final frequency offset value to each frequency offset estimated for each preamble symbol according to the gain set in the low noise amplifier when the corresponding preamble symbol is received. As a result, it is possible to relatively reduce the IQ unbalanced component included in the frequency offset estimation value, and finally to obtain a more accurate frequency offset.

  In the receiver, usually, a large gain is determined in the low noise amplifier at the start of signal detection, and thereafter, the gain is switched to a low gain in accordance with the received signal power. Therefore, the frequency offset estimator gives a small weight to the frequency offset estimated in the first few preamble symbols in which a large gain is set in the low noise amplifier, while the preambles after switching to a small gain. A larger weight is given to the frequency offset estimated in the symbols, and these are added to obtain a final frequency offset. The assignment of the weight to the frequency offset here specifically corresponds to multiplying the output of the frequency offset estimation vector (described later) or the differential filter by a weighting coefficient.

  Specifically, the frequency offset estimation unit calculates a weighting factor based on the absolute value of the frequency offset estimated for each preamble symbol, weights and adds the frequency offset for each preamble symbol with each weighting factor, and finally To obtain a typical frequency offset value.

  For example, the frequency offset estimator gives a weighting factor of 0 to a frequency offset whose absolute value exceeds a predetermined threshold, and gives a weighting factor of 1 to a frequency offset whose pair value does not exceed a predetermined threshold, The final frequency offset value is obtained by weighting and adding the frequency offset for each symbol with each weighting factor. This corresponds to ignoring the frequency offset estimated in the preamble section in which a large gain is set in the low noise amplifier. The predetermined threshold mentioned here can be determined based on, for example, received signal strength.

  Alternatively, the frequency offset estimator gives a weighting factor consisting of the reciprocal of the absolute value to each frequency offset, and weights and adds the frequency offset for each preamble symbol with each weighting factor to obtain a final frequency offset value. It may be.

  According to the present invention, an excellent radio that employs a direct conversion method and can perform OFDM demodulation with higher characteristics by suitably removing the frequency offset in a situation where the DC offset fluctuates with time. A communication device can be provided.

  Also, according to the present invention, when a DC offset, IQ imbalance, and frequency offset that change with time are simultaneously present in the received OFDM symbol, the DC offset can be removed and the frequency offset can be accurately estimated. An excellent wireless communication device can be provided.

  The radio communication apparatus according to the present invention receives an OFDM signal by the direct conversion method, but even if the OFDM signal includes a DC offset, the radio communication apparatus can perform high speed by removing the DC offset using a differential filter. In addition, accurate frequency offset estimation can be performed. In addition, when the DC offset changes suddenly due to switching of the gain of the low noise amplifier, etc., the change is detected from the output of the differential filter and excluded from the estimation of the frequency offset, thereby improving the estimation accuracy of the frequency offset. Can do.

  In addition, according to the present invention, when vector signals including frequency offset information are integrated, the weighting coefficient corresponding to the power of the corresponding preamble signal, that is, the gain of the low noise amplifier, is multiplied to thereby estimate the frequency offset. The influence can be suppressed by IQ imbalance and time variation of DC offset, and more accurate frequency offset can be estimated by simple signal processing.

  Other objects, features, and advantages of the present invention will become apparent from more detailed description based on embodiments of the present invention described later and the accompanying drawings.

  Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.

  The present invention relates to a wireless communication apparatus that receives an OFDM signal by a direct conversion method. According to the direct conversion method, since the IF filter is not used, it is easy to increase the bandwidth of the receiver, and the flexibility of the configuration of the receiver is increased.

  In the OFDM communication system, there is a problem of frequency offset due to a subtle error in the frequency of the oscillator mounted on each transceiver, which is observed as a phenomenon of phase rotation of the received signal in the digital part on the receiver side. The For this reason, it is common practice to observe the frequency offset using a known training sequence added to the beginning of the packet and correct it.

  However, the direct conversion type receiver has a problem that a direct current component, that is, a DC offset, is generated in the downconverter output due to self-mixing of the local signal. The accuracy of frequency offset estimation and timing detection is easily affected by the DC offset, and accurate estimation cannot be performed even if the frequency offset is estimated while the DC offset exists.

  On the other hand, in the wireless communication apparatus according to the present invention, fast and accurate frequency offset estimation is performed by removing the DC offset using a differential filter.

(1) First Embodiment FIG. 1 shows the configuration of a receiver system of a wireless communication apparatus according to the first embodiment of the present invention. The illustrated apparatus comprises a direct conversion receiver configuration for an OFDM signal and includes a module that compensates for frequency offset.

  Of the OFDM signal received at the antenna, only a desired frequency band passes through the bandpass filter (BPF) 1 and is amplified by the low noise amplifier (LNA) 2. Here, a frequency offset resulting from the frequency error of the local oscillator between the transceivers is added to the received RF signal.

In order to keep the power of the received signal at an appropriate constant level, the gain of the low noise amplifier 2 is adjusted by an automatic gain control circuit (AGC). For example, in IEEE802.11a / g, gain adjustment is required in a range of 50 dB or more. In general, a large gain is set in the low-noise amplifier 2 at the start of signal detection. For example, in the vicinity of the center of the short preamble section (in this embodiment, the last end of t ₄ ), the gain is adapted to the received signal power. Switch to a lower gain. The gain switching level is about 20 dB. Since the AGC mechanism itself is well known, description thereof is omitted here.

The amplified received signal is multiplied by the local frequency f _LO generated by the local oscillator in the mixer 3 and is frequency-converted into a baseband signal by the direct conversion method. The baseband signal is digitally converted by an AD converter (ADC) 4.

  In the direct conversion of the reception system, the reception frequency and the local frequency are equal, so that a direct current component, that is, a DC offset is generated in the downconverter output due to self-mixing of the local signal. When the gain of the low noise amplifier is switched by automatic gain control, the DC offset fluctuates with time accordingly (see, for example, Non-Patent Document 9). Therefore, a DC offset that changes with time is added to the received baseband signal at this time.

  A predetermined section of the preamble signal of the digital baseband signal is branched and input to the differential filter 5, the DC offset component is removed and then input to the frequency offset estimator 6, and the estimated DC offset is subtracted. A more accurate frequency offset is estimated from the signal from which the DC offset has been removed. FIG. 21 shows a specific configuration example of the differential filter 5 and the frequency offset estimation unit 6. The differential filter 5 includes a delay unit 201 and an adder 202. The frequency offset estimation unit 6 includes a delay unit 203, a complex conjugate calculation circuit 204, a multiplier 205, an adder 206, a storage element 207, and a phase detection circuit 208.

The differential filter 5 is a kind of high-pass filter, but has a simple circuit configuration and good response. If the low gain of the low noise amplifier 2 is switched to a low gain in accordance with the received signal power at the position of the end of the short preamble t ₄ (described above), the DC offset level changes, but the differential filter 5 is sufficient for this. Following this, it is possible to suppress the passage of high frequency components.

Also, increasing the cutoff frequency f _c for the DC offset removed, the problem that the low-noise amplifier may become although the tracking of the DC offset switches due to gain switching of degrading the demodulation characteristic will shaved to near-DC signal (See FIG. 17B). On the other hand, the illustrated receiver has a configuration in which a preamble portion (STS having a relatively large subcarrier interval) used for frequency offset estimation is branched to remove DC offset. In other words, the DC offset is removed only for the preamble portion for which the frequency offset is estimated, but the DC offset is not removed for the preamble portion and the payload portion after the frequency offset is used for estimation, although the frequency offset is removed. . Thus, very large cut-off frequency f _c, there is no fear that adversely affect the demodulation characteristics of the data portion of the subsequent (subsequent LTS subcarrier interval is shortened).

  Further, when the DC offset changes abruptly due to gain switching of the low noise amplifier 2, the DC offset may pass through the differential filter 5. If such a waveform on an impulse is input to the frequency offset estimation unit 6, there is a risk of increasing the mean square error (MSE). Therefore, the differential filter 5 inputs a detection signal to the frequency offset estimation unit 6 when detecting a sudden change in the DC offset. When the detection signal is input, the frequency offset estimation unit 6 can maintain high estimation accuracy by excluding the sample output from the differential filter 5 from the target of frequency offset estimation as a signal including a DC offset. it can.

  Note that the threshold value for detecting a sudden change in the DC offset in the differential filter 5 can be calculated based on, for example, the number of gain switching and the received signal level.

  The estimated value of the frequency offset is input to the frequency offset correction unit 7, and the frequency offset of the baseband signal of the OFDM symbol portion after the frequency offset is used for estimation is compensated.

  The output of the frequency offset unit 7 is input to a discrete Fourier transform unit (DCT) 7 to demodulate subcarrier signals arranged on the frequency axis.

  The OFDM signal input to the receiver according to the present invention does not use DC subcarriers (DC corresponds to the 0 Hz position of the baseband signal in the OFDM modulation scheme). In addition, the frequency offset estimation unit 6 estimates a frequency offset using a packet preamble (STS having a relatively large subcarrier interval), and there is a preamble for transmitting two symbols of the same OFDM signal.

  An example of a wireless communication system to which the present invention is applied is a wireless LAN system conforming to IEEE 802.11a / g. FIG. 2 shows a subcarrier configuration of the OFDM symbol of the wireless LAN system. As shown in the figure, one OFDM symbol consists of 64 subcarriers, of which 52 subcarriers are modulated into information signals, and 4 subcarriers are used as pilot signals. Further, it is assumed that no signal is transmitted to the remaining subcarriers including a DC component (a null signal).

  On the other hand, a portion of the received baseband signal that has been used for frequency offset estimation (after LTS) and a portion that is affected by the frequency offset is input to the frequency offset correction unit 7, and is then converted into a DC by the high-pass filter. The frequency offset can be accurately compensated and demodulated without being affected by the deterioration of the demodulation characteristics associated with the removal of the offset.

  FIG. 15 shows a preamble configuration defined by IEEE 802.11a / g. Of these, in the long preamble, an OFDM symbol consisting of the same long training sequence (LTS) of 3.2 microseconds is transmitted twice in succession. The i-th sample of the time waveform of this OFDM symbol is s (i). However, {s (0), s (1),..., S (63)} corresponds to the first OFDM symbol, and {s (64), s (65),. Corresponding to the second OFDM symbol (when the order of the discrete Fourier transform is N, the first OFDM symbol is {s (0), s (1),..., S (N / 4-1)}). Yes, the second OFDM symbol is {s (N / 4), s (N / 4 + 1),..., S (2N / 4-1)}).

  When the frequency offset at this time is Δf and the DC offset is D, the received baseband signal for the i-th short preamble is expressed by the following equation.

  The differential filter 5 includes a delay unit 201 and an adder 202. An AD conversion signal from the AD converter 4 is input to the input of the delay unit 201. In addition, the AD conversion signal is input to the first input of the adder 202, and the output of the delay device 201 is inverted to the second input to subtract both signals. Therefore, the differential filter 5 performs processing such as the following equation (8) on the received baseband signal (the equation is an output signal of the differential filter 5 for the i-th short preamble).

The subsequent frequency offset estimation unit 6 includes a delay unit 203, a complex conjugate calculation circuit 204, a multiplier 205, an adder 206, a storage element 207, and a phase detection circuit 208. The output of the adder 202 is input to the first inputs of the delay unit 203 and the multiplier 205. The delay unit 203 delays the input signal by N / 4 (= 16) samples corresponding to the short preamble length, and outputs the delayed input signal to the complex conjugate calculation circuit 204 at the subsequent stage. The output of the complex conjugate calculation circuit 204 is input to the second input of the multiplier 205. Therefore, the multiplier 206 performs the cross-correlation process shown in the following equation for each short preamble t ₁ , t ₂ .

  The output of the multiplier 205 is connected to the first input of the adder 206, and the output of the storage element 207 is connected to the second input thereof. The output of the adder 206 is input to the storage element 207 and the phase detection circuit 208. As a result, the cross-correlation results of the above equation (9) over all the short preambles are added using the adder 206, and the frequency offset Δf is estimated.

  In the frequency offset correction unit 7, the frequency offset correction unit 7 compensates the frequency offset of the received baseband signal that has been affected by the frequency offset subsequent to the frequency offset estimation. Specifically, the frequency offset is corrected by reversely rotating the data phase in response to the frequency shift. Note that even if a short training sequence (STS) is used, the frequency offset can be estimated in the same manner as described above (however, the number of samples in this case is 16).

  FIG. 3 shows the square error of the estimated frequency offset value when the DC offset power versus the OFDM signal power is 30 dB and the frequency offset value normalized by the subcarrier interval. From the figure, it can be seen that the frequency offset can be accurately estimated by removing the DC offset from the received preamble signal used for frequency offset estimation using the differential filter 5. Since the same training sequence is repeated in the STS (see FIG. 15), the frequency offset can be estimated using a similar process.

  Here, in the above equation (5), it is assumed that the DC offset is constant, that is, does not vary with time, and the DC offset can be canceled by the differential filter 5. However, when the gain of the low noise amplifier 2 is switched, a change in the DC offset level appears as a high frequency component, so the output of the differential filter 5 represents the magnitude of the DC offset.

Specifically, the automatic gain control circuit sets a large gain for the low noise amplifier 2 at the start of signal detection, and uses the first to fourth short preambles t _{1 to} t ₄ (however, (Due to multipath effects, t ₁ and t ₂ are not used.) An appropriate gain is determined so that the power of the received signal becomes a constant level. Then, the gain is switched to a low gain at the head of the _fifth short preamble t ₅ . The gain switching level is about 20 dB, and the DC offset level fluctuates accordingly. Therefore, the output of the differential filter 5 is also affected at the beginning of t ₅ (see FIG. 22).

  Here, if the DC offset value in the i-th sample is set to D (i), the output of the differential filter 5 is output when the DC offset changes in the OFDM sample due to gain switching in the low noise amplifier 2 or the like. Is as shown in the following equation (10).

  From the above equation, when the DC offset changes suddenly, the difference {D (i + 1) −D (i)} between the DC offsets of the (i + 1) th and ith samples remains, and the absolute value of the output d (i) of the differential filter 5 remains. The value increases. As a result, when a change in the DC offset is detected, the DC offset is not removed but passes through.

  Therefore, in the receiver shown in FIG. 1, when the differential filter 5 detects that the absolute value of the output d (i) exceeds a predetermined value, it gives an instruction (detection signal) to the subsequent frequency offset estimation unit 6. Thus, the output d (i) of the i-th sample including the influence of the DC offset is excluded from the frequency offset estimation. Accordingly, the frequency offset estimation unit 6 can improve the estimation accuracy by estimating the frequency offset by excluding the DC offset. Moreover, even when automatic gain control of the low noise amplifier 2 is frequently performed, high estimation accuracy can be maintained by excluding the component through which the DC offset has passed.

  As described above, according to the receiver configuration shown in FIG. 1, since the frequency offset is estimated after the DC offset is removed by the differential filter 6, the frequency offset can be accurately determined while supporting both the DC offset and the frequency offset. Estimation can be performed. Moreover, since the frequency offset estimated value in the section where the gain of the low noise amplifier 2 is switched is excluded, high estimation accuracy can be maintained.

  In the receiver configuration shown in FIG. 1, the frequency offset correction unit 7 is estimated by the frequency offset estimation unit 6 from the received baseband signal after being used for estimation of the frequency offset (that is, the DC offset is not removed). Remove the frequency offset. In other words, the DC offset is removed only for the preamble portion where the frequency offset is estimated, but the DC offset is not removed although the frequency offset is removed for the preamble portion and the payload portion after the frequency offset is used for estimation. In such a case, even if the frequency offset estimation itself is highly accurate, there is a problem that the influence of the DC offset is propagated to the demodulation circuit at the subsequent stage.

  FIG. 4 shows an example of a receiver circuit that solves such a problem. In the receiver configuration shown in FIG. 1, the frequency offset estimated by the frequency offset estimator 6 after being input to the differential filter 5 and removed from the DC offset is used, and the frequency offset corrector 7 uses the frequency offset for estimation. Thereafter, frequency offset compensation is performed on the received baseband signal including the DC offset. On the other hand, in the receiver configuration shown in FIG. 4, the phase of the local frequency oscillated in the local oscillator 11 is reversely rotated based on the frequency offset estimated by the frequency offset estimation unit 6 after the DC offset is removed by the differential filter 5. I am letting. As a result, the influence of the DC offset and the frequency offset can be simultaneously removed from the received baseband signal including the DC offset after the frequency offset is used for estimation.

  Also in this case, the differential filter 5 inputs a detection signal to the frequency offset estimation unit 6 when detecting a sudden change in the DC offset. The frequency offset estimation unit 6 excludes the sample output from the differential filter 5 at the time when this detection signal is input so as not to be affected by the DC offset that has passed through the differential filter 5 (same as above).

  FIG. 5 shows another receiver configuration example that can remove the influence of the DC offset from the received baseband signal after the frequency offset is used for estimation (after the LTS).

  In the illustrated receiver, a predetermined section of the preamble signal of the received baseband signal is branched and input to the differential filter 5 to remove the DC offset, and the frequency offset correction unit 7 estimates the frequency offset. The differential filter 5 inputs a detection signal to the frequency offset estimator 6 when detecting a sudden change in the DC offset. The frequency offset estimation unit 6 excludes the sample output from the differential filter 5 at the time when this detection signal is input so as not to be affected by the DC offset that has passed through the differential filter 5 (same as above).

  In parallel with the frequency offset estimation process, the DC offset estimation unit 9 estimates a DC offset for the received baseband signal, and the DC offset correction unit 10 removes the DC offset from the received baseband signal. After that, the frequency offset correction unit 7 converts the received baseband signal from which the DC offset after the frequency offset has been used for estimation in this way to a highly accurate frequency offset value that has been estimated after the DC offset. Based on this, frequency offset compensation is performed.

  Here, DC offset estimation is normally performed by averaging digitally converted baseband signals. For this reason, when the DC offset changes abruptly due to gain switching of the low noise amplifier 2 or the like, the average value has no meaning and the accuracy deteriorates if the DC offset estimation is continued. Therefore, when the differential filter 5 detects a sudden change in the DC offset, the differential filter 5 inputs the detection signal not only to the frequency offset estimation unit 6 but also to the DC offset estimation unit 9. Then, the DC offset estimation unit 9 excludes the estimation data estimated until the detection signal is input, and recalculates it to prevent a decrease in estimation accuracy.

  FIG. 6 shows still another configuration example of the receiver that can remove the influence of the DC offset from the received baseband signal after the frequency offset is used for estimation (after the LTS).

In the illustrated receiver, the mixer 3 multiplies the local frequency f _LO generated by the local oscillator 11 and down-converts by the direct conversion method. The received baseband signal is added with a DC offset generated by self-mixing of the local signal (see FIG. 32A), but is passed through the high-pass filter (HPF) 12 to obtain the DC offset. Is removed. However, since all signals pass through, the high-pass filter 12 sets a relatively small cut-off frequency so as not to delete signals near the DC of the OFDM symbol. The baseband signal that has passed through the high-pass filter 12 is digitally converted by an AD converter (ADC) 4.

If you have set smaller cut-off frequency f _c of the high-pass filter 12, while can keep the demodulation characteristics in the subsequent stage, the response to changes in DC offset slower. For this reason, if the DC offset changes due to gain switching of the low noise amplifier 2 or the like, the effect remains long and the DC offset continues to pass (see FIG. 32B). Therefore, a predetermined section of the preamble signal in the received baseband signal is branched into two branches and input to the differential filter 5 in which a higher cutoff frequency is set in one branch to remove the residual DC offset. . As shown in FIG. 32 (c), the differential filter 5 blocks the residual DC offset, a timing when the gain switching took place at the head position of the t _5, and outputs only the waveform on steep impulse.

  Subsequently, the frequency offset estimation unit 6 estimates the frequency offset based on the autocorrelation value of the output of the differential filter 5. Then, the frequency offset correction unit 7 removes the frequency offset from the received baseband signal.

  The differential filter 5 inputs this detection signal to the frequency offset estimation unit 6 when detecting a sudden change in the DC offset. When the waveform on the impulse is input to the frequency offset estimation unit 6, there is a risk of increasing the MSE. Therefore, the frequency offset estimation unit 6 excludes the sample output from the differential filter 5 at the time when this detection signal is input so as not to be affected by the DC offset that has passed through the differential filter 5 (same as above).

  The threshold for detecting a sudden change in DC offset in the differential filter 5 can be calculated based on the number of gain switching and the received signal level (same as above). For example, a received signal strength indicator (RSSI) circuit can be provided after the high-pass filter 12 to detect the received signal level.

  On the other hand, for the received baseband signal after the frequency offset is used for estimation, the DC offset estimating unit 9 estimates the DC offset, and the DC offset correcting unit 10 removes the DC offset from the received baseband signal.

  Here, the differential filter 5 inputs a detection signal to the DC offset estimation unit 9 when detecting a sudden change in the DC offset. Then, the DC offset estimation unit 9 excludes the estimation data estimated until the detection signal is input, and calculates again to prevent a decrease in estimation accuracy (same as above).

  After that, the frequency offset correction unit 7 removes the DC offset from the received baseband signal from which the DC offset used for estimating the frequency offset in this way has been removed. Frequency offset compensation is performed based on the offset value.

  FIG. 7 shows still another receiver configuration example that can remove the influence of the DC offset from the received baseband signal after the frequency offset is used for estimation (after the LTS).

In the receiver shown in the figure, the mixer 3 multiplies the local frequency f _LO generated by the local oscillator 11 and down-converts by the direct conversion method. Then, it passes through the high-pass filter 12 at the subsequent stage, and removes the DC offset generated by the self-mixing of the local signal from the received baseband signal. High-pass filter 12, since all the signals contained in the packet passes sets a relatively small cutoff frequency f _c so as not cut a near-DC signal in the OFDM symbol. The received baseband signal is digitally converted by an AD converter (ADC) 4.

If you have set smaller cut-off frequency f _c of the high-pass filter 12, while can keep the demodulation characteristics in the subsequent stage, the response to changes in DC offset slower. For this reason, if the DC offset changes due to gain switching of the low noise amplifier 2 or the like, the effect remains long and the DC offset continues to pass (see FIG. 32B). Therefore, a predetermined section of the preamble signal in the received baseband signal is branched into two branches and input to the differential filter 5 in which a higher cutoff frequency is set in one branch to remove the residual DC offset. . As shown in FIG. 32 (c), the differential filter 5 blocks the residual DC offset, a timing when the gain switching took place at the head position of the t _5, and outputs only the waveform on steep impulse.

  Then, based on the frequency offset estimated in this way, the phase of the local frequency oscillated in the local oscillator 11 is reversely rotated. As a result, the influence of the DC offset and the frequency offset can be simultaneously removed from the received baseband signal including the DC offset after the frequency offset is used for estimation.

  The differential filter 5 inputs a detection signal to the frequency offset estimator 6 when detecting a sudden change in the DC offset. When the waveform on the impulse is input to the frequency offset estimation unit 6, there is a risk of increasing the MSE. Therefore, the frequency offset estimation unit 6 excludes the sample output from the differential filter 5 at the time when this detection signal is input so as not to be affected by the DC offset that has passed through the differential filter 5 (same as above).

  The threshold for detecting a sudden change in DC offset in the differential filter 5 can be calculated based on the number of gain switching and the received signal level. For example, an RSSI circuit can be provided after the high-pass filter 12 to detect the received signal level (same as above).

  In the present invention, the frequency offset is estimated using a preamble in which the same OFDM signal is repeatedly transmitted for two symbols. In the preamble configuration shown in FIG. 15, it is necessary to perform signal processing for packet detection, coarse timing detection, frequency offset estimation, and correction using STS. These signal processes are generally sensitive to characteristic deterioration with respect to DC offset.

  In FIG. 6, the frequency offset estimation is performed with high accuracy by removing the influence of the DC offset by using the output signal of the differential filter 5 which is a kind of high-pass filter. Similarly, with respect to packet detection and coarse timing detection, it is considered that deterioration of characteristics with respect to DC offset can be suppressed by using an output signal of a high-pass filter.

  FIG. 8 shows a configuration example around the synchronization circuit that performs frequency offset, packet detection, and coarse timing detection using the output of the high-pass filter.

  The illustrated synchronization circuit is provided with a path to the high-pass filter 21 and a path of the DC offset estimation unit 24 for each of the I-axis and Q-axis input signals, and the two switches 26 and 27 are exclusively turned on / off. By controlling, the path is switched.

  In STS, the subcarrier closest to DC is 1.25 MHz, and the subcarrier spacing is relatively wide. By paying attention to this, the cutoff frequency of the high-pass filter 21 is increased to ensure the response characteristics with respect to the fluctuation of the DC offset. Thereby, it is possible to suppress the influence of the DC offset while minimizing the loss of the SNR characteristic.

During the synchronization process, the DC offset estimation unit 25 estimates the DC offset from the received baseband signal. Since the STS has a periodicity of 0.8 microseconds, the DC offset can be estimated using a moving average or the like. If taking the first four symbols t ₁ ~t ₄ of STS in an automatic gain control and DC offset, it is possible to obtain up to 6 symbols t ₅ ~t ₁₀ i.e. the estimated time of 4.8 microseconds in synchronous circuit Therefore, highly accurate DC offset estimation can be performed.

  Until the STS is completed, the IQ input terminal is connected to the path of the high-pass filter 21. The frequency offset estimation unit 22, the packet detection and coarse timing detection unit 23 receive the received baseband from which the DC offset is removed. From the signal, the frequency offset is estimated with high accuracy, and the packet and coarse timing are detected.

  The switches 26 and 27 are turned on / off at the timing when the STS ends, and the IQ input terminal is switched from the high-pass filter 21 path to the DC offset correction path. The output signal of the coarse timing detection unit 23 is used as the timing when the STS ends.

  Since the DC offset estimation unit 25 performs DC offset estimation using a sufficiently long time until the STS is completed, it is possible to realize highly accurate DC offset correction. Further, after the timing when the STS ends, there is no path to the high-pass filter 21. That is, DC offset removal by the high-pass filter 21 using a large cut-off frequency is not performed for received baseband signals after the LTS in which the subcarrier interval is shortened, so there is no need to consider degradation of SNR characteristics. .

  Further, the frequency offset correction unit 24 corrects the frequency offset for the received baseband signal after the STS is finished, using the frequency offset estimated by the frequency offset estimation unit 22 until the STS is finished. .

  In the synchronous circuit configuration shown in FIG. 8, the coarse timing detection signal is used as it is for switching the path to the high-pass filter 21 and the path to the frequency offset correction unit 24. However, in consideration of the processing delay of the digital circuit, there may be a case where it is better to switch at an earlier stage than the end of the STS.

  Therefore, in the synchronous circuit configuration shown in FIG. 9, the switches 26 and 27 are not directly switched by the detection signal of the coarse timing detector 23, but a switch controller 28 is provided separately. FIG. 10 shows an example of a method for adjusting the switch switching timing by the switch control unit 28.

  The switch control unit 28 calculates the switch switching timing using the moving average value of the correlation value of the input signal. Then, a threshold value is set in advance, and the timing at which the moving average value exceeds the threshold value is output as a control signal to the switches 26 and 27. As shown in FIG. 10, by setting the threshold value, the switch switching timing can be flexibly adjusted in the section from the packet detection to the coarse timing.

  FIG. 11 shows a configuration example of the high pass filter 21. FIG. 11A is a configuration example using a differential filter, and FIG. 11B is a configuration example using DC offset estimation and correction by moving average.

  FIG. 12 shows a configuration example of the DC offset estimation unit 25. Using the output signal of the switch control unit 28 or the coarse timing detection unit 23, the DC offset estimation value at the switching timing of the switches 26 and 27 can be held thereafter. Therefore, the DC offset correction unit 25 can perform DC offset correction on the received baseband signal after the completion of the STS by using a highly accurate DC offset estimation value at the time when the STS is completed.

  Each of the synchronization circuits shown in FIG. 8 and FIG. 9 is configured to perform frequency offset correction in consideration of DC offset using only the STS portion of the preamble configuration defined by IEEE 802.11a. The method of using the LTS after the process is completed is arbitrary.

  In general, coarse frequency offset correction is performed using STS, and then finer frequency offset correction and channel estimation are performed using LTS. Therefore, in order to realize more accurate channel estimation, it is necessary to appropriately estimate the frequency offset in the STS.

FIG. 13 shows a configuration example around a synchronous circuit including a circuit module that performs frequency offset correction and channel estimation in the long preamble section. In the illustrated example, an LTS frequency offset estimation unit 31 and a frequency offset correction unit 32 are provided independently of the STS frequency offset estimation unit 22 and the frequency offset correction unit 24, and further, channel estimation is performed in the subsequent stage. A portion 33 is provided. On the other hand, the frequency offset estimator 22 performs rough frequency offset estimation using a short preamble (however, the _first and _second short preambles t ₁ and t ₂ may be affected by multipath. Therefore, estimation is performed using the _third and subsequent short preambles t _{3 to} t ₁₀ ). The other frequency offset estimation unit 31 performs more precise frequency offset estimation using the long preambles T ₁ and T ₂ .

  At the timing when the short preamble section ends, the IQ input terminal is switched from the high-pass filter 21 path to the DC offset correction path. After the LTS, the DC offset correction unit performs DC offset correction by subtracting an accurate DC offset estimated using a sufficiently long time until the STS is completed from each input signal of IQ. The subsequent frequency offset correction unit 24 corrects the frequency offset estimated by the frequency offset estimation unit 22 until the STS ends.

  When the frequency offset estimation unit 31 receives the received baseband signal after the LTS from which the DC offset and the frequency offset estimated using the STS have been removed, the frequency offset estimation unit 31 estimates the frequency offset more precisely. The frequency offset correction unit 32 removes the frequency offset estimated by the frequency offset estimation unit 31 from the received baseband signals after the LTS.

  And the channel estimation part 33 can perform a more accurate channel estimation using the received baseband signal from which the secondary (residual) frequency offset was removed using LTS.

  In the circuit configuration shown in FIG. 13, a circuit module for estimating and removing the frequency offset is separately provided for the received baseband signal after the LTS, but the frequency offset estimating unit 22 and the frequency offset correcting unit 24 receive the signal after the LTS. It can also be configured to perform estimation and removal of the frequency offset of the baseband signal, respectively. FIG. 14 shows a configuration example around the synchronization circuit in the latter case.

  At the timing when the STS ends, the IQ input terminal is switched from the high-pass filter 21 path to the DC offset correction path. After the LTS, the DC offset correction unit performs DC offset correction by subtracting an accurate DC offset estimated using a sufficiently long time until the STS is completed from each IQ input signal. Subsequently, the frequency offset correction unit 24 corrects the frequency offset estimated by the frequency offset estimation unit 22 until the STS ends.

  Further, at the timing when the STS ends, the switch 27 ′ is also turned on, and a path for feeding back the output terminal of the frequency offset correction unit 24 to the frequency offset estimation unit 22 is made.

  The frequency offset estimator 22 receives the DC offset estimated using the STS and the received baseband signal after the LTS from which the frequency offset has been removed, and further estimates the frequency offset. The frequency offset correction unit 24 removes the frequency offset estimated by the frequency offset estimation unit 31 from the received baseband signals after the LTS.

  And the channel estimation part 33 can perform a more accurate channel estimation using the received baseband signal from which the secondary (residual) frequency offset was removed using LTS.

  So far, in the direct conversion type OFDM receiver, when the DC offset fluctuates with the gain fluctuation of the low noise amplifier, the method of reducing the influence of the fluctuation of the DC offset on the frequency offset estimation has been described.

  On the other hand, in the direct conversion type OFDM receiver, not only the DC offset due to the self-mixing of the local signal but also the phase difference of the local signal input to the mixer of each IQ axis and the amplitude difference between the mixers. There is a problem of IQ imbalance. IQ imbalance is a factor that degrades the estimation accuracy of the frequency offset together with the DC offset, and also affects the decoding characteristics. Therefore, in the following, the frequency offset estimation method in the case where IQ imbalance and time-varying DC offset exist will be described in detail.

  In the receiver configuration shown in FIG. 1, when the level of the DC offset fluctuates with the gain switching of the low noise amplifier 4 and the output of the differential filter 5 has a fluctuation of the DC offset exceeding a predetermined value ( By excluding the symbol from the frequency offset estimation, the influence of DC offset variation on the frequency offset estimation is reduced (see FIG. 22). However, sufficient consideration is not given to the effects of IQ imbalance.

  In the direct conversion type OFDM receiver, there is a problem of IQ imbalance as well as DC offset due to self-mixing of local signals. In the direct conversion method, since there is no IF signal in the digital domain, IQ quadrature demodulation must be performed in the analog domain instead of the digital domain. For this reason, IQ imbalance arises by the unbalanced component of in phase (I) and quadrature phase (Quadrature: Q). In particular, the IQ imbalance related to the phase is caused by the fact that the phase difference between the local signals input to the mixers of the I and Q channels is not exactly 90 degrees. This is due to the channel signal gain difference.

  FIG. 23 illustrates the cause of IQ imbalance. In the figure, a local signal from one local oscillator is distributed to two branches, and the phase of one branch is changed by 90 ° to generate a cosine signal and a sine signal. Yes. If the phase relationship between the two signals is shifted from 90 ° or the amplitude is different, distortion occurs in the baseband signal after frequency conversion. Such a case is called IQ imbalance. The IQ imbalance deteriorates the estimation accuracy of the frequency offset because the influence of the distortion passes through the differential filter.

Here, if the phase difference of the cosine signal and the sine signal theta, when put between the amplitude difference lambda [dB], I and Q components of the local signal of the local frequency f _c becomes as follows, the input to each of the mixers Is done.

  However, α is expressed as follows by the amplitude difference λ.

These local signals are frequency-multiplied with the received signal r (i) by each mixer. However, the complex transmission symbols and _{X n = a n + jb n} .

  Therefore, the received baseband signal when IQ imbalance is not taken into consideration is expressed as in the above equation (7), but when affected by IQ imbalance, the complex received baseband signal is expressed by the following equation (11). ). Here, i is a sample number in the short preamble, and a superscript * in the formula represents a complex conjugate.

  Therefore, the differential signal output from the differential filter 5 when affected by IQ imbalance is expressed by the following equation (12).

  Then, the frequency offset estimation unit 6 performs frequency offset based on the difference signal of the above equation (12). Multiplier 205 multiplies the difference signal delayed by N / 4 samples to obtain a frequency offset estimation vector for each sample. The autocorrelation value (that is, frequency offset estimation vector) of the differential filter 5 output including IQ imbalance is expressed by the following equation (13).

  Here, the frequency offset information represented by the above equation (13) is composed of four terms. When each term is expressed as a vector on the complex space, it is as shown in FIG.

  The first term of the above equation (13) is a vector that depends only on the frequency offset. That is, when IQ imbalance is not included in the received baseband signal, only the first term is output from the multiplier 205, and the frequency offset can be estimated from the angle of this vector.

In addition, the second to fourth terms of the above equation (13) are terms generated by IQ imbalance, and these terms deteriorate the estimation accuracy of the frequency offset. Here, with respect to the fourth term, | ψ | ² is an extremely small value and is ignored (α is about 0.1, θ is about 0.05 °, and it is clear that ψ is small). In addition, the second and third terms have a complex conjugate relationship, and when these two terms are added, only the real component remains, which is considered to be the main factor that degrades the frequency offset estimation accuracy.

  The multiplier 205 adds the cross-correlation results for each short preamble over all the short preambles using the adder 206 to estimate the frequency offset Δf. The distortion included in the sum of the cross-correlation results depends on the number of samples to be summed and the pattern of the preamble signal.

  FIG. 21 shows a configuration example of the differential filter 5 and the frequency offset estimator 6, which considers the DC offset that varies with time, but fully considers the influence of IQ imbalance when estimating the frequency offset. It was n’t. Therefore, FIG. 25 is proposed as a configuration example of the differential filter 5 and the frequency offset estimation unit 6 that suppress the influence of IQ imbalance. In the figure, the differential filter 5 includes a delayer 301 and an adder 302. The frequency offset estimation unit 6 includes a delay unit 303, a complex conjugate calculation circuit 304, a multiplier 305, a multiplier 306, a coefficient calculation circuit 307, an adder 309, a storage element 310, and a phase detection circuit. 311.

  Hereinafter, an operation for estimating the frequency offset more accurately by suppressing the influence due to the IQ imbalance and the time variation of the DC offset at the time of frequency offset estimation will be described with reference to FIG.

  A signal received by the antenna passes through the band-pass filter 1 and the low noise amplifier 2, and only a desired OFDM signal is amplified. The amplified signal is multiplied by the local signal from the local oscillator 11 by the mixer 3 and converted into a baseband signal. The received baseband signal is converted into a digital signal by the AD converter 4. Here, if the amplitude difference due to IQ imbalance is α and the phase difference is θ, the received baseband signal is expressed as in the above equation (11).

  The differential filter 5 includes a delay unit 301 and an adder 302. An AD conversion signal from the AD converter 4 is input to the delay unit 301. In addition, the AD conversion signal is input to the first input of the adder 302, and the output of the delay device 301 is inverted and input to the second input to subtract both signals. Therefore, the differential filter 5 performs processing such as the following equation (12) on this received baseband signal.

The subsequent frequency offset estimation unit 6 includes a delay unit 303, a complex conjugate calculation circuit 304, a multiplier 305, a multiplier 306, a coefficient calculation circuit 307, an adder 309, a storage element 310, and a phase detection circuit 311. Consists of. The output of the adder 302 is input to the first inputs of the delay unit 303 and the multiplier 305. The delay unit 303 delays the input signal by the short preamble length N / 4 (= 16) samples and outputs it to the complex conjugate calculation circuit 304 at the subsequent stage. The output of the complex conjugate calculation circuit 304 is input to the second input of the multiplier 305. Therefore, the frequency offset estimation unit 6 performs the processing shown in the above equation (13) for each short preamble t ₁ , t ₂ .

  The first term of the above equation (13) is a vector including the frequency offset Δf. By adding the vectors by the adder 309 and the storage element 310 and detecting the rotation angle by the phase detector 311, the value of the frequency offset value Δf can be estimated.

  As already described, the second to fourth terms of the above equation (13) are terms generated by IQ imbalance, and these terms deteriorate the estimation accuracy of the frequency offset. Of these, the fourth term can be ignored because the absolute value of ψ is small (as described above). The second and third terms have a complex conjugate relationship, and when these two terms are added, only the real number component remains, and that component is the main factor that degrades the frequency offset estimation accuracy.

  Here, the second and third terms are vectors depending on the pattern of the short preamble r (i) and the frequency offset, and their directions are not constant. Therefore, if the estimated frequency offset value for each preamble symbol output from multiplier 305 is added over a sufficient number of samples using storage element 310 and adder 309, the first term of equation (13) increases. On the other hand, the distortion components of the second to fourth terms are relatively smaller than those of the first term. If detected by the phase detection circuit 311 at the subsequent stage, a more accurate frequency offset can be estimated.

  However, if the gain of the low noise amplifier 2 is switched while the frequency offset estimation unit 6 is estimating the frequency offset, the IQ imbalance included in the frequency offset estimated when a large gain is set. Since the component becomes large, there is a concern that if the frequency offset estimated over a plurality of preamble symbols is simply added, the proportion of the IQ unbalanced component cannot be sufficiently reduced.

In the receiver, normally, a large gain is determined in the low noise amplifier 2 at the start of signal detection, and thereafter, the gain is switched to a low gain in accordance with the received signal power. Specifically, the automatic gain control circuit at the time of signal detection start have configured the large gain for the low-noise amplifier 2, with a short preamble t ₁ ~t ₄ from the head to 4 th appropriate gain Is switched to a low gain at the beginning of the _fifth short preamble t ₅ . The gain switching level is about 20 dB.

Assuming that the first to second short preambles t _{1 to} t ₂ are not used for frequency offset estimation in consideration of the influence of multipath, before and after the gain switching of the low noise amplifier 2 is performed (multiplier 305 The frequency offset estimation vector in which the frequency offset is expressed in a complex space has a significantly different vector size as shown in FIG. 26, for example. That is, the absolute value of the output of the multiplier 305 corresponding to the short preamble symbols t ₃ and t ₄ is large, while the absolute value of the output of the multiplier 305 corresponding to the short preamble symbols t ₅ and t ₁₀ is Get smaller. The output of the multiplier 305 corresponding to t ₃ and t ₄ is only 15 samples. Therefore, the combined vector of the second term and the third term of the above equation (13) included in the output of the multiplier 305 corresponding to t ₃ and t ₄ remains a relatively large value compared to the first term. Remain at 310.

In the configuration shown in FIG. 21, the frequency offset is estimated by summing up the cross-correlation results of all the short preambles from t ₃ to t ₁₀ . The larger the number of samples having cross-correlation, the more the distortion depending on the second and third terms in the above equation (13) can be suppressed. However, when the gain of the low-noise amplifier 2 is switched, the amplitude of the sample relating to the short preamble symbols t ₃ and t ₄ is extremely larger than that of the other preamble symbols, and thus depends on the second and third terms. The distortion component remains in the estimated vector. As a result, frequency estimation based on a combined estimation vector (see FIG. 27) obtained by sequentially adding 79 outputs of the multiplier 205 corresponding to the short preamble symbols t ₅ and t ₁₀ to the output of the storage element 310. The accuracy of is degraded.

  On the other hand, in the configuration of the frequency offset estimation unit 6 shown in FIG. 25, the frequency offset estimated for each rumble symbol corresponds to the gain set in the low noise amplifier 2 when receiving the corresponding preamble symbol. The final frequency offset value is obtained by weighted addition. Therefore, the IQ included in the frequency offset estimated value is obtained by performing greater weighting on the samples after the gain switching of the low noise amplifier 2 (or not using the samples before the gain switching for the frequency offset estimation). The unbalanced component can be made relatively small, and finally a more accurate frequency offset can be obtained.

  Specifically, the absolute value of the output signal of the multiplier 305 is input to the coefficient calculation circuit 307 and the corresponding coefficient is input to the multiplier 306. The coefficient calculation circuit 307 calculates a weighting coefficient by one of the following methods, for example, according to the absolute value of the output signal from the multiplier 305. The calculation of the weighting factor according to the absolute value of the output signal of the multiplier 305 is almost equivalent to the calculation of the weighting factor according to the magnitude of the gain set for the low noise amplifier 2.

(Method 1)
If a threshold value is provided and the absolute value of the output signal of the multiplier 305 exceeds the threshold value, the coefficient is set to 0. If the threshold value is not exceeded, the coefficient is set to 1. This threshold value is calculated from, for example, RSSI (Received Signal Strength Indicator). In this case, the frequency offset is estimated from the following equation (14).

In other words, (Method 1) corresponds to estimating the frequency offset from the short preambles t _{5 to} t ₁₀ after the gain switching of the low noise amplifier 2 is performed. In this case, the number of samples that take the cross-correlation increases, and the distortion term that depends on the IQ imbalance can be reduced.

(Method 2)
The reciprocal of the absolute value of the output signal of the multiplier 305 is defined as a coefficient. In this case, the frequency offset is estimated from the following equation.

However, here, it is assumed that the gain switching of the low noise amplifier 2 is instantaneously performed between the short preambles t ₄ and t ₅ as shown in FIG. By such processing, it is possible to reduce the influence of distortion components that appear in the output of the adder 302 due to IQ imbalance.

FIG. 28 shows a combined estimation vector obtained by the frequency offset estimation unit 6 when the above (Method 1) is used. As shown in the figure, the estimated vector obtained by using the short preamble symbols t ₃ and t ₄ before the gain of the low noise amplifier 2 is reduced is canceled, which is different from the example shown in FIG. If the addition is performed over a sufficient number of samples, the first term of Equation (13) increases, while the distortion components of the second to fourth terms become relatively smaller than the first term.

  In addition, FIG. 29 compares the frequency offset estimation accuracy (mean square error versus normalized frequency offset value) when using the above (Method 1) with that using the configuration shown in FIG. As shown.

  As described above, in the receiver shown in FIG. 1, the differential filter 5 and the frequency offset estimation unit 6 are configured as shown in FIG. 25, so that it is easy even when there is IQ imbalance and time-varying DC offset. The frequency offset can be accurately estimated by simple signal processing.

(2) Second Embodiment In the first embodiment described above, the direct conversion type OFDM receiver can accurately perform frequency measurement by simple signal processing even in an environment where IQ imbalance and time-varying DC offset exist. A technique for estimating the offset is employed. However, such an apparatus configuration is only for realizing an appropriate frequency offset estimation, and is not intended for correcting IQ imbalance together with the frequency offset. On the other hand, under the basic configuration of the present invention, not only simple frequency offset correction but also IQ imbalance correction can be realized.

  Therefore, in the following, a second embodiment will be described in which the basic configuration of the present invention is ensured and the IQ imbalance is corrected.

  33 and 34 show the configuration of the receiver system of the wireless communication apparatus according to the second embodiment of the present invention. Both apparatuses estimate IQ imbalance and estimate IQ imbalance. It should be fully understood that it has a mechanism for performing correction. In each drawing, the same reference numerals are assigned to the same components as those shown in FIGS.

  In the OFDM receiver shown in each figure, the output signal from the differential filter 5 is input not only to the frequency offset estimation unit 6 but also to the IQ imbalance estimation unit 1000.

  The output signal from the differential filter 5 is as shown in the above equation (12). A signal delayed by N / 4 (= 16) samples with respect to this output signal is expressed by the following equation (16). (Where η = r (i) −r (i−1), γ = exp (j2πΔf (N / 4))).

  Similarly, a signal advanced by N / 4 samples with respect to the signal represented by the above equation (12) is represented by the following equation (17).

  In each of the above equations (16) and (17), γ is a value obtained from the above equation (13), and thus is obtained by processing by the phase detection circuit 208 in the frequency offset estimation unit 6. Therefore, by feeding back the calculated value in the phase detection circuit 208 to the IQ imbalance estimation unit 1000, the unknown can be reduced by one from the above equations (16) and (17), and each equation (16) and (17) can be treated as a function of d, φ and η. As a result, three equations (12), (16), and (17) are obtained for the three unknown variables φ, η, and d, and it can be seen that solutions can be derived for all the variables. .

  Therefore, the processing corresponding to the following equations (18) and (19) is performed on the three samples corresponding to each of the above three equations (12), (16), and (17).

  Therefore, the relationship shown in the following equation (20) is obtained from these equations (18) and (19).

  On the other hand, from the above equation (11), φ and ψ are each expressed by the following equations.

  Accordingly, when both equations are approximated, φ and ψ are expressed as follows.

Note that α and θ included in these expressions represent the following as described above.
(I) α = amplitude value of I component and Q component (ii) θ = phase difference between cosine signal and sine signal

  Here, the IQ imbalance is caused by the fact that the I component and the Q component of the signal have the following two relationships (a) and (b).

(A) In order to perform frequency conversion, the local signal input to the mixer 3 (that is, the output signal of the local signal oscillator 11) is divided into two output signals from the PLL, one of which is a 90-degree phase shifter. Generated by passing through. Here, when the output signal from the PLL is a high-frequency signal, the phase rotation amount of this phase shifter is not exactly 90 degrees (ie, the I component and Q component of the signal are not orthogonal), and as a result, A phase difference θ is generated.
(B) The bridegroom should be in the I and Q components at the A / D converter input and the value of α should not be zero due to loss due to the phase shifter or error in the amplifier amplification between the IQ components.

  From these relationships (a) and (b), the values of α and θ are calculated based on the three samples d (i−N / 4), d (i), and d (i + N / 4). The IQ imbalance can be corrected by correcting the complex reception baseband signal based on the calculated values of α and θ.

  Therefore, in the OFDM receiver according to the present embodiment, the IQ imbalance estimation unit 1000 calculates α and θ (that is, estimates IQ imbalance), and calculates a correction coefficient corresponding to the calculated value as IQ imbalance. By correcting the complex reception baseband signal by the correction unit 1100, the IQ imbalance is corrected.

  Below, the specific method at the time of calculating (alpha) and (theta) in the IQ imbalance estimation part 1000 is demonstrated.

  First, substituting approximate equations for φ and ψ into the above equation (20), the following equation (21) is obtained.

  Here, when the above equation (21) is divided into a real part and an imaginary part, respectively, the following expressions (22) and (23) are obtained.

  From the above equations (22) and (23), α and θ are obtained as values of the following equations (24) and (25), respectively.

  Here, ε is expressed by the above equation (20). On the other hand, the variable d in the equation (20) is obtained as an output signal from the differential filter 5. Further, γ is calculated by the phase detection circuit 208, and ε is calculated from the output signal from the differential filter 5 (ie, corresponding to d) and the feedback signal from the phase detection circuit 208 (ie, corresponding to γ). It can be seen that α and θ can be calculated.

  In the present embodiment, the IQ imbalance estimation unit 1000 executes the above processing, calculates α and θ, and outputs a correction coefficient corresponding to the calculated values to the IQ imbalance correction unit 1100 including a multiplier. . As a result, the complex reception baseband signal is multiplied by the correction coefficient by the IQ imbalance correction unit 1100, and the IQ imbalance generated in the complex reception baseband signal is corrected.

  The IQ imbalance correction unit 1100 is disposed anywhere between the ADC 4 and the DFT 8. However, it has been experimentally found that the reception characteristic can be improved by disposing the IQ imbalance correction unit 1100 upstream of the frequency offset correction unit 7. Also, when IQ imbalance correction is performed upstream of the branch point x to the differential filter 5 and the corrected signal is branched to the differential filter 5, IQ imbalance correction is performed downstream of the branch point x. It has been experimentally found that the reception characteristics can be further improved than the case. From this point of view, the OFDM receiver according to the present embodiment employs a configuration in which an IQ imbalance correction unit 1100 is disposed between the ADC 4 and the branch point x as shown in FIG.

  Of course, even if the position where the IQ imbalance correction unit 1100 is disposed is changed, a desired effect can be obtained. Therefore, it should be fully understood that the gist of the present invention is not limited to a specific arrangement position of the IQ imbalance correction unit 1100.

  The gist of the present invention is not limited to a specific method for determining the correction coefficient based on the values of α and θ. For example, a correction coefficient corresponding to the calculated values of α and θ is experimentally obtained in the IQ imbalance estimation unit 1000, and a table in which the experimental value is associated with each value of α and θ is used as the IQ imbalance estimation unit 1000. You may make it equip to. As another method, it is also possible to calculate directly from the values of α and θ. In the latter case, the following method can be employed.

  That is, from the above equation (11), the complex reception baseband signal is obtained as the following equation (26).

  Therefore, IQ imbalance can be corrected by obtaining a correction coefficient represented by the following equation (27) and multiplying the complex reception baseband signal. The configuration and operation of other elements are the same as those in the first embodiment.

  Thus, according to the OFDM receiver according to the present embodiment, the IQ imbalance correction unit 1000 calculates the correction coefficient based on the complex reception baseband signal from the differential filter 5. Then, the IQ imbalance included in the complex reception baseband signal can be corrected by multiplying the complex reception baseband signal by the IQ imbalance correction unit 1100 by the calculated correction coefficient.

  The MSE characteristics in the OFDM receiver when such a method is adopted are shown in FIGS. 35 and 36. It will be understood that the MSE characteristics are in a good state. The experimental values in both figures are the MSE when α = 0.05 and θ = 5 degrees in an environment where the gain switching is not performed in the LNA, and FIG. 35 is the MSE for estimating the error of α. FIG. 36 shows the MSE for estimating the error of θ.

  So far, the IQ imbalance correction method corresponding to the differential filter 5 and the frequency offset estimation unit 6 shown in FIG. 21 in the OFDM receiver according to the first embodiment shown in FIG. 1 has been described. Further, the IQ imbalance correction method described above can be similarly applied to other receiver configurations, and will be described below.

(2-1) Other Configurations of Differential Filter 5 and Frequency Offset Estimator 6 FIG. 37 shows IQ imbalance correction for the OFDM receiver corresponding to differential filter 5 and frequency offset estimator 6 shown in FIG. (Ie, an IQ imbalance estimation unit 1000 is added). However, in the figure, the same reference numerals are given to the same components as those in FIG.

  In FIG. 37, similarly to the configuration shown in FIG. 34, the output signal of the differential filter 5 (that is, the delay unit 301 and the adder 302) is input to the IQ imbalance estimation unit 1000, and from the phase detection circuit 311. A signal corresponding to y is fed back. Based on these input signals, the IQ imbalance estimation unit 1000 executes processes corresponding to the above equations (16) to (25), and calculates IQ imbalance correction coefficients. Then, the IQ imbalance generated in the complex reception baseband signal can be corrected by multiplying the calculated correction coefficient by the complex reception baseband signal in the IQ imbalance correction unit 1100. Note that the processes in the other components in FIG. 37 are the same as those in FIG.

(2-2) Other configuration of OFDM receiver (2-2-1) Other configuration 1
Although FIG. 33 shows the configuration of the OFDM receiver, the receiver configurations shown in FIGS. 38 to 41 are also possible. FIG. 38 shows a configuration of a receiver in which components for performing IQ imbalance correction are added to the OFDM receiver shown in FIG. 39, 40, and 41 show configurations of receivers in which components for performing IQ imbalance correction are added to the OFDM receivers shown in FIGS. 5, 6, and 7, respectively. Yes.

  In any of the receivers shown in FIGS. 38 to 41, (i) the output signal from the differential filter 5 (see FIG. 34) is input to the IQ imbalance estimation unit 1000, ii) A signal corresponding to y is fed back from the phase detection circuit 311, and the IQ imbalance estimation unit 1000 executes processing corresponding to the above equations (16) to (25) based on these input signals to perform correction. The coefficient is calculated. Based on the correction coefficient, the IQ imbalance correction unit 1100 corrects the IQ imbalance generated in the complex reception baseband signal. Since other components and their operations are the same as those shown in FIGS. 4, 5, 6, and 7, description thereof is omitted here.

  Here, in any of the OFDM receiver configurations of FIGS. 38 to 41, the position where the IQ imbalance correction unit 1100 is arranged is arbitrary as described above, and any position is provided between the ADC 4 and the DFT 8. It can also be arranged. However, each of the OFDM receivers shown in FIGS. 38 to 41 employs the following method in order to improve reception characteristics.

(I) In the case of the OFDM receiver shown in FIGS. 38 and 41 In order to input a signal after correcting the IQ imbalance to the differential filter 5, an IQ imbalance correction unit is provided between the ADC 4 and the branch point x. 1100 is provided.

(Ii) In the case of the OFDM receiver shown in FIGS. 39 and 40 For the following two reasons, the output signal from the DC offset correction unit 10 is subjected to IQ imbalance correction, and then frequency offset correction is performed. The configuration is adopted.
<Reason 1>
The reception characteristics are improved by performing the IQ imbalance correction after the DC offset correction rather than performing the IQ imbalance correction before the DC offset correction.
<Reason 2>
The reception characteristic is improved by performing the frequency offset correction after correcting the IQ imbalance rather than performing the frequency offset correction before performing the IQ imbalance correction.

(2-2-2) Other form 2
Also in the synchronous circuit configuration shown in FIGS. 8 and 9, a configuration for performing IQ imbalance correction can be added. FIGS. 42 and 43 show configuration examples in which components for performing IQ imbalance correction are added to the synchronization circuits shown in FIGS. 8 and 9, respectively. However, in FIG. 42 and FIG. 43, the same components as those shown in FIG. 8 and FIG. The apparatus configuration shown in each figure will be described below.

  In the synchronization circuit shown in FIG. 42, an IQ imbalance correction unit 1100 is provided between the DC offset correction unit 25 and the frequency offset correction unit 24. The output signal from the HPF 21 is also input to the IQ imbalance estimation unit 1000 in addition to the frequency offset estimation unit 22 and the packet detection / coarse timing detection unit 23. Based on these input signals, processing corresponding to the above equations (16) to (25) is executed to calculate the correction coefficient. The calculated correction coefficient is supplied from the IQ imbalance estimation unit 1000 to the IQ imbalance correction unit 1100, and IQ imbalance correction is performed on the complex reception baseband signal based on the correction coefficient.

  In the OFDM receiver having the synchronization circuit shown in FIG. 42, as in FIG. 8, each of the I-axis and Q-axis input signals is controlled by turning on and off the two switches 26 and 27, whereby the HPF 21 Alternatively, the input is switched so as to be input to any one of the DC offset estimation units 24. When the input signal is input to the HPF 21, the IQ imbalance estimation unit 1000 calculates a correction coefficient. Since the other components in the figure are the same as those in FIG. 8, the description thereof is omitted here.

  The synchronization circuit shown in FIG. 43 is not directly controlled on / off of the switches 26 and 27 by the detection signal of the packet detection and coarse timing detector 23, but is controlled by the switch control unit 28 as in FIG. It is comprised so that. 42, an IQ imbalance correction unit 1100 is disposed between the DC offset correction unit 25 and the frequency offset correction unit 24. The output signal from the HPF 21 is input to the frequency offset estimation unit 22 and the packet detection and coarse timing detection unit 23 as well as to the IQ imbalance estimation unit 1000. Based on these input signals, processing corresponding to the above equations (16) to (25) is executed to calculate the correction coefficient. The calculated correction coefficient is supplied from the IQ imbalance estimation unit 1000 to the IQ imbalance correction unit 1100, and IQ imbalance correction is performed on the complex reception baseband signal based on the correction coefficient.

  On the other hand, as in the synchronous circuit configuration shown in FIG. 42 and FIG. 43, not only DC offset correction, IQ imbalance correction, and frequency offset correction are performed using only STS, but also by using the subsequent LTS. These correction processes can be performed more appropriately. FIGS. 13 and 14 show the configuration of a synchronization circuit that performs frequency offset correction and channel estimation in LTS, but FIGS. 44 and 45 show IQ imbalance correction for the synchronization circuit shown in FIGS. The example of a structure which added the component for performing is shown, respectively. However, in FIG. 44 and FIG. 45, the same components as those shown in FIG. 13 and FIG. The apparatus configuration shown in each figure will be described below.

In the synchronization circuit shown in FIG. 44, an IQ imbalance estimation unit 1400 and an IQ imbalance correction unit 1500 for LTS are provided independently of the IQ imbalance estimation unit 1200 and IQ imbalance correction unit 1300 for STS. It has been. The IQ imbalance estimation unit 1200 for STS calculates a correction coefficient for performing coarse IQ imbalance correction using the short preamble, and the IQ imbalance correction unit 1300 performs the calculation on the complex reception baseband signal based on the correction coefficient. IQ imbalance correction is performed. On the other hand, the IQ imbalance estimation unit 1400 for LTS calculates a correction coefficient for performing more precise IQ imbalance correction using the long preambles T ₁ and T ₂ , and the IQ imbalance correction unit 1500 Based on this correction coefficient, IQ imbalance correction is performed on the complex reception baseband signal.

  In the OFDM receiver having the synchronization circuit shown in FIG. 44, the two switches 26 and 27 are turned on / off at the timing when the short preamble section ends, as in FIG. 13, and the IQ input terminal is connected to the HPF 21. Switching to the DC offset estimation unit 24 is performed. After the LTS, the IQ imbalance correction unit 1300 performs IQ imbalance correction by multiplying the complex reception baseband signal by the correction coefficient calculated until the STS ends. Thereafter, a secondary (ie, remaining) IQ imbalance is estimated by the IQ imbalance estimation unit 1400 using the LTS, and the IQ imbalance correction unit 1500 performs correction.

  In the circuit configuration shown in FIG. 44, circuit modules 1400 and 1500 for performing IQ imbalance correction on complex reception baseband signals after the LTS are separately provided. In contrast, in the synchronization circuit shown in FIG. 45, IQ imbalance estimation unit 1600 and IQ imbalance correction unit 1700 perform IQ imbalance correction using STS, and also perform IQ imbalance correction after LTS. It is configured as follows.

  The present invention has been described in detail above with reference to specific embodiments. However, it is obvious that those skilled in the art can make modifications and substitutions of the embodiment without departing from the gist of the present invention.

  In the present specification, the embodiment applied to the wireless communication system defined by IEEE 802.11 has been mainly described, but the gist of the present invention is not limited to this. By applying the receiver according to the present invention to a wireless communication system in which the DC subcarrier is a null signal and the same OFDM symbol is repeatedly transmitted in the preamble portion, it is possible to similarly estimate the frequency offset accurately. The present invention can be applied to various digital communication technologies of the OFDM transmission system such as terrestrial digital broadcasting, fourth generation mobile communication, and power line communication (Power Line Communication) in addition to the wireless LAN.

  Further, the present invention can solve the problem of DC offset occurring in a receiver to which the direct conversion method is applied, but the gist of the present invention is not limited to this. Even in the case of a receiver that down-converts an RF reception signal by another frequency conversion method, the present invention can be suitably applied when the problems of DC offset and frequency offset occur simultaneously.

  In short, the present invention has been disclosed in the form of exemplification, and the description of the present specification should not be interpreted in a limited manner. In order to determine the gist of the present invention, the claims should be taken into consideration.

FIG. 1 is a diagram showing a configuration of a receiver system of a wireless communication apparatus according to the first embodiment of the present invention. FIG. 2 is a diagram showing a subcarrier configuration of an OFDM symbol in a wireless LAN system according to IEEE 802.11a / g. FIG. 3 is a diagram showing the frequency offset value normalized by the square error of the frequency offset estimated value and the subcarrier interval when the DC offset power versus the OFDM signal power is 30 dB. FIG. 4 is a diagram illustrating another configuration example of the receiver system of the wireless communication device. FIG. 5 is a diagram illustrating another configuration example of the receiver system of the wireless communication device. FIG. 6 is a diagram illustrating another configuration example of the receiver system of the wireless communication device. FIG. 7 is a diagram illustrating another configuration example of the receiver system of the wireless communication device. FIG. 8 is a diagram showing a configuration example around a synchronization circuit that performs frequency offset, packet detection, and coarse timing detection using the output of the high-pass filter. FIG. 9 is a diagram showing another configuration example around the synchronization circuit that performs frequency offset, packet detection, and coarse timing detection using the output of the high-pass filter. FIG. 10 is a diagram illustrating an example of a method for adjusting the switch switching timing by the switch control unit 28. FIG. 11A is a diagram illustrating a configuration example of the high-pass filter 21 using a difference filter. FIG. 11B is a diagram illustrating a configuration example of the high-pass filter 21 using DC offset estimation and correction based on a moving average. FIG. 12 is a diagram illustrating a configuration example of the DC offset estimation unit 25. FIG. 13 is a diagram showing a configuration example around a synchronous circuit including a circuit module that performs frequency offset correction and channel estimation in LTS. FIG. 14 is a diagram illustrating a configuration example around the synchronization circuit configured such that the frequency offset estimation unit 22 and the frequency offset correction unit 24 perform estimation and removal of the frequency offset of the received baseband signal after the LTS, respectively. . FIG. 15 is a diagram showing a preamble configuration defined in IEEE 802.11a / g. FIG. 16 is a diagram schematically showing a configuration of a receiver that removes a DC offset using HPF. FIG. 17A is a diagram illustrating a manner in which the DC component of the OFDM signal is removed with a sufficiently small HPF with respect to the subcarrier interval. FIG. 17B is a diagram illustrating a state in which a DC component of an OFDM signal is removed with a large HPF with respect to the subcarrier interval, and a nearby signal is deleted. FIG. 18 is a diagram schematically illustrating a configuration of a receiver that simultaneously estimates a DC offset and a frequency offset. FIG. 19 is a diagram schematically illustrating a configuration of a receiver that estimates a DC offset and a frequency offset in parallel. FIG. 20 is a diagram schematically illustrating the configuration of a receiver that repeatedly performs DC offset estimation and frequency offset compensation. FIG. 21 is a diagram illustrating a specific configuration example of the differential filter 5 and the frequency offset estimation unit 6. FIG. 22 is a diagram for explaining the influence of the gain switching of the low noise amplifier 2 by automatic gain control on the output of the differential filter 5. FIG. 23 is a diagram for explaining the cause of IQ imbalance. FIG. 24 is a diagram in which the frequency offset information is represented as a vector on a complex space. FIG. 25 is a diagram illustrating a specific configuration example of the differential filter 5 and the frequency offset estimation unit 6. FIG. 26 is a diagram illustrating an estimated vector of the frequency offset before and after the gain switching of the low noise amplifier 2 is performed. 27 shows 15 samples of the output of the multiplier 305 corresponding to the short preamble symbols t ₃ and t ₄ and 79 samples of the output of the multiplier 305 corresponding to the short preamble symbols t ₅ and t ₁₀ . It is the figure which showed the synthetic | combination estimation vector obtained by adding to an output sequentially. FIG. 28 is a diagram showing a combined estimated vector obtained by weighted addition using a weighting coefficient corresponding to the absolute value of the output signal of multiplier 305. FIG. 29 is a diagram illustrating the frequency offset estimation accuracy (mean square error versus normalized frequency offset value) of the conventional method and the proposed method. FIG. 30 is a diagram showing a subcarrier arrangement in IEEE 802.11a / g. FIG. 31 is a diagram for explaining DC offset to frequency offset. FIG. 32 is a diagram showing how the residual DC offset is removed by the differential filter 5. FIG. 33 is a diagram illustrating a configuration of a receiver system of the wireless communication device according to the second embodiment of the present invention. FIG. 34 is a diagram illustrating specific configurations of the differential filter 5, the frequency offset estimation unit 6, and the IQ imbalance estimation unit 1000. FIG. 35 is a diagram showing an MSE for estimating an error of α when α = 0.05 and θ = 5 degrees in an environment where gain switching is not performed in the LNA. FIG. 36 is a diagram showing an MSE of θ error estimation when α = 0.05 and θ = 5 degrees in an environment where gain switching is not performed in the LNA. FIG. 37 is a diagram illustrating specific configurations of the differential filter 5, the frequency offset estimation unit 6, and the IQ imbalance estimation unit 1000. FIG. 38 is a diagram illustrating another configuration example of the receiver system of the wireless communication device. FIG. 39 is a diagram illustrating another configuration example of the receiver system of the wireless communication device. FIG. 40 is a diagram illustrating another configuration example of the receiver system of the wireless communication device. FIG. 41 is a diagram illustrating another configuration example of the receiver system of the wireless communication device. FIG. 42 is a diagram showing a configuration example of the periphery of a synchronization circuit that performs frequency offset correction, packet detection, and coarse timing detection using the output of the high-pass filter. FIG. 43 is a diagram showing a configuration example around a synchronization circuit that performs frequency offset correction, packet detection, and coarse timing detection using the output of the high-pass filter. FIG. 44 is a diagram showing a configuration example around a synchronous circuit including a circuit module that performs frequency offset correction, IQ imbalance correction, and channel estimation in LTS. FIG. 45 is a diagram showing a configuration example around a synchronous circuit including a circuit module that performs frequency offset correction, IQ imbalance correction, and channel estimation in LTS.

Explanation of symbols

1 ... Bandpass filter (BPF)
2. Low noise amplifier (LNA)
3 ... Mixer 4 ... AD converter (ADC)
DESCRIPTION OF SYMBOLS 5 ... Differential filter 6 ... Frequency offset estimation part 7 ... Frequency offset correction part 8 ... Discrete Fourier-transform part (DFT)
DESCRIPTION OF SYMBOLS 9 ... DC offset estimation part 10 ... DC offset correction | amendment part 11 ... Local oscillator 12 ... High pass filter (HPF)
DESCRIPTION OF SYMBOLS 21 ... High-pass filter 22 ... Frequency offset estimation part 23 ... Packet detection and rough timing detection part 24 ... Frequency offset correction part 25 ... DC offset estimation part 26, 27 ... Switch 28 ... Switch control part 31 ... Frequency offset estimation part 32 ... Frequency offset correction unit 33 ... channel estimation unit 201 ... delay unit 202 ... adder 203 ... delay unit 204 ... complex conjugate calculation circuit 205 ... multiplier 206 ... adder 207 ... storage element 208 ... phase detection circuit 301 ... delay unit 302 ... Adder 303 ... Delayer 304 ... Complex conjugate calculation circuit 305 ... Multiplier 306 ... Multiplier 307 ... Coefficient calculation circuit 309 ... Adder 310 ... Storage element 311 ... Phase detection circuit 1000, 1200, 1400, 1600 ... IQ imbalance estimation Part 1100, 1300, 1500 1700 ... IQ imbalance correction unit

Claims (31)

A wireless communication device for receiving a packet composed of an OFDM modulated signal,
A bandpass filter for extracting an OFDM signal in a desired band;
A low-noise amplifier that amplifies the desired OFDM signal, the gain of which is controlled according to the intensity of the received signal;
A frequency converter that down-converts the amplified OFDM signal into a baseband signal;
An analog-to-digital converter that converts a baseband signal into a digital signal;
A first high pass filter for removing a DC offset for a baseband signal corresponding to a predetermined preamble portion of the packet;
A frequency offset estimator for estimating a frequency offset from the signal after removing the DC offset by the first high-pass filter;
A frequency offset correction unit that removes the estimated frequency offset from the received baseband signal used for estimating the frequency offset;
A demodulator that demodulates the subcarrier signals arranged on the frequency axis from the baseband signal after compensating for the frequency offset;
A wireless communication apparatus comprising: The first high-pass filter comprises a differential filter;
The wireless communication apparatus according to claim 1. When the first high-pass filter detects a sudden change in DC offset, the first high-pass filter inputs a detection signal to the frequency offset estimation unit,
The frequency offset estimation unit excludes a sample output from the first high-pass filter at the time when the detection signal is input from a target of frequency offset estimation;
The wireless communication apparatus according to claim 1. The input OFDM signal does not use DC subcarriers,
The frequency offset estimation unit estimates a frequency offset using a preamble in which two symbols of the same OFDM signal are transmitted.
The wireless communication apparatus according to claim 1. One OFDM symbol consists of n subcarriers, and the i-th sample of the time waveform of the OFDM symbol transmitted twice is s (i), and the OFDM symbol sample transmitted the first time is {s ( 0), s (1),..., S (n−1)}, and samples of the OFDM symbol transmitted for the second time are {s (n), s (n + 1),. When the frequency offset is Δf and the DC offset is D, the first high-pass filter performs processing represented by the following equation (2) on the received baseband signal represented by the following equation (1). The sample signal d (i) is output, and the frequency offset correction unit estimates the frequency offset Δf by performing processing represented by the following expression (3) using the sample signal d (i).

The wireless communication apparatus according to claim 4. When the absolute value of the sample output d (i) increases due to the change in DC offset D (i + 1) -D (i) between the i-th and i + 1-th samples, the first high-pass filter Outputting a detection signal for excluding estimation of the frequency offset using the above equation (3) in the second sample to the frequency offset estimation unit;

The wireless communication apparatus according to claim 5. The frequency converting unit converts the amplified desired OFDM signal into a baseband signal by a direct conversion method using a local frequency generated by a local oscillator,
Based on the frequency offset estimated by the frequency offset estimation unit, the phase of the local frequency oscillating in the local oscillator is reversely rotated.
The wireless communication apparatus according to claim 1. A DC offset estimator for estimating a DC offset in a baseband signal after digital conversion by the analog-digital converter;
A DC offset correction unit that removes the estimated DC offset from the baseband signal after digital conversion;
The wireless communication apparatus according to claim 1, further comprising: When the first high-pass filter detects a sudden change in DC offset, the first high-pass filter inputs a detection signal to the DC offset estimation unit,
The DC offset estimation unit excludes a DC offset estimation value calculated before the detection signal is input, and recalculates an estimation value.
The wireless communication apparatus according to claim 8. A second high-pass filter for filtering a baseband signal output from the frequency converter;
The analog-to-digital converter converts the baseband signal after passing through the second high-pass filter into a digital signal;
The wireless communication apparatus according to claim 1. Setting the cutoff frequency of the second high-pass filter to a value smaller than the cutoff frequency of the first high-pass filter;
The wireless communication apparatus according to claim 10. A detection unit for detecting packet detection and coarse timing using an output signal of the first high-pass filter;
The wireless communication apparatus according to claim 8. A switch that exclusively connects the output terminal of the analog-digital converter to the path of the first high-pass filter or the DC offset correction unit;
The wireless communication apparatus according to claim 12. The switch connects the output terminal of the analog-digital conversion unit from the path to the first high-pass filter at the timing when the coarse timing detection unit detects the end of a predetermined preamble portion for performing frequency offset estimation. Switching to the path to the DC offset correction unit;
The wireless communication apparatus according to claim 13. The frequency offset estimation unit estimates a frequency offset using a reception baseband signal from which a DC offset has been removed using the first high-pass filter, using a period until the predetermined preamble portion ends. Do,
The wireless communication apparatus according to claim 14. The DC offset estimation unit performs DC offset estimation using a period until the predetermined preamble part ends,
The DC offset correction unit corrects the estimated DC offset for the received baseband signal after the predetermined preamble portion is completed.
The wireless communication apparatus according to claim 14. The frequency offset correction unit uses the frequency offset estimated by the frequency offset estimation unit until the end of the predetermined preamble portion, and uses the frequency offset for the received baseband signal after the end of the predetermined preamble portion. Offset correction,
The wireless communication apparatus according to claim 14. A switch control unit that calculates a moving average value of a correlation value of an output signal of the analog-digital conversion unit and controls switching of the switch at a timing when the moving average value exceeds a predetermined threshold;
The wireless communication apparatus according to claim 13. A packet to be received has a short preamble portion made up of a short training sequence having a relatively small subcarrier interval and a long preamble portion made up of a long training sequence having a relatively small subcarrier interval,
In accordance with the timing at which the short preamble portion ends and the long preamble portion starts, in the switch portion, the output terminal of the analog-digital conversion portion is passed from the path to the first high-pass filter to the DC offset correction portion. Switch to the path,
The wireless communication apparatus according to claim 13. The frequency offset compensation unit removes the frequency offset estimated by the frequency offset estimation unit in the short preamble unit from the long preamble unit,
The DC offset correction unit removes the DC offset estimated by the DC offset estimation unit in the short preamble unit from the long preamble unit,
A second frequency offset estimator for estimating a frequency offset from a received baseband signal after the long preamble portion after the frequency offset and the DC offset estimated by the short preamble portion are removed; and the second frequency offset A second frequency offset correction unit that removes the frequency offset estimated by the estimation unit from the received baseband signal after the long preamble unit;
The wireless communication apparatus according to claim 19. The frequency offset correction unit removes the frequency offset estimated by the frequency offset estimation unit in the short preamble unit from the long preamble unit,
The DC offset correction unit removes the DC offset estimated by the DC offset estimation unit in the short preamble unit from the long preamble unit,
The received baseband signal after the long preamble part after the frequency offset and the DC offset estimated by the short preamble part are removed is fed back to the frequency offset estimation part to return the frequency from the received baseband signal after the long preamble part. Estimating an offset and removing it from the received baseband signal after the long preamble part in the frequency offset correction part,
The wireless communication apparatus according to claim 19. A channel estimator for performing channel estimation from a received baseband signal in which the frequency offset and the DC offset estimated in the short preamble part are removed and frequency offset correction is performed based on the frequency offset estimation after the long preamble part; In addition,
The wireless communication apparatus according to claim 20, wherein the wireless communication apparatus is a wireless communication apparatus. An IQ imbalance estimation unit for estimating an IQ imbalance from the signal after the DC offset is removed by the first high-pass filter;
An IQ imbalance correction unit that performs IQ imbalance correction on the intended received baseband signal used for estimating the IQ imbalance;
The wireless communication apparatus according to claim 1, further comprising: The IQ imbalance correction unit performs IQ imbalance correction on the received baseband signal after the DC offset is corrected.
24. The wireless communication apparatus according to claim 23. The frequency offset correction unit removes the estimated frequency offset from the received baseband signal after IQ imbalance correction is performed by the IQ imbalance correction unit;
24. The wireless communication apparatus according to claim 23. Preamble symbols that can be used for frequency offset estimation are transmitted multiple times,
The frequency offset estimation unit adds a frequency offset estimated for each preamble symbol to obtain a final frequency offset estimation value.
The wireless communication system according to claim 1. A low noise amplifier that amplifies the received signal, and a gain control unit that adjusts the gain of the low noise amplifier;
The frequency offset estimation unit weights and adds a final frequency offset value to each frequency offset estimated for each preamble symbol according to a gain set in the low noise amplifier when the corresponding preamble symbol is received. Get the
The wireless communication system according to claim 28. The frequency offset estimator calculates a weighting factor based on the absolute value of the frequency offset estimated for each preamble symbol, and weights and adds the frequency offset for each preamble symbol with each weighting factor to obtain a final frequency offset. Get value,
27. The wireless communication system according to claim 26. The frequency offset estimator gives a weighting factor of 0 to a frequency offset whose absolute value exceeds a predetermined threshold, and gives a weighting factor of 1 to a frequency offset whose absolute value does not exceed the predetermined threshold. The final frequency offset value is obtained by weighting and adding the frequency offset for each symbol with the respective weighting factors.
The wireless communication system according to claim 28. The frequency offset estimation unit determines the predetermined threshold based on received signal strength;
30. The wireless communication system according to claim 29. The frequency offset estimator gives each frequency offset a weighting factor consisting of the reciprocal of the absolute value, and weights and adds the frequency offset for each preamble symbol with each weighting factor to obtain a final frequency offset value.
The wireless communication system according to claim 28.

Images