Correction of a sampling frequency offset in an orthogonal frequency division multiplexing system by sidelobe analysis of pilot subcarriers

[0001] (Field of invention) The present invention relates to the processing of orthogonal frequency division multiplexing (OFDM) signals. [0002] (Background of invention) Orthogonal Frequency Division Multiplexing (OFDM) is a robust technique for effectively transmitting data over channels. This technique uses multiple subcarrier frequencies (subcarriers) within the channel bandwidth to carry the data. These subcarriers are configured for optimal bandwidth efficiency when compared to many traditional transmission techniques such as frequency division multiplexing (FDM). Frequency division multiplexing (FDM) divides the subcarrier frequency spectrum and<u style="single">Separation</u>Between carriers by doing<u style="single">interference</u>Much of the channel bandwidth is wasted to prevent (ICI). In contrast, the frequency spectra of the OFDM subcarriers overlap significantly within the OFDM channel bandwidth, but nonetheless, OFDM decomposes and decomposes the modulated information on each subcarrier.<u style="single">Regeneration</u>Is possible. [0003] Transmission of data over channels over OFDM signals offers several advantages over many traditional transmission techniques.<u style="single">Have</u>.. One of its advantages is<u style="single">Multipath</u>The tolerance of OFDM to multipath delay spread. This tolerance is due to the relatively long symbol interval Ts compared to the typical time duration of the channel impulse response. By increasing the symbol interval, between symbols<u style="single">interference</u>(ISI) has been avoided. Another advantage is the resistance of OFDM to frequency-selective fading. By providing redundancy in the OFDM signal, from other subcarriers<u style="single">Regeneration</u>From the data, the data encoded on the fading subcarrier can be restored. Yet another advantage is that in the case of OFDM, the spectrum can be used effectively. Since the OFDM subcarriers are placed very close to each other without the need for an unused frequency space between the subcarriers, OFDM can effectively fill the channel. Yet another advantage is that OFDM subchannel equalization is simplified. OFDM goes from the time domain (as is the case with single carrier transmission systems) to the frequency domain where a simple 1-tap equalizer bank can individually adjust the phase and amplitude distortion of each subchannel. Shift channel equalization. Yet another advantage is the goodness of OFDM<u style="single">interference</u>It is a characteristic.<u style="single">interference</u>The OFDM spectrum can be modified to take into account the distribution of signal power, and out-of-band by avoiding the use of OFDM subcarriers near the edges of the channel bandwidth.<u style="single">interference</u>Can be alleviated. [0004] Although OFDM has these advantages, there are several embodiments of OFDM prior art.<u style="single">problem</u>And there are practical restrictions.<u style="single">problem</u>One is the problem of measuring and correcting carrier frequency offsets, which is a major aspect of OFDM synchronization. Ideally, the receiving carrier frequency f_crIs exactly<u style="single">Send</u>Carrier frequency f_ctMust be consistent with. If this condition is not met, a non-zero carrier frequency offset Δ (delta) f of the received OFDM signal due to inconsistency._cWill cause. OFDM signals are extremely sensitive to such carrier frequency offsets, which impairs orthogonality between OFDM subcarriers and between carriers.<u style="single">interference</u>(ICI) and at the receiver<u style="single">Regeneration</u>It is responsible for the significant increase in data error rate (BER). [0005] other<u style="single">problem</u>Is to synchronize the sample rate of the transmitter with the sample rate of the receiver in order to eliminate the sampling rate offset. If the frequency offset is small, any inconsistency between the two sample rates will be 2^mIt causes the term sub-symbol array to rotate from symbol to symbol in the frame, but if the frequency offset is large, the frequency spectrum of the received signal will shrink or expand. Both contraction and expansion of the spectrum are responsible for the increase in BER. One of the causes of the sampling rate offset is that the sampling frequency is offset. Sampling frequency offset occurs when the receiver samples the received signal at a frequency higher or lower than the sample rate used by the transmitter. Sampling frequency offsets are detrimental to receiver performance and must be corrected for proper receiver synchronization. The present invention corrects this problem.<u style="single">Target</u>It was. [0006] (Outline of the invention) The OFDM receiver samples the input signal in the time domain and multiplies the sample data by a window function to reduce the sublobe of each frequency domain spectrum of the given subcarriers to prevent interference between the given subcarriers. Sample signals to relax and expand the main lobe so that samples of sufficient amplitude (relative to the noise floor) are available on both sides of the main peak and analyze frequency domain samples for each given subcarrier. FFT (Fast Fourier Transform) is taken to detect the difference in amplitude of frequency domain samples on both sides of each predetermined subcarrier, and a sampling frequency error is generated based on the difference in detected amplitude. [0007] The present invention will be more fully understood by reading the detailed description of preferred embodiments with the accompanying drawings. [0008] (Detailed description of preferred embodiments) The features and advantages of the present invention will become clearer from the following description provided as an example. [0009] The general principle of OFDM signal transmission will be described with reference to FIGS. 1 to 4. Referring to FIG. 1, a block diagram of a conventional OFDM transmitter 10 is shown. The OFDM transmitter 10 receives a stream 12 of baseband data bits as an input. These input data bits 12 are sent to the encoder 14. The encoder 14 sets these data bits 12 to T._g+ T_sBringing into the B-bit segment every second. T_sIs the OFDM symbol interval and T_gIs a cyclic prefix or guard interval. The encoder 14 typically introduces error correction and / or error detection redundancy into the B-bit segment and blocks and / or superimposes the encoded bits into 2N sub-segments of m-bits. You are using a coding scheme. Usually, the range of the integer m is from 2 to 6. [0010] A typical OFDM transmission system has 2N + 1 OFDM subcarriers, including a zero frequency DC subcarrier that is not normally used for data transmission because it has no frequency and therefore no phase. Therefore, the encoder 14 usually has an m-bit subsegment, 2^m2 of 2N subsegments of m bits to map to the corresponding given complex digitization points in the term array^mPerforming term quadrature amplitude modulation (QAM) coding. Each complex digitization point in the array represents a discrete value of phase and amplitude. Thus, the encoder 14 corresponds to each of the 2N subsegments of m bits by -N k N in order to generate a sequence of frequency domain subsymbols that encode the B data bits. Complex quantified 2^mTerm QAM subsymbol c_k= a_k+ jb_kIs assigned. Also, zero frequency subcarriers are usually c₀= 0 is assigned. The encoder 14 then passes the sequence of subsymbols, along with all the additional zeros needed to simplify filtering, to the inverse discrete Fourier transform (IDFT), preferably the inverse fast Fourier transform (IFFT) 16. ing. [0011] When IFFT16 receives a sequence of OFDM frequency domain subsymbols from encoder 14, it performs an inverse fast Fourier transform on the sequence of subsymbols. That is, IFFT16 is a complex digitized subsymbol c._kEach of these is used to modulate the phase and amplitude of one of the 2N + 1 subcarrier frequencies over the symbol interval Ts over the symbol interval Ts. Subcarrier is e^-2^π^jfk^tGiven in, therefore, k is the number of frequencies and is an integer in the range -N k N, f_k= k / T_sHas a baseband frequency of. Therefore IFFT16 is 0 &lt;t &lt;T_sIs [0012] [Number 1]<img file="JP4808888B2_D0001.tif" />[0013] Produces a digital time domain OFDM symbol of duration Ts given in. [0014] T_sAs a result of this discrete modulation of the OFDM subcarriers by the frequency domain subsymbols over the symbol interval of seconds, the OFDM subcarriers each show a sink x = (sin x) / x spectrum in the frequency domain. 1 / T for each of the 2N + 1 subcarriers in the frequency domain_sThe first peak of the sink x spectrum of each subcarrier coincides with the null of the spectrum of all other subcarriers. In this way, even if the spectra of the subcarriers overlap, their orthogonality with each other is maintained. Figure 2 shows the carrier frequency f within the OFDM channel bandwidth BW._ctThe arrangement of the OFDM subcarriers centered on and the envelope of the modulation spectrum thereof are shown. Note that the modulation subcarriers fill the channel bandwidth very effectively. [0015] Returning to FIG. 1, the digital time domain OFDM symbol generated by IFFT 16 is then passed to the digital signal processor (DSP) 18. DSP18 performs additional spectral shaping for digital time domain OFDM symbols and also has a length T for each symbol._gCyclic prefix or guard interval is added. Cyclic prefixes are usually just repeating parts of a symbol. This cyclic prefix is usually longer than the OFDM channel impulse response and thus acts to prevent intersymbol interference (ISI) between consecutive symbols. [0016] The real and imaginary digital components that make up the periodically expanded and spectrally shaped digital time domain OFDM symbol are then passed to the digital-to-analog converters (DACs) 20 and 22, respectively. Is done. DACs 20 and 22 convert the real and imaginary digital components of the time domain OFDM symbols into conversion rates, or sampling rates f, determined by the clock circuit 24._{ck_t}It is converted into an in-phase OFDM analog signal and an orthogonal OFDM analog signal, respectively. The common mode and orthogonal OFDM signals are then passed to mixers 26 and 28, respectively. [0017] Mixers 26 and 28 use in-phase and orthogonal OFDM signals from DACs 20 and 22, respectively, to generate in-phase IF OFDM and orthogonal IF OFDM signals, using in-phase intermediate frequency (IF) and 90-degree signals, respectively. The phase shift (orthogonal) IF signal is modulated. The in-phase IF signal supplied to the mixer 26 is directly generated by the local oscillator 30, and the 90 ° phase shift IF signal supplied to the mixer 28 is generated by the local oscillator 30 prior to being supplied to the mixer 28. It is generated by passing the IF signal through the 90 ° phase shifter 32. These two common mode IF OFDM signals and orthogonal IF The OFDM signals are then coupled by a coupler 34 to form a composite signal. Some traditional transmitters use a digital synthesizer and a digital mixer to perform the IF coupling in the time domain prior to performing the digital-to-analog conversion. [0018] This composite IF OFDM signal is then passed to the radio frequency (RF) transmitter 40. The RF transmitter 40 has many variants and is well known in the art, but typically the RF transmitter 40 is an IF bandpass filter 42, RF mixer 44, RF carrier frequency local oscillator 46, RF. It has a bandpass filter 48, an RF power amplifier 50, and an antenna 52. The RF transmitter 40 receives an IF OFDM signal from the coupler 34 to generate an RF OFDM modulated carrier that occupies the channel bandwidth BW, and uses that signal to generate the frequency f generated by the RF local oscillator 46._ctThe transmission carrier is modulated. The width of the channel bandwidth must be at least (1 / T) to adapt to all modulated OFDM subcarriers, as the entire OFDM signal must fit into this channel bandwidth._s) · (2N + 1) Hz must be. FIG. 2 shows the frequency domain characteristics of this RF OFDM modulation carrier. The RF OFDM modulation carrier is then transmitted from the antenna 52 to a remote OFDM receiver via a channel. In an alternative embodiment of the RF transmitter 40, an OFDM signal is used to modulate the transmission carrier using frequency modulation (FM), single sideband modulation (SSM), or other modulation technique. Therefore, the resulting RF OFDM modulation carrier does not necessarily have exactly the same waveform as the RF OFDM modulation carrier shown in FIG. 2 (ie, the RF OFDM modulation carrier is not centered on the transmission carrier, but on both sides of it. Existing). [0019] In order to receive the OFDM signal at a remote location and recover the baseband data bits encoded in the OFDM subcarrier, the OFDM receiver is essentially performed by the OFDM transmitter described above. The reverse operation of all operations must be performed. These operations will be described in light of FIG. 3, which is a block diagram of a conventional OFDM receiver. [0020] The first component of a typical OFDM receiver 60 is the RF receiver 70. Like the RF transmitter 40, the RF receiver 70 has many variants and is well known in the art, but typically the RF receiver 70 has an antenna 72, a low frequency amplifier (LNA) 74. , RF bandpass filter 76, automatic gain control (AGC) circuit 77, RF mixer 78, RF carrier radio frequency local oscillator 80, and IF band pass filter 82. [0021] [0021] Through the antenna 72, the RF receiver 70 is coupled to the RF OFDM modulation carrier after the RF OFDM modulation carrier has passed through the channel. Then RF OFDM modulated carrier, frequency f generated by RF local oscillator 80_crBy combining with the receiving carrier of, the RF receiver 70 down-converts the RF OFDM modulated carrier to obtain the received IF OFDM signal. The difference between the frequencies of the receiving carrier and the transmitting carrier is the carrier frequency offset Δf._cContributes to. [0022] This received IF OFDM signal is then fed to both Mixer 84 and Mixer 86 and combined with an in-phase IF signal and a 90 ° phase-shift (orthogonal) IF signal, respectively, to produce an in-phase OFDM signal and an orthogonal OFDM signal, respectively. Orthogonal. The common mode IF signal supplied to the mixer 84 is generated by the IF local oscillator 88. The 90 ° phase shift IF signal supplied to the mixer 86 is extracted from the in-phase IF signal of the IF local oscillator 88 by passing the in-phase IF signal through the 90 ° phase shifter 90 prior to the supply to the mixer 86. .. [0023] The in-phase OFDM signal and the orthogonal OFDM signal are then passed to the analog-to-digital converters (ADCs) 92 and 93, respectively, and the sampling rate f determined by the clock circuit 94._{ck_r}It is digitized with. ADCs 92 and 93 generate digital samples that form in-phase discrete-time OFDM signals and orthogonal discrete-time OFDM signals, respectively. The difference between the sampling rate of the receiver and the sampling rate of the transmitter is the sampling rate offset Δf._ck= f_{ck_r}-f_{ck_t}Is. [0024] The unfiltered in-phase discrete-time OFDM signals and orthogonal discrete-time OFDM signals from ADCs 92 and 93 then pass through digital low-pass filters 96 and 98, respectively. The outputs of the low pass digital filters 96 and 98 are filtered into in-phase and orthogonal samples of the received signal OFDM, respectively. In this way, the received OFDM signal is the complex-valued OFDM signal r._i= q_i+ jp_iHomeomorphism (q) representing the real and imaginary components of_i) Samples and orthogonal (p)_i) Converted to a sample. These in-phase and orthogonal (real and imaginary) samples of the received OFDM signal are then passed to the DSP100. Note that in some conventional embodiments of the receiver 60, analog-to-digital conversion is performed prior to the IF coupling process. In such an embodiment, a digital mixer and a digital frequency synthesizer must be used in the coupling process. Also note that in many conventional embodiments of the receiver 60, the digital / analog conversion is performed after filtering. [0025] The DSP100 performs various operations on in-phase and orthogonal samples of the received OFDM signal. These operations include a) synchronizing receiver 60 to the timing of symbol frames and data frames in the received OFDM signal, b) removing cyclic prefixes from the received OFDM signal, c) during each OFDM symbol interval. Compute the discrete Fourier transform (DFT), or preferably the fast Fourier transform (FFT), of the received OFDM signal to recover the sequence of frequency region subsymbols used to modulate the subcarriers, d) subcarriers Perform all necessary channel equalization for, and e) A sequence of frequency region subsymbols from each symbol of the OFDM signal by demodating the subcarriers of the OFDM signal by FFT calculation._kContains operations to calculate. The DSP100 then passes the sequence of these subsymbols to the decoder 102. [0026] The decoder 102 recovers the transmitted data bits from the sequence of frequency domain subsymbols passed from the DSP 100 to the decoder 102. This recovery is performed by decoding the frequency domain subsymbols to obtain a stream 104 of the data bits that should ideally match the stream 12 of the data bits fed to the OFDM transmitter 10. This decoding process can include, for example, soft-viterby decoding and / or Reed-Solomon decoding to recover data from blocks and / or superposed coded subsymbols. [0027] In a typical OFDM data transmission system, such as a system for implementing digital television or wireless local area networks (WLANs), data is transmitted in an OFDM signal with a set of symbols known as frames. Figure 4 illustrates this concept, where data frame 100 contains M consecutive symbols 112a, 112b, ..., 112M, each of which has a guard interval T._gAnd the OFDM symbol interval Ts is included. Therefore, the total duration of each symbol is T_g+ T_sSeconds. Depending on the application, data frames can be transmitted continuously, for example in the case of digital TV broadcast communication, or in the case of WLAN embodiments, for example, data frames in the form of bursts at random times. Can be transmitted. [0028] Next, with reference to FIGS. 5 to 8, exemplary embodiments according to the present invention are shown. Although the present invention is shown separately from the components of the OFDM receiver of FIG. 3, engineers in the art can easily devise the integration of the present invention with the components of the OFDM receiver. Yeah. For example, the present invention can combine the ADCs 92 and 93 of the OFDM receiver 60 and the clock circuit 94, but the present invention is to make it easy to understand and refer to, and to understand the present invention. For ease of use, it is shown as an individual sampling frequency loop. [0029] The present invention is ETSI-BRAN Although described as operating in a receiver that complies with the HIPERLAN / 2 (Europe) and IEEE802.11a (US) wireless LAN standards, the present invention is within the skill of engineers in the art. It is considered that the teaching is carried out within other OFDM systems. [0030] The above wireless LAN standard proposes the use of training lines to detect OFDM transmissions. Briefly, a set of short OFDM training symbols (known) transmitted to a training column (eg, training column A or B) via a predetermined number of pilot subcarriers or bins (eg, 12 pilot subcarriers). Has the amplitude and phase of). All other subcarriers (eg 52 subcarriers) maintain zero while the training sequence is being transmitted. The use of training columns in the LAN standard above will be discussed below, but the use of alternative training columns and symbols will also be considered within the scope of the invention, as defined in each claim of the claims. ing. [0031] More specifically, HIPERLAN / 2 short training column B consists of 12 non-zero pilot subcarriers, all other carriers being zero (64 subcarriers in total). The presence of a sampling frequency offset when sampling a time domain OFDM signal results in the frequency domain spectrum being scaled along the frequency axis. Such scaling means that the 12 non-zero pilot subcarriers are no longer present on frequencies that are multiples of the fundamental frequency. Therefore, a DFT leak will occur, and therefore, instead of having a single peak in the FFT output corresponding to the pilot subcarrier, a peak with a secondary lobe (samples on both sides of the main peak), as shown in FIG. Will exist. By calculating the difference between the sample on the left side of the peak and the sample on the right side of the peak, it can be used to successfully update the sampler frequency, thereby generating a significant error that eliminates the sampling frequency offset. Can be done. [0032] Next, referring to FIG. 6, the sampling frequency correction network 120 is shown. It should be noted that the network 120 can be embodied in software, hardware, or some combination thereof. The network 120 receives a sample signal from a sampler 122 having an analog-to-digital converter (eg ADC 92 and 93 in FIG. 3) driven by a variable clock circuit (eg clock circuit 94 in FIG. 3). As discussed above, the sampler 122 samples the received signal at a frequency that is different from the transmitter sample rate. This difference in sample rate creates a sampling frequency offset that is detrimental to receiver performance. Network 120 corrects the frequency offset so that the receivers are properly synchronized. [0033] The network 120 includes a time domain windowing module 124 that multiplies the sample time domain data received from the sampler 122 by a window function such as a humming window or a humming window. Applying a window function to the sample data offers two benefits. The first is that when the sample data is analyzed in the frequency domain, that is, when the sample data is processed by the FFT module 126 and analyzed by the error calculation module 128, the main lobe of each pilot bin expands or expands. The expansion of the main lobe increases the number of frequency domain samples for each pilot bin, as discussed in more detail below. Second, the pilot bin's secondary lobe is narrowed, resulting in less interference with adjacent pilot bins. [0034] The time domain windowing module 124 passes the sample time domain data to the FFT module 126, which transforms the time domain sample into the frequency domain. As discussed above, the application of the window function in the time domain increases the number of frequency samples for the main lobe of each pilot bin. For example, FIG. 7 shows three frequency domain samples, namely two pilot bins with a main peak and smaller peaks on either side of the main peak, respectively. If the window function is not applied in the time domain, there will be only one frequency domain sample per pilot bin. That is, if the window function is not applied in the time domain, only the main peak exists in the frequency domain. As discussed in more detail below, frequency domain samples must be added to generate frequency offset errors. [0035] The FFT module 126 passes the frequency domain sample to the error calculation module 128. The error calculation module 128 compares the received sample with the stored pilot bin index table 130 of the known training column in order to detect the occurrence of the main peak frequency domain sample of the training column. When a main peak frequency domain sample in the training column is detected, the error calculator 128 analyzes the frequency bins adjacent to each pilot bin in the training column to determine if a sampling frequency offset is present. In the absence of a sampling frequency offset, the left and right frequency bins of the pilot bin will have the same amplitude as shown in FIG. However, if there is a sampling frequency offset, the left and right frequency bins of the pilot bin will have different amplitudes as shown in FIG. [0036] When the calculation module 128 detects a sampling frequency offset error, it calculates an error value for each pilot bin by calculating the amplitude difference between the frequency bins on both sides of each pilot bin. The difference in amplitude is either positive or negative, as shown in Figure 5, and the sign of the difference contains information about whether the receiver's sampling frequency is too fast or too slow. The sign of this difference must be maintained. For a given sampling frequency offset, the amplitude of the sample to the left of the main peak (the amplitude of this sample is indicated by LEFT) to the value of the sample to the right of the main peak (the amplitude of this sample is RIGHT). The value obtained by subtracting (shown) produces a value having a given sign for the first half of the FFT output spectrum, and the sign of the difference is inverted in the second half of the FFT output spectrum. become. To give the error terms the same sign throughout the spectrum, the definition of the error for the first half 136 of the spectrum (ie the positive frequency bin from the FFT) contradicts the error for the second half 138. Must be defined as what to do. That is, the error for the first half can be defined as LEFT-RIGHT, while the error for the second half (ie, the negative frequency bin from the FFT) can be defined as RIGHT-LEFT. Alternatively, as conflicting definitions, (RIGHT-LEFT for the first half and LEFT-RIGHT for the second half) can also be used. [0037] Returning to FIG. 6, the error calculation module 128 outputs the calculation error value for the pilot bin to the average error module 132. The average error module 132 calculates the average error value by taking the average of the calculated error values. When the average error module 132 calculates the average error value, it outputs the average error value to the gain module 134. The gain module 134 generates a scaled error signal that is fed back to the sampler 122 to adjust the sampling frequency by multiplying the average error value by the gain coefficient. The gain value controls the convergence rate of the sampling frequency of the OFDM receiver. Therefore, the gain module 134 functions as a primary loop filter. [0038] Therefore, according to the principles of the present invention, there is provided a method for correcting the sampling frequency offset of the OFDM receiver. Methods for correcting the sampling frequency offset include the step of receiving the OFDM signal containing the training symbol on the OFDM pilot subcarrier, the step of sampling the received OFDM signal at the sampling frequency, and the main lobe of the OFDM pilot subcarrier. So, the step of applying a time domain window to the sample OFDM signal, the step of taking a fast Fourier transform (FFT) of the windowed OFDM signal so that there are multiple frequency domain samples of the OFDM pilot subcarrier, and of the OFDM pilot subcarrier. It includes the step of generating an error proportional to the sampling frequency offset by analyzing multiple frequency domain samples. [0039] According to the features of the present invention, there is provided an OFDM receiver for receiving an OFDM signal having a training symbol on a pilot subcarrier. The OFDM receiver is an analog-to-digital converter (ADC) that samples the received analog OFDM signal at the sampling frequency to generate a digital OFDM sample, and a window function that extends the main lobe of the pilot subcarrier to the digital sample output from the ADC. Sampling by analyzing multiple frequency samples of the pilot subcarrier, FFT that fast Fourier transforms the windowed sample output from the window module so that there are multiple frequency samples for the pilot subcarrier. It has an error calculation module that calculates frequency errors. [0040] Although the present invention has been described above in the light of preferred embodiments, various modifications may be made to these embodiments without departing from the spirit and scope of the invention as defined by each claim of the claims. It is clear that you can. [Simple explanation of drawings] FIG. 1 is a block diagram of a conventional OFDM transmitter. FIG. 2 shows an OFDM signal within the OFDM channel bandwidth, showing the positioning of the OFDM subcarriers and their modulation spectrum in the frequency domain. FIG. 3 is a block diagram of a conventional OFDM receiver. FIG. 4 shows a typical array of OFDM symbols and their corresponding guard intervals within a data frame. FIG. 5 shows 64 FFT output bins in a training column in the presence of sampling frequency offsets. FIG. 6 is a block diagram of an exemplary sampling frequency correction network according to the present invention. FIG. 7 is a diagram showing an FFT output bin when there is no sampling frequency offset. FIG. 8 is a diagram showing an FFT output bin in the presence of a sampling frequency offset.