JP4640870B2 - Receiver - Google Patents

Description

  The present invention relates to a receiving apparatus, and more particularly to a receiving apparatus that receives an orthogonal frequency division multiplexed (OFDM) signal and performs an FFT operation on the received signal to demodulate transmission data.

As a next-generation mobile communication system, a multicarrier modulation system has attracted attention. By using the multi-carrier modulation scheme, not only can broadband high-speed data transmission be realized, but also the influence of frequency selective fading can be reduced by narrowing each subcarrier. Further, by using an orthogonal frequency division multiplexing (Orthogonal Frequency Division Multiplexing) scheme, not only the frequency use efficiency can be improved, but also by providing a guard interval for each OFDM symbol, the influence of intersymbol interference can be eliminated.
FIG. 12A is an explanatory diagram of the multi-carrier transmission system. The serial / parallel conversion unit 1 converts serial data into parallel data and inputs the parallel data to the orthogonal modulation units 3a to 3d via the low-pass filters 2a to 2d. In the figure, it is converted into parallel data consisting of 4 symbols. Each symbol includes an in-phase component (In-Phase component) and a quadrature component (Quadrature component). The quadrature modulation units 3a to 3d perform quadrature modulation of each symbol with subcarriers having frequencies f _{1 to} f ₄ shown in FIG. 12B, the synthesis unit 4 synthesizes the respective quadrature modulation signals, and a transmission unit (not shown) The signal is upconverted to a high frequency signal and transmitted. In the multicarrier transmission method, in order to satisfy the orthogonality between subcarriers, the frequencies are arranged as shown in (b) so that the spectra do not overlap.
In the orthogonal frequency division multiplexing system, the frequency interval is set so that the correlation between the modulated waveband signal transmitted by the nth subcarrier of the multicarrier transmission and the modulated waveband signal transmitted by the (n + 1) th subcarrier becomes zero. Is placed. FIG. 13A is a configuration diagram of a transmission apparatus using an orthogonal frequency division multiplexing system, and the serial / parallel conversion unit 5 converts serial data into parallel data including a plurality of symbols (I + jQ, complex numbers). An IDFT (Inverse Discrete Fourier Transform) 6 performs inverse discrete Fourier transform on frequency data and converts it to time data on the assumption that each symbol is transmitted by a subcarrier having a frequency of the interval shown in FIG. Are input to the quadrature modulation unit 8 through the low-pass filters 7a and 7b. The quadrature modulation unit 8 subjects the input data to quadrature modulation, and a transmission unit (not shown) upconverts the modulated signal to a high-frequency signal and transmits it. According to the orthogonal frequency division multiplexing method, the frequency arrangement shown in FIG. 13B is possible, and the frequency use efficiency can be improved.
In recent years, research on multi-carrier CDMA (MC-CDMA) has been actively conducted, and application to next-generation broadband mobile communication is being studied. In MC-CDMA, transmission data is divided into a plurality of subcarriers by performing serial / parallel conversion and frequency domain orthogonal code spreading. Due to frequency selective fading, subcarriers that are spaced apart from each other undergo independent fading. Therefore, by spreading the code-spread subcarrier signal on the frequency axis by frequency interleaving, the despread signal can obtain frequency diversity gain.
Furthermore, an orthogonal frequency / code division multiple access (OFDM / CDMA) system combining OFDM and MC-CDMA is also being studied. This is a method in which frequency use efficiency is improved by orthogonal frequency multiplexing of signals divided into subcarriers by MC-CDMA.
As shown in FIG. 14, the CDMA (Code Division Multiple Access) system multiplies transmission data of a bit period T _s by spreading codes C _{1 to} C _{N of} a chip period Tc by a multiplier 9, and modulates the multiplication result to transmit. To do. By the multiplication can be spread modulated and transmitted to the broadband signal DS 2 / Tc narrowband signal NM of 2 / T _s as shown in FIG. 15. T _s / Tc is a spreading factor, and is the code length N of the spreading code in the example of the figure. This CDMA transmission system has the advantage that the interference signal can be reduced to 1 / N.
The principle of the multi-carrier CDMA system is that N pieces of copy data are created from one transmission data D as shown in FIG. 16, and each code C _{1 to} C _N constituting a spread code (orthogonal code) is individually set to each of the above-mentioned codes. multiplied by the multiplier ₉ 1 to 9 _N to copy data, to multi-carrier transmit the multiplication results _DC 1 to DC _N in N subcarrier frequencies _f 1 ~f _N shown in FIG. 17 (a). The above is a case where 1-symbol data is transmitted by multicarrier. Actually, as will be described later, transmission data is converted into parallel data of M symbols, and the processing shown in FIG. N total multiplication results are transmitted by multicarrier using M × N subcarriers having frequencies f _{1 to} f _NM . Further, an orthogonal frequency / code division multiple access scheme can be realized by using subcarriers having the frequency arrangement shown in FIG.
FIG. 18 is a block diagram of the MC-CDMA transmission side (base station). The data modulation unit 11 modulates user transmission data and converts it into a complex baseband signal (symbol) having an in-phase component and a quadrature component. The time multiplexing unit 12 time-multiplexes pilots of a plurality of symbols before transmission data. The serial / parallel conversion unit 13 converts the input data into parallel data of M symbols, and each symbol is N-branched and input to the spreading unit 14. The spreading unit 14 includes M multiplying units 14 _{1 to} 14 _M , and each of the multiplying units 14 _{1 to} 14 _M includes codes (codes) C ₁ , C ₂ ,. . By multiplying the C _N individually branch symbol output. As a result, the subcarrier signals S _{1 to} S _MN for multicarrier transmission using N × M subcarriers are output from the spreading unit 14. That is, the spreading unit 14 spreads in the frequency direction by multiplying the symbol for each parallel series by the orthogonal code. Different codes (Walsh codes) C ₁ , C ₂ ,. . _CN is shown, but in reality, station identification codes (gold codes) G _{1 to} G _MN are further multiplied by subcarrier signals S _{1 to} S _MN .
The code multiplexing unit 15 code-multiplexes the subcarrier signal generated as described above with the subcarrier signal of another user generated by the same method. That is, the code multiplexing unit 15 synthesizes and outputs subcarrier signals of a plurality of users corresponding to the subcarriers for each subcarrier. The frequency interleaving unit 16 rearranges the code-multiplexed subcarrier signals by frequency interleaving and distributes them on the frequency axis in order to obtain a frequency diversity gain. An IFFT (Inverse Fast Fourier Transform) unit 17 performs IFFT (Inverse Fourier Transform) processing on the subcarrier signals input in parallel to convert them into OFDM signals (real part signals and imaginary part signals) on the time axis. The guard interval insertion unit 18 inserts a guard interval into the OFDM signal, the orthogonal modulation unit performs orthogonal modulation on the OFDM signal into which the guard interval is inserted, and the radio transmission unit 20 up-converts to a radio frequency and amplifies it at a high frequency. Transmit from antenna.
The total number of subcarriers is (spreading factor N) × (number of parallel sequences M). Since the propagation path undergoes different fading for each subcarrier, the pilot is time-multiplexed on all the subcarriers, and fading compensation can be performed for each subcarrier on the receiving side. Here, the time-multiplexed pilot is a common pilot that all users use for channel estimation.
FIG. 19 is an explanatory diagram of serial-parallel conversion, in which a common pilot P is time-multiplexed ahead of one frame of transmission data. As will be described later, the common pilot P can be dispersed within the frame. If the common pilot per frame is, for example, 4 × M symbols and the transmission data is 28 × M symbols, the serial / parallel converter 13 outputs pilot M symbols up to the first four times as parallel data. As a result, M symbols of the transmission data are output 28 times. As a result, the pilot can be time-multiplexed on all subcarriers and transmitted four times in one frame period, and channel compensation (fading compensation) can be performed by estimating the channel for each subcarrier on the receiving side. Become.
FIG. 20 is an explanatory diagram of guard interval insertion. The guard interval insertion is to copy the tail part to the head part of the IFFT output signal corresponding to M × N subcarrier samples (= 1 OFDM symbol) as one unit. By inserting the guard interval GI, it is possible to eliminate the influence of intersymbol interference due to multipath.
FIG. 21 is a block diagram of the MC-CDMA receiving side. The wireless reception unit 21 performs frequency conversion processing on the received multicarrier signal, and the orthogonal demodulation unit performs orthogonal demodulation processing on the received signal. The timing synchronization / guard interval removal unit 23 obtains timing synchronization of the received signal, removes the guard interval GI from the received signal, and inputs the guard interval GI to an FFT (Fast Fourier Transform) unit 24. The FFT unit 24 performs FFT calculation processing at the FFT window timing to convert the time domain signal into N × M subcarrier signals (subcarrier samples), and the frequency deinterleave unit 25 performs the reverse permutation on the transmission side. Output in the order of subcarrier frequencies.
After deinterleaving, the channel compensation unit 26 performs channel estimation for each subcarrier using a pilot time-multiplexed on the transmission side, and compensates for fading. Although the channel estimation unit 26a ₁ for only one subcarrier is shown in figure, the channel estimator is provided for each sub-carrier. That is, channel estimation section 26a ₁ estimates the phase effect exp (jφ) due to fading using the pilot signal, and multiplier 26b ₁ multiplies the subcarrier signal of the transmission symbol by exp (−jφ) to perform fading. To compensate.
The despreading unit 27 includes M multiplying units 27 _{1 to} 27 _M , and the multiplying unit 27 ₁ configures the codes C ₁ , C ₂ ,... Constituting the orthogonal code (Walsh code) assigned to the user. . . _CN is individually multiplied by N subcarriers and output, and the other multiplication units perform the same arithmetic processing. As a result, the fading-compensated signal is despread by the spreading code assigned to each user, and the signal of the desired user is extracted from the signals multiplexed by this despreading. In practice, the station identification code (Gold code) is multiplied before the Walsh code is multiplied.
The combining units 28 _{1 to} 28 _M add N multiplication results output from the multiplying units 27 _{1 to} 27 _M, respectively, to create parallel data including M symbols, and the parallel-serial conversion unit 29 converts the parallel data into the parallel data. The data demodulator 30 demodulates the transmission data by converting the serial data.
FIG. 22 is a diagram for explaining the operation of the channel estimator. Four pilot symbols (4 OFDM pilot symbols) are dispersed and multiplexed in one frame composed of 32 OFDM symbols. Since one pilot symbol is composed of subcarrier samples of the number of subcarriers (M × N, for example, 1024), the channel (amplitude) for each subcarrier is monitored by monitoring the FFT output at the pilot reception timing on the receiving side. Characteristics and phase characteristics) can be estimated. That is, the channel estimation is performed by collecting four subcarrier samples in the frequency direction and collecting four subcarrier samples in the time direction to form one group with a total of 32 subcarrier samples as shown by PG1 in FIG. The channel value of the central subcarrier (amplitude and phase of the received pilot signal) is set as the value, and this channel value is compared with the known channel value (amplitude and phase of the known pilot signal) to estimate the channel of the subcarrier. . The channel estimation value of the next subcarrier is composed of 8 subcarrier samples shifted by 1 subcarrier in the frequency direction as shown by PG2, and 4 groups in the time direction constitute one group with a total of 32 subcarrier samples. And it calculates similarly using the average value in this group PG2. The reason why the channel values are obtained by averaging as described above is that noise is added to each symbol, and averaging eliminates the influence of the noise and improves the S / N ratio. If the subcarriers are close in frequency, the channel values are almost the same, so there is no problem even if they are averaged.
FIG. 23 is an explanatory diagram of FFT window timing in the case of two paths (two waves), where A is a direct wave and B is a delayed wave (reflected wave). If the FFT window is determined to be one OFDM period width Wa from the head of the OFDM symbol D1 of the direct wave, the OFDM symbol D1 of the direct wave only overlaps the guard interval GI1 of the delayed wave and the D1 symbol portion in the FFT window width Wa. Therefore, there is no intersymbol interference (Inter-Symbol Interference, ISI) due to multipath delay waves. However, if the FFT window is determined to be one OFDM period width Wb from the beginning of the OFDM symbol D1 of the delayed wave, the D1 symbol portion of the delayed wave and the guard interval GI2 of the D2 symbol of the direct wave overlap and receive intersymbol interference ISI. The FFT window should be determined so as not to receive intersymbol interference ISI, and if the FFT window is determined to be Wa in FIG. Absent.
By the way, if the Fourier transform of the function f (t) is expressed as F [f (t)], the Fourier transform of the time delay function f (t−t0) becomes exp (−2πjft0) F [f (t)]. . When exp (−2πjft0) = cos2πft0−jsin2πft0 is considered, the Fourier transform of the time delay function f (t−t0) rotates according to the change in frequency. For example, as shown in FIG. 24A, the Fourier transform of impulse δ (t) at time t = 0 is 1 (= constant) at an arbitrary frequency, but the Fourier transform of δ (t−t0) is FIG. As shown in (b), it rotates according to the frequency so that the projection onto the IjQ complex plane becomes a unit circle.
As described above, when the received power of the direct wave is large and the received power of the delayed wave is small as shown in FIG. Is large and the rotation amplitude is small. Since the fluctuation portion is small in this way, a correct channel value can be obtained by averaging 32 subcarrier samples, and a channel can be estimated correctly. Also, it does not receive intersymbol interference ISI.
However, when the received power of the direct wave is small and the received power of the delayed wave is large, as shown in FIG. Smaller and larger rotational amplitude. When the fluctuation portion becomes large in this way, an error is included in the channel value obtained by averaging 32 subcarrier samples, and the channel cannot be estimated correctly.

An object of the present invention is to correctly perform channel estimation even when the received power of a delayed wave is larger than the received power of a direct wave.
Another object of the present invention is to prevent inter-symbol interference ISI even when the received power of the delayed wave is larger than the received power of the direct wave, and to perform channel estimation correctly. .

The present invention extracts a signal portion corresponding to a predetermined FFT window size from a received signal including a preceding wave and a delayed wave, and performs an FFT process on the signal portion to transmit a transmission signal to which an orthogonal frequency division multiplexing system is applied. A receiving device that performs reception, a signal cutout unit that cuts out a signal portion corresponding to the FFT window size from the leading timing of the preceding wave excluding the guard interval, and a detecting unit that detects the leading timing of the delayed wave excluding the guard interval And a shift unit that shifts a part of the signal portion cut out by the signal cutout unit from the leading timing of the preceding wave to the end of the cutout signal portion from the leading timing of the delayed wave, and the shift An FFT processing unit that performs FFT processing on the signal portion obtained by shifting by the unit, That.

  According to the present invention, intersymbol interference ISI can be eliminated, the average value of FFT conversion can be increased, the rotation amplitude can be decreased, and channel estimation can be performed correctly.

FIG. 1 is a diagram illustrating the principle of the present invention. FIG. 2 is a block diagram showing a main part of the OFDM receiving apparatus according to the present invention. FIG. 3 is a partial detail view of the OFDM receiver. FIG. 4 is an explanatory diagram of a delay profile. FIG. 5 is a block diagram of the FFT window timing recovery unit. FIG. 6 is a diagram for explaining the operation of the FFT window timing recovery unit. FIG. 7 shows various embodiments of the shift timing detection unit in the FFT calculation position shift unit. FIG. 8 is a configuration diagram of a shift timing detection unit including a footing part. FIG. 9 is a block diagram of a shift timing detection unit provided with a sufficient level determination unit in addition to the configuration of FIG. FIG. 10 is an explanatory diagram for explaining why the shift operation is necessary. FIG. 11 is a block diagram of a shift timing detection unit having means for shifting the delay profile. FIG. 12 is an explanatory diagram of a conventional multicarrier transmission system. FIG. 13 is an explanatory diagram of a conventional orthogonal frequency division multiplexing system. FIG. 14 is an explanatory diagram of CDMA code spread modulation. FIG. 15 is an explanatory diagram of band spreading in CDMA. FIG. 16 is a diagram for explaining the principle of the multicarrier CDMA system. FIG. 17 is an explanatory diagram of subcarrier arrangement. FIG. 18 is a block diagram of the conventional MC-CDMA transmission side. FIG. 19 is an explanatory diagram of serial-parallel conversion. FIG. 20 is an explanatory diagram of a guard interval. FIG. 21 is a configuration diagram on the receiving side of the conventional MC-CDMA. FIG. 22 is an explanatory diagram of the operation of the channel estimation unit. FIG. 23 is an explanatory diagram of FFT window timing in the case of two paths (two waves). FIG. 24 is an explanatory diagram of Fourier transform. FIG. 25 is an explanatory diagram of FFT conversion when the received power of the direct wave is large and the received power of the delayed wave is small. FIG. 26 is an explanatory diagram of FFT conversion when the received power of the direct wave is small and the received power of the delayed wave is large.

により求め、このアドレスｉよりデータを読み出してＦＦＴ演算部５３に入力する。ただし、バッファメモリ５２ａにはＯＦＤＭシンボル長と同じＮ個の複素データｘ_ｉが蓄えられ、ｉ＝０，…，（Ｎ−１）であり、Ｓ＝シフト量である。又、ｍｏｄはモジュロ演算を現し、（ｊ＋Ｓ）をＮで割った時の余りに相当する。
すなわち、タイミングシフト部５２ｃはバッファメモリ５２ａからの読み出し時、ｊを０から（Ｎ−１）まで変化させながら、式（１）で計算したアドレスｉのサンプルデータｘ_ｉを順次読み出す。これにより、Ｓサンプル分シフトさせたサンプルデータをＦＦＴ演算部５３に入力し、かつ、先頭Ｓサンプル分のデータを最後尾にシフトすることができる。
（ｄ）動作
以上より、タイミング同期／ＯＦＤＭシンボル取り出し部５１は、ＦＦＴウインドウタイミング、理想的には直接波のＯＦＤＭシンボルの先頭時刻を基準にして、入力データ列よりＮサンプルで構成された１ＯＦＤＭシンボルを取り出してＦＦＴ演算位置シフト部５２に入力する。このＯＦＤＭシンボルを用いてＦＦＴ演算することによりシンボル間干渉ＩＳＩの発生を防止できる。
ＦＦＴ演算位置シフト部５２は、入力されたＮサンプル分の１ＯＦＤＭシンボルデータをバッファメモリ５２ａに格納すると共に、ＩＦＦＴ演算部５６から入力する遅延プロファイルに基づいてＦＦＴウインドウタイミングから受信電力最大の波までの遅延時間をサンプル数で換算したシフト量Ｓを求め、（１）式が示すバッファメモリ５２ａのアドレスｉから順次サンプルデータを読み出してＦＦＴ演算部５３に入力する。ＦＦＴ演算部５３はＦＦＴ演算位置シフト部５２から入力するＳサンプル分シフトしたＯＦＤＭシンボルデータにＦＦＴ演算処理を施し、サブキャリア数Ｎの信号ＳＣ_０〜ＳＣ_Ｎ−１を出力する。Ｓサンプル分シフトさせることにより、ＦＦＴ変換の回転変動分を小さくできる。
チャネル推定部５４はサブキャリア毎にチャネルを推定し、チャネル補償部５５はサブキャリア毎のチャネル推定値Ｃ_０〜Ｃ_Ｎ−１をＦＦＴ演算部５３から出力するＮ個のサブキャリア信号ＳＣ_０〜ＳＣ_Ｎ−１に乗算してチャネル補償を施す。又、ＩＦＦＴ演算部５６はチャネル推定部５４から出力するサブキャリア数Ｎのチャネル推定値Ｃ_０〜Ｃ_Ｎ−１にＩＦＦＴ演算を施し、遅延プロファイルを出力する。
以後、上記動作をＯＦＤＭシンボル毎に行うことにより、シンボル間干渉ＩＳＩの発生を防止でき、且つ、ＦＦＴ出力の回転変動部の振幅を小さくでき、正確にチャネルを推定することが可能になる。
（Ｃ）シフトタイミング検出部の実施例
図７はＦＦＴ演算位置シフト部５２におけるシフトタイミング検出部５２ｂの各種実施例説明図である。尚、以下の説明において、ｓは求めるシフト量、Ｎはサブキャリア数、ｃ_ｉはｉ番目のサブキャリアのチャネル推定値（複素数値）、ｐ_ｉはチャネル推定値ｃ_ｉをＩＦＦＴして得られる遅延プロファイル（複素数値）であり、ｉ＝０，…，（Ｎ−１）である。
（ａ）最大電力に基づいた算出法
図７（ａ）は電力｜ｐ_ｉ｜^２を最大にするｉをシフト量ｓとする実施例である。
電力計算部７１は遅延プロファイルの各サンプル値ｐ_ｉの電力｜ｐ_ｉ｜^２を計算し、最大電力検出部７２はＮ個のサンプルｐ_ｉ（ｉ＝０，…，（Ｎ−１））の電力のうち最大電力ｐ_ｓを検出し、シフト量決定部７３は該最大電力ｐ_ｓを示すサンプル位置ｓをシフト量として決定する。すなわち、任意のｐ_ｉに対して｜ｐ_ｓ｜^２≧｜ｐ_ｉ｜^２となるシフト量ｓを求める。
（ｂ）電力重み付けに基づいた算出法
図７（ｂ）は電力｜ｐ_ｉ｜^２で重み付け平均したｉをシフト量とする実施例であり、次式

で求まるｓをシフト量とする。すなわち、電力計算部７１は遅延プロファイルの各サンプル値ｐ_ｉの電力｜ｐ_ｉ｜^２を計算し、第１演算部７４ａは（２）式の分子Ｐ_１を算出する。分子Ｐ_１は電力｜ｐ_ｉ｜^２で重み付けしたｉの合計値である。又、第２演算部７４ｂは（２）式の分母Ｐ_２、すなわち電力の総和を演算し、シフト量決定部７５はＰ_１／Ｐ_２により電力｜ｐ_ｉ｜^２で重み付けしたｉの重み付け平均を演算し、得られた平均値をシフト量ｓとして出力する。ただし、このままではｓは実数となるので、シフト量決定部７５は四捨五入や切捨てなどにより整数に丸める。
（ｃ）振幅重み付けに基づいた算出法
図７（ｃ）は振幅｜ｐ_ｉ｜で重み付け平均したｉをシフト量とする実施例であり、次式

で求まるｓをシフト量とする。すなわち、絶対値化部７６は遅延プロファイルの各サンプル値ｐ_ｉの振幅｜ｐ_ｉ｜を計算し、第１演算部７７ａは（３）式の分子Ａ_１を算出する。分子Ａ_１は振幅｜ｐ_ｉ｜で重み付けしたｉの合計値である。又、第２演算部７７ｂは（３）式の分母Ａ_２、すなわち振幅の総和を演算し、シフト量決定部７８はＡ_１／Ａ_２により振幅｜ｐ_ｉ｜で重み付けしたｉの重み付け平均を演算し、得られた平均値をシフト量ｓとして出力する。ただし、このままではｓは実数となるので、シフト量決定部７８は四捨五入や切捨てなどにより整数に丸める。
（ｄ）重み付け関数を用いた重み付け平均に基づいた算出法
図７（ｄ）はｐ_ｉを引数とする重み付け関数ｆ（）で重み付け平均したｉをシフト量とする実施例であり、次式

で求まるｓをシフト量とする。すなわち、重み付け関数算出部７９は重み付け関数値ｆ（ｐ_ｉ）を算出し、第１演算部８０ａは（４）式の分子Ｆ_１を算出する。分子Ｆ_１は重み付け関数値ｆ（ｐ_ｉ）で重み付けしたｉの合計値である。又、第２演算部８０ｂは（４）式の分母Ｆ_２、すなわち重み付け関数値ｆ（ｐ_ｉ）の総和を演算し、シフト量決定部８１はＦ_１／Ｆ_２により重み付け関数値ｆ（ｐ_ｉ）で重み付けしたｉの重み付け平均を演算し、得られた平均値をシフト量ｓとして出力する。ただし、このままではｓは実数となるので、シフト量決定部８１は四捨五入や切捨てなどにより整数に丸める。
この一般化した方式は、以下の（５），（６）式のように重み付け関数ｆ（）を定めると上記（ａ）の最大電力に基づいた算出法と同じになる。
ｆ（ｐi）＝１ （ｉ＝ｓのとき） （5）
ｆ（ｐi）＝０ （ｉ≠ｓのとき） （6）
また、以下の（７）式のように重み付け関数ｆ（）を定めると上記（ｂ）の電力重み付けに基づいた算出法と同じになる。
ｆ（ｐi）＝｜ｐi｜^２ (7)
更に、以下の（８）式のように重み付け関数ｆ（）を定めると上記（ｃ）の振幅重み付けに基づいた算出法と同じになる。
ｆ（ｐi）＝｜ｐi｜ (8)
（ｅ）足きりによるシフト量の算出法
遅延プロファイル（複素数値）ｐ_ｉは足切りをしてからシフト量の検出に使用する。このようにすれば雑音を除去でき、シフト量の計算精度を向上できる。図８は足きり部９１を備えたシフトタイミング検出部の構成図であり、図７（ａ）〜（ｄ）のいずれかのシフトタイミング検出部９２の前段に足きり部９１を設けた構成になっている。足きり部９１は適当なしきい値Ｔを決め、Ｔよりも小さければｑ_ｉ＝０、大きければｑ_ｉ＝ｐ_ｉとする。すなわち、
ｑi＝ｐi (｜ｐi｜^２≧Tのとき) （9）
ｑi＝０ (｜ｐi｜^２＜Tのとき) （10）
にしたがってｑ_ｉを決定し、シフトタイミング検出部９２はこのｑ_ｉを用いて上記（ａ）〜（ｄ）と同じ操作をしてシフト量ｓを求める。
足切りレベルＴの決定法は各種考えられるが、例として、次式

に従って、｜ｐ_ｉ｜^２の平均値の４倍を足切りレベルＴとする。あるいは、次式
T=|Pmax|²／４ （|Pmax|²はｐiの最大値） (12)
にしたがって、｜ｐ_ｉ｜^２の最大値の４分の１を足切りレベルＴとする。図９は図８の構成に加えて足きりレベル決定部９３を備えたシフトタイミング検出部の構成図であり、足きりレベル決定部９３は（１１）または（１２）式に従って足切りレベルＴを決定して足きり部９１に入力する。足きり部９１はしきい値Ｔよりも小さければｑ_ｉ＝０、大きければｑ_ｉ＝ｐ_ｉとし、シフトタイミング検出部９２はこのｑｉを用いて上記（ａ）〜（ｄ）と同じ操作をしてシフト量ｓを求める。
（ｆ）足きりの変形例
足切り方式における（９），（１０）式を以下のように、
ｑi＝１ (｜ｐi｜^２≧Tのとき) （13）
ｑi＝０ (｜ｐi｜^２＜Tのとき) （14）
変形し、得られたｑ_ｉを用いて次式

に従ってシフト量ｓを求める。このように変形して足切りすると、マルチパスの単純な平均遅延時間（シフト量）が求まる。例えば２パスの場合において、それぞれの遅延プロファイルのタイミングをｔ_１とｔ_２とすると、変形方式によればパスの大小関係に関わらず、２パスの丁度まん中のタイミング（ｔ_１＋ｔ_２）／２がパスの平均遅延時間（シフト量ｓ）として得られる。
（ｇ）遅延プロファイルのシフト操作
ＦＦＴウインドウタイミングよりも前にパスが存在すると、すなわち、ＦＦＴウインドウタイミングよりも前に到来する波が存在すると、該パスに応じた遅延プロファイルは図４（ｂ）に示すように後尾に現れる。かかる場合、正しい位置に遅延プロファイルが存在しないため、正確にシフト量を検出することができなくなる。図１０に従って説明する。例えば図１０（ａ）に示すような２パスのとき、ＦＦＴウインドウタイミングの開始タイミングＷ_ｒが２つのパスＡ，Ｂのちょうど中間に有った場合、パスＡ，Ｂの遅延プロファイルＰ_Ａ，Ｐ_Ｂは図１０（ｂ）の実線位置に現れる。このため、重み付けによりシフト量ｓを求めるとｓは図１０（ｂ）の様に遅延プロファイル範囲の中央付近になる。所望のシフト量ｓは０付近なので、これは好ましくない。そこで、図１０（ｃ）に示すように、予めｓｐだけパスＡ，Ｂの遅延プロファイルＰ_Ａ，Ｐ_ＢをシフトさせてＰ_Ａ′，Ｐ_Ｂ′としてからｓを求め、その後で、ｓ＝ｓ−ｓ_ｐとして元に戻すことで正確なシフト量ｓを求める。
図１１は遅延プロファイルをシフトする手段を備えたシフトタイミング検出部の構成図である。遅延プロファイルシフト部９５は遅延プロファイルｐ_ｉ（ｉ＝０，…，（Ｎ−１））をｓ_ｐだけ次式
ｑ_ｉ＝ｐ_ｋ
但し、ｋ＝(ｉ＋ｓ_p)mod N (i=0,……,(N-1)) (16)
によりシフトしてｑ_ｉとし、しかる後、図７（ａ）〜（ｄ）及び図８〜図９のいずれかのシフトタイミング検出部９２は、このｑ_ｉを用いて仮のシフト量ｓを求める。仮のシフト量ｓが求まった後、シフト量補正部９６は次式
ｓ＝ｓ−ｓ_p (17)
によりシフト量ｓからｓｐを減じて本来のｓを算出して出力する。シフト量ｓ_ｐは遅延プロファイルの広がり（最大遅延時間）よりも十分に大きければよいので、カードインターバル長の２倍程度でよいと思われる。
以上本発明によれば、シンボル間干渉ＩＳＩをなくせ、かつ、ＦＦＴ変換の平均値を大きく、回転振幅を小さくでき、チャネル推定を正しく行うことができる。 (A) Principle of the Present Invention As shown in FIG. 1A, when the received power of the direct wave A is small and the received power of the delayed wave B is large, the FFT window Wa is set based on the OFDM symbol D0 of the direct wave. If determined, there is no inter-symbol interference ISI, but the average value of the FFT transform is small and the rotation amplitude is large as shown in FIG. On the other hand, when the FFT window Wb is determined with reference to the OFDM symbol D0 of the delayed wave, the average value of FFT conversion is large and the rotation amplitude is small as shown in FIG. 1C, but intersymbol interference ISI occurs.
Therefore, in the present invention, one OFDM symbol is taken from the input data sequence with reference to the head (time Ta) of the OFDM symbol D0 of the direct wave, and the head (time Tb) of the OFDM symbol D0 of the delayed wave is set as the FFT calculation start position. . Specifically, the data between the times Ta and Tb is shifted later to perform the FFT operation. That is, as shown in FIG. 1 (d), the direct wave portion D01 and the delayed wave portion GI0 ′ between Ta and Tb are shifted to the tail, and the FFT operation is performed on the shifted 1 OFDM symbol from the beginning. Even if it is shifted to the end, data continuity is maintained and no problem occurs. In this way, intersymbol interference ISI can be eliminated, the average value of FFT conversion can be increased, the rotation amplitude can be reduced, and channel estimation can be performed correctly.
The above is a case where the received power of the delayed wave is large, but the received power of the direct wave may be large. Therefore, the time Tb is the start time of the OFDM symbol of the wave (direct wave or delay wave) with the maximum received power, and this start time is determined from the delay profile obtained by IFFT of the channel estimation value. it can.
In order to perform the FFT operation by shifting as described above, one OFDM symbol is fetched from the input data string at the time Ta and stored in the buffer, and the data is sequentially read from the buffer storage position corresponding to the time Tb. When the data is input to the arithmetic unit and read to the end of the buffer, the data is read from the beginning of the buffer to the buffer storage position corresponding to time Tb and input to the FFT arithmetic unit.
(B) OFDM Receiver (a) Overall Configuration FIG. 2 is a block diagram of the main part of the OFDM receiver of the present invention, and FIG. 3 is a partial detailed view of the OFDM receiver. The timing synchronization / OFDM symbol extracting unit 51 detects the leading timing Ta (see FIG. 1) of the OFDM symbol of the direct wave, and extracts and outputs one OFDM symbol that does not include a guard interval based on the timing. The timing synchronization / OFDM symbol extraction unit 51 corresponds to the timing synchronization / guard interval removal unit 23 in FIG.
The FFT calculation position shift unit 52 determines the delay time (Tb-Ta) from the direct wave of the wave having the maximum received power (actually, the delay time from the FFT window start timing) based on the delay profile described later. Then, the FFT calculation start position for the OFDM symbol extracted by the timing synchronization / OFDM symbol extraction unit 51 is shifted by (Tb−Ta), and the leading (Tb−Ta) portion is shifted to the end.
The FFT operation unit 53 performs FFT operation processing on the OFDM symbol data (time domain signal) input from the FFT operation position shift unit 52, and converts the signals SC _{0 to} SC _N-1 (frequency domain signals) having N subcarriers. Convert. The channel estimation unit 54 estimates a channel for each subcarrier by the method described in FIG. 22, and outputs channel estimation values C _{0 to} C _N−1 . Each multiplier unit ₅₅ 0 _{to 55 N-1} of the channel compensation unit 55, a channel estimation value for each subcarrier _C 0 _-C N subcarrier signal _SC the _N-1 outputs from the FFT computation block 53 0 to SC _{N The} channel compensation (fading compensation) is performed by multiplying by ₋₁ , and the N subcarrier signals subjected to the channel compensation are input to a subsequent despreading unit (not shown).
The IFFT calculation unit 56 performs an IFFT (Inverse Fast Fourier Transform) calculation on the channel estimation values C _{0 to} C _N−1 of the number of subcarriers N output from the channel estimation unit 54, and N per symbol period shown in FIG. Output a delay profile consisting of samples. Each sample indicates the strength of each wave of the direct wave and the delayed wave, and each sample value of the delay profile exceeding the maximum delay time M from the FFT window position (= 0) becomes a small value below the set level. A sample value corresponding to a wave arriving before the FFT window timing (= 0) appears at the tail of the delay profile as shown in FIG.
(B) Timing Synchronization / OFDM Symbol Extraction Unit In the timing synchronization / OFDM symbol extraction unit 51, the AD conversion unit 51a converts a baseband signal (complex I component and Q component) output from a quadrature demodulation unit (not shown) into digital data. Then, the FFT window timing recovery unit 51b detects the head position of the direct wave guard interval by correlation calculation and inputs it to the OFDM symbol cutout unit 51c. The OFDM symbol cutout unit 51 c extracts 1 OFDM symbol data from the input data string at the timing Ta when the guard interval period has elapsed from the head position of the guard interval, and inputs the data to the FFT calculation position shift unit 52.
FIG. 5 is a configuration diagram of the FFT window timing recovery unit 51b, and FIG. 6 is an operation explanatory diagram of the FFT window timing recovery unit. As shown in FIG. 6A, the guard interval GI is created by copying the _NG end portion of the sample number N to the beginning portion of the 1 OFDM symbol having the N number of samples. correlation value a guard interval GI moiety as shown in FIG. 6 (b) by calculating the correlation between the current received signal is maximized, it is possible to detect the FFT window timing to time t ₀ to the maximum. That is, the delay unit 61a of the correlation calculation unit 61 delays the received signal by one OFDM symbol (number of samples N), the conjugate complex number calculation unit 61b calculates the complex conjugate of the reception signal delayed by the delay unit 61a, and the multiplication unit 61c. Multiplies the complex conjugate of the received signal one OFDM symbol before output from the conjugate complex number calculation unit 61b and the current received signal, and outputs the multiplication result (correlation value).
The shift register 62 has a number of samples N length of _G component of the guard interval, and stores the multiplication result of the most recent N _G hysterectomy (correlation value), the addition unit 63 adds the N _G number of correlation values output To do. Since the received signal one OFDM symbol before and the current received signal (I + jQ) are ideally the same in the guard interval period, I ² + Q ² is obtained from the multiplier 61c in the guard interval period. Thus, Figure 6 adds value gradually increases as shown in (b) according to the number of correlation values of the guard interval period to be stored in the shift register 62 is increased, all the correlation values N _G number of the guard interval period When the value is stored in the shift register 62, the added value becomes the maximum. Thereafter, the number of correlation values stored in the shift register 62 in the guard interval period decreases, and the added value gradually decreases. Therefore, the head timing of the direct wave guard interval GI can be detected by detecting the peak timing of the added value. However, in the multipath environment, the addition value of the addition unit 63 is the sum CR of the direct wave addition value CR1 (one-dot chain line) and the delayed wave addition value CR2 (dotted line) as shown in FIG. Is shifted by Δt from the peak position of the direct wave.
Peak timing detection unit 64 detects the peak position of the added value, and inputs the OFDM symbol extraction unit 51c the peak position t ₀ as the start position of the guard interval.
The OFDM symbol cutout unit 51 c extracts one OFDM symbol data from the input data string after the guard interval period has elapsed from the head position of the guard interval, and inputs the data to the FFT calculation position shift unit 52.
(C) FFT operation position shift unit The buffer memory 52a stores N samples for one OFDM symbol input from the OFDM symbol cutout unit 51c. The shift timing detection unit 52b has a shift amount S (see FIG. 4) obtained by converting the delay time from the direct wave (FFT window position) of the wave having the maximum received power from the delay profile input from the IFFT calculation unit 56 into the number of samples. Ask. The timing shift unit 52c shifts the FFT calculation start position for the OFDM symbol extracted by the timing synchronization / OFDM symbol extraction unit 51 by the S samples and shifts the first S samples of the OFDM symbol to the end. Specifically, the timing shift unit 52c calculates the address i of the data read out jth from the buffer memory 52a by the following equation.

The data is read from the address i and input to the FFT operation unit 53. However, N complex data x _i having the same length as the OFDM symbol length are stored in the buffer memory 52a, i = 0,..., (N−1), and S = shift amount. Further, mod represents a modulo operation and corresponds to the remainder when (j + S) is divided by N.
That is, when reading from the buffer memory 52a, the timing shift unit 52c sequentially reads the sample data x _i at the address i calculated by the equation (1) while changing j from 0 to (N−1). Thereby, the sample data shifted by S samples can be input to the FFT operation unit 53, and the data for the first S samples can be shifted to the end.
(D) Operation As described above, the timing synchronization / OFDM symbol extraction unit 51 uses one OFDM symbol composed of N samples from the input data sequence with reference to the FFT window timing, ideally, the start time of the direct wave OFDM symbol. Is extracted and input to the FFT calculation position shift unit 52. Generation of intersymbol interference ISI can be prevented by performing an FFT operation using this OFDM symbol.
The FFT calculation position shift unit 52 stores 1 OFDM symbol data for the input N samples in the buffer memory 52a, and based on the delay profile input from the IFFT calculation unit 56, from the FFT window timing to the wave with the maximum received power. A shift amount S obtained by converting the delay time into the number of samples is obtained, and the sample data is sequentially read from the address i of the buffer memory 52a indicated by the equation (1) and input to the FFT operation unit 53. The FFT calculation unit 53 performs FFT calculation processing on the OFDM symbol data shifted by S samples input from the FFT calculation position shift unit 52, and outputs signals SC _{0 to} SC _N−1 having N subcarriers. By shifting by S samples, the rotational variation of the FFT transform can be reduced.
The channel estimation unit 54 estimates a channel for each subcarrier, and the channel compensation unit 55 outputs N subcarrier signals SC ₀ to SC that output channel estimation values C _{0 to} C _N−1 for each subcarrier from the FFT calculation unit 53. Channel compensation is performed by multiplying SC _N-1 . The IFFT calculation unit 56 performs IFFT calculation on the channel estimation values C _{0 to} C _N−1 of the number N of subcarriers output from the channel estimation unit 54, and outputs a delay profile.
Thereafter, by performing the above operation for each OFDM symbol, the occurrence of intersymbol interference ISI can be prevented, the amplitude of the rotation fluctuation part of the FFT output can be reduced, and the channel can be estimated accurately.
(C) Examples of Shift Timing Detection Unit FIG. 7 is an explanatory diagram of various examples of the shift timing detection unit 52 b in the FFT calculation position shift unit 52. In the following description, s is the obtained shift amount, N is the number of subcarriers, c _i is the channel estimation value (complex value) of the i-th subcarrier, and p _i is obtained by IFFT of the channel estimation value c _i. It is a delay profile (complex value), i = 0,..., (N−1).
(A) Calculation based on the maximum power diagram. 7 (a) power | examples which the i that maximizes ² and the shift amount s | p _i.
The power calculation unit 71 calculates the power | p _i | ² of each sample value p _i of the delay profile, and the maximum power detection unit 72 calculates N samples p _i (i = 0,..., (N−1)). detecting the maximum power p _s of the power shift amount determining unit 73 determines the sample position s indicating the said maximum power p _s as a shift amount. That is, for any _{p i |} Request ² become the shift amount ^{_{s | p s | 2 ≧ |}} p i.
(B) Calculation Method Based on Power Weighting FIG. 7B is an embodiment in which i that is weighted averaged by power | p _i | ² is used as a shift amount.

S obtained by the above is used as a shift amount. That is, the power calculation unit 71 calculates the power | p _i | ² of each sample value p _i of the delay profile, and the first calculation unit 74a calculates the numerator P ₁ of the equation (2). The numerator P ₁ is the total value of i weighted by power | p _i | ² . The second calculation unit 74b calculates the denominator P _{2 of} equation (2), that is, the total sum of power, and the shift amount determination unit 75 calculates the weighted average of i weighted by power | p _i | ² by P ₁ / P _2. And the average value obtained is output as the shift amount s. However, since s becomes a real number as it is, the shift amount determination unit 75 rounds it to an integer by rounding off or truncating.
(C) Calculation Method Based on Amplitude Weighting FIG. 7C is an example in which i that is weighted and averaged by amplitude | p _i | is used as a shift amount.

S obtained by the above is used as a shift amount. That is, the absolute value conversion unit 76 calculates the amplitude | p _i | of each sample value p _i of the delay profile, and the first calculation unit 77a calculates the numerator A ₁ of the equation (3). The numerator A ₁ is the total value of i weighted by the amplitude | p _i |. The second calculation unit 77b calculates the denominator A _{2 of} equation (3), that is, the sum of the amplitudes, and the shift amount determination unit 78 calculates the weighted average of i weighted by the amplitude | p _i | by A ₁ / A _2. The calculated average value is output as the shift amount s. However, since s is a real number as it is, the shift amount determination unit 78 rounds it to an integer by rounding off or truncating.
(D) weighting calculation method based on a weighted average using the function diagram. 7 (d) are examples of a shift amount i weighted average weighting function f as an argument the p _i (), the following equation

S obtained by the above is used as a shift amount. That is, the weighting function calculation unit 79 calculates the weighting function value f (p _i ), and the first calculation unit 80a calculates the numerator F ₁ of the formula (4). The numerator F ₁ is the total value of i weighted by the weighting function value f (p _i ). In addition, the second calculation unit 80b calculates the denominator F _{2 of} equation (4), that is, the sum of the weighting function values f (p _i ), and the shift amount determination unit 81 calculates the weighting function value f (p) by F ₁ / F _2. The weighted average of i weighted in _i ) is calculated, and the obtained average value is output as the shift amount s. However, since s becomes a real number as it is, the shift amount determination unit 81 rounds it to an integer by rounding off or rounding down.
This generalized method is the same as the calculation method based on the maximum power in (a) above when the weighting function f () is defined as in the following equations (5) and (6).
f (pi) = 1 (when i = s) (5)
f (pi) = 0 (when i ≠ s) (6)
Further, when the weighting function f () is defined as in the following equation (7), the calculation method based on the power weighting in (b) is the same.
f (pi) = | pi | ² (7)
Furthermore, when the weighting function f () is defined as in the following equation (8), the calculation method is the same as that based on the amplitude weighting in the above (c).
f (pi) = | pi | (8)
(E) Method of calculating shift amount by footing The delay profile (complex value) p _i is used to detect the shift amount after cutting off. In this way, noise can be removed and the shift amount calculation accuracy can be improved. FIG. 8 is a configuration diagram of a shift timing detection unit provided with a footing part 91. In the configuration in which the footing part 91 is provided in the preceding stage of any one of the shift timing detection parts 92 in FIGS. It has become. The truncation unit 91 determines the appropriate threshold T, smaller than T _q i = 0, and larger if _{q i} = _p i. That is,
qi = pi (when | pi | ² ≧ T) (9)
qi = 0 (when | pi | ² <T) (10)
Q _i is determined according to the above, and the shift timing detection unit 92 obtains the shift amount s by performing the same operation as the above (a) to (d) using this q _i .
Various methods for determining the cut-off level T can be considered.

Accordingly, the cut-off level T is set to 4 times the average value of | p _i | ² . Or
^{T = | Pmax | 2/4} (| Pmax | 2 is the maximum value of pi) (12)
_{According, ^|} p i _| ² of a quarter of the maximum value and cutback level T. FIG. 9 is a block diagram of a shift timing detection unit provided with a footstep level determination unit 93 in addition to the configuration of FIG. 8. The footstep level determination unit 93 sets the footstep level T according to the equation (11) or (12). It is determined and input to the footing portion 91. If truncation unit 91 is smaller than the threshold value T _q i = 0, and _q i = _{p i} is greater, the shift timing detector 92 the same operation as described above with reference to the qi (a) ~ (d) Thus, the shift amount s is obtained.
(F) Modification of foot-cutting Equations (9) and (10) in the foot-cut method are as follows:
qi = 1 (when | pi | ² ≧ T) (13)
qi = 0 (when | pi | ² <T) (14)
Using the obtained q _i

The shift amount s is obtained according to When the transformation is made in this way, a simple average delay time (shift amount) of the multipath is obtained. For example, in the case of two paths, if the timing of each delay profile is t ₁ and t ₂ , the timing in the middle of the two paths (t ₁ + t ₂ ) / 2 regardless of the magnitude relationship of the paths according to the modification method. Is obtained as the average delay time (shift amount s) of the path.
(G) Delay profile shift operation When a path exists before the FFT window timing, that is, when a wave arriving before the FFT window timing exists, the delay profile corresponding to the path is shown in FIG. Appears at the tail as shown. In such a case, since the delay profile does not exist at the correct position, the shift amount cannot be detected accurately. A description will be given with reference to FIG. For example, when the two-pass as shown in FIG. 10 (a), the start timing _{W r} are two paths A of the FFT window timing, just when there in the middle of B, the path A, a delay profile _P A of B, P _B appears at the position of the solid line in FIG. Therefore, when the shift amount s is obtained by weighting, s is near the center of the delay profile range as shown in FIG. This is not preferred because the desired shift amount s is near zero. Therefore, as shown in FIG. 10C, the delay profiles P _A and P _B of the paths A and _B are shifted in advance by sp to obtain P _A ′ and P _B ′, and then s is obtained. Thereafter, s = s An accurate shift amount s is obtained by returning it to -s _p .
FIG. 11 is a block diagram of a shift timing detection unit having means for shifting the delay profile. Delay profile shift unit 95 is a delay profile _{p i (i = 0, ...} , (N-1)) only _{s p} equation q _i = _{p k}
However, k = (i + s _p ) mod N (i = 0, ..., (N-1)) (16)
And q _i are shifted by, thereafter, FIGS 7 (a) ~ (d) and any shift timing detection unit 92 of FIGS. 8-9 obtains a temporary shift amount s by using this q _i . After the provisional shift amount s is obtained, the shift amount correction unit 96 calculates the following equation: s = s−s _p (17)
By subtracting sp from the shift amount s, the original s is calculated and output. Since the shift amount s _p may be sufficiently larger than the extent of the delay profile (maximum delay time) appears to be about twice the guard interval length.
As described above, according to the present invention, intersymbol interference ISI can be eliminated, the average value of FFT conversion can be increased, the rotation amplitude can be decreased, and channel estimation can be performed correctly.

A Direct wave (second received signal)
B Delay wave (first received signal)

Claims (1)

Reception for receiving a transmission signal to which an orthogonal frequency division multiplexing method is applied by extracting a signal portion corresponding to a predetermined FFT window size from a reception signal including a preceding wave and a delayed wave, and performing FFT processing on the signal portion. In the device
A signal cutout unit that cuts out a signal portion corresponding to the FFT window size from the leading timing of the preceding wave excluding the guard interval;
A detection unit for detecting the leading timing of the delayed wave excluding the guard interval;
Of the signal portion cut out by the signal cutout unit, a shift unit that shifts a partial signal between the leading timing of the preceding wave and the leading timing of the delayed wave to the end of the cut out signal portion ;
An FFT processing unit that performs an FFT process on the signal portion obtained by shifting by the shift unit;
A receiving apparatus comprising:

Images