JP4945751B2

JP4945751B2 - Transmission system, transmission method, reception filter, and decoding method

Info

Publication number: JP4945751B2
Application number: JP2006301013A
Authority: JP
Inventors: 浩古川
Original assignee: Kyushu University NUC
Current assignee: Kyushu University NUC
Priority date: 2006-11-06
Filing date: 2006-11-06
Publication date: 2012-06-06
Anticipated expiration: 2026-11-06
Also published as: JP2008118483A

Description

本発明は、伝送システム、伝送方法、受信フィルタ、及び復号方法に関し、特にブロック変調信号の等化についての伝送システム、伝送方法、受信フィルタ、及び復号方法に関する。 The present invention relates to a transmission system, a transmission method, a reception filter, and a decoding method, and more particularly to a transmission system, a transmission method, a reception filter, and a decoding method for equalization of a block modulation signal.

図２８は一般的な符号分割多重伝送システムの様子を示した図である。以下簡単に説明する。 FIG. 28 is a diagram showing a general code division multiplex transmission system. This will be briefly described below.

システムの説明を行う前に、符号分割多重伝送について簡単に説明する。マルチコード多重による直接拡散スペクトラム変調（ＭＣ−ＤＳＳＳ）は伝送する情報信号を複数ビット単位でブロック化し、１ブロック内の各ビットをそれぞれ直交する拡散符号で拡散し、時間軸上で多重して伝送する手法である。各ビットの拡散処理は、ブロック符号化とも捉えられるため、以下ではＭＣ−ＤＳＳＳにおける拡散符号のことを時間ブロック符号と呼ぶことにし、またＭＣ−ＤＳＳＳ変調と同様に、複数の時間ブロック符号に情報信号を乗じて並列多重伝送する変調方式を総称して符号分割多重変調と呼ぶことにする。 Before describing the system, code division multiplexing transmission will be briefly described. Direct spread spectrum modulation (MC-DSSS) by multi-code multiplexing blocks the information signal to be transmitted in units of multiple bits, spreads each bit in one block with spreading codes orthogonal to each other, and multiplexes and transmits on the time axis It is a technique to do. Since the spreading process of each bit is also regarded as block coding, the spreading code in MC-DSSS will be referred to as a time block code below, and information in a plurality of time block codes as in MC-DSSS modulation. Modulation schemes for multiplying signals and performing parallel multiplexing transmission are collectively referred to as code division multiplexing modulation.

図２８を参照して、符号分割多重伝送システム１００は、送信機１００ａと受信機１００ｂとを備え、送信機１００ａから送信された信号は伝送路１０９を介して受信機１００ｂに受信される。情報信号Ｓ［ｎ］は変調処理部１０１により変調処理され、その変調信号ｘ（ｔ）は、アップコンバータ１０３による処理後にＲＦ１０５によって無線周波数（高周波）処理されて送信アンテナ１０７から送信される。送信された信号は、伝送路１０９を介して受信アンテナ１１１により受信される。受信された信号は、ＬＮＡ＆受信フィルタ１１３により低雑音増幅処理とフィルタ処理され、ダウンコンバータ１１５により処理される。そして、受信信号ｒ［ｔ］が復調処理部１１７により処理され、復号信号Ｓ＾［ｎ］が得られる。 Referring to FIG. 28, code division multiple transmission system 100 includes a transmitter 100a and a receiver 100b, and a signal transmitted from transmitter 100a is received by receiver 100b via transmission path 109. The information signal S [n] is modulated by the modulation processing unit 101, and the modulated signal x (t) is processed by the RF 105 after being processed by the up-converter 103 and transmitted from the transmitting antenna 107. The transmitted signal is received by the receiving antenna 111 via the transmission path 109. The received signal is subjected to low noise amplification processing and filtering processing by the LNA & reception filter 113 and processed by the down converter 115. Then, the received signal r [t] is processed by the demodulation processing unit 117, and a decoded signal S ^ [n] is obtained.

ところで、ブロック変調における各変調ブロックは有限時間に制限された信号である。このような有限時間信号に対する逆特性等化（Frequency Domain Inverse Filter, ＦＤＩＦ）の議論においては、被等化信号を時間軸上で繰り返し出現する周期信号とみなすことや（これを周期信号仮定と呼ぶことにする）、ブロック外はすべて０とすること（これをゼロ充填仮定と呼ぶことにする）から議論が展開される。このような仮定を行なう目的は、元来有限時間信号である被等化信号を無限に続く信号と捉えることで周波数軸上での議論に置き換えるためである。言うまでもなく、周期信号仮定ではフーリエ級数もしくは離散フーリエ変換（ＤＦＴ）が用いられ、ゼロ充填仮定ではフーリエ変換もしくは離散時間フーリエ変換（ＤＴＦＴ）が用いられる。 By the way, each modulation block in the block modulation is a signal limited to a finite time. In the discussion of inverse characteristic equalization (Frequency Domain Inverse Filter, FDIF) for such a finite time signal, the equalized signal is regarded as a periodic signal that repeatedly appears on the time axis (this is called a periodic signal assumption). The argument is developed from setting all outside the block to 0 (this is called the zero-filling assumption). The purpose of making such an assumption is to replace the equalized signal, which is originally a finite-time signal, with a discussion on the frequency axis by taking it as an infinite signal. Needless to say, a Fourier series or discrete Fourier transform (DFT) is used in the periodic signal assumption, and a Fourier transform or discrete time Fourier transform (DTFT) is used in the zero filling assumption.

図２９は図２８の復調処理部１１７のためにＬＮＡ＆受信フィルタ１１３等において用いられる通常の周波数軸上での逆特性等化（ＦＤＩＦ）を説明するための図である。 FIG. 29 is a diagram for explaining inverse characteristic equalization (FDIF) on a normal frequency axis used in the LNA & reception filter 113 and the like for the demodulation processing unit 117 of FIG.

ブロック変調信号をベクトルＸで表す。フィルタ２００は、トランスバーサルフィルタ２０１によって回路が実現され、ベクトルＲで表される受信信号が得られる（非特許文献１参照）。フィルタ２００のように、通常の逆特性等化では各受信シンボルｒ＿ｉは等しいＦＩＲフィルタが施された結果の信号となる。 The block modulation signal is represented by a vector X. The filter 200 is realized by a transversal filter 201, and a received signal represented by a vector R is obtained (see Non-Patent Document 1). Like the filter 200, in normal inverse characteristic equalization, each received symbol r_i is a signal that is the result of applying the same FIR filter.

Proakis編,″DigitalCommunications, 4th edition,″McGraw-Hill ,pp.616-618,2001Proakis, ″ DigitalCommunications, 4th edition, ″ McGraw-Hill, pp.616-618,2001

しかしながら、逆特性等化（ＦＤＩＦ）は周波数軸上で逆特性となるようなフィルタ伝達特性を実現するために、等化対象の伝送路のある周波数成分に深い落ち込みが存在する場合、これを等化しようと雑音成分を強調してしまい伝送特性が劣化する問題があった。既知の信号を伝送し、これと等化後の信号との誤差の自乗平均を最小にするようにフィルタ係数を制御する手法を適用すれば、雑音強調を抑制できることが知られているが、一方でこの場合、伝送路歪を受けた信号を完全に等化することができなくなり、等化後の信号に歪成分が残ってしまう。さらに逆特性等化は、歪成分を分解し合成するパスダイバーシチとは正反対に歪成分を無くするように働くため、パスダイバーシチ効果は得られず、前記自乗誤差を最小にするフィルタ係数制御を行なった場合でもパスダイバーシチと等化とが相反する効果として現れ、両者の両立は行えなかった。 However, inverse characteristic equalization (FDIF) realizes a filter transfer characteristic that has an inverse characteristic on the frequency axis. If there is a deep drop in a certain frequency component of the transmission path to be equalized, this is equalized. There is a problem that noise components are emphasized to reduce transmission characteristics and transmission characteristics deteriorate. It is known that noise enhancement can be suppressed by applying a technique that controls a filter coefficient so that a known signal is transmitted and the mean square error between this signal and the equalized signal is minimized. In this case, a signal subjected to transmission path distortion cannot be completely equalized, and a distortion component remains in the equalized signal. Furthermore, the inverse characteristic equalization works to eliminate the distortion component in the opposite direction to the path diversity that decomposes and synthesizes the distortion component. Therefore, the path diversity effect cannot be obtained, and the filter coefficient control is performed to minimize the square error. Even in this case, the effects of path diversity and equalization appear to be contradictory, and both cannot be achieved.

本発明は、Ｎの有限時間長のブロック変調信号に対して、逆特性等化と同様の効果とパスダイバーシチ効果を同時に行う受信等化を実現する伝送システム、伝送方法、受信フィルタ及び復号方法を提供することを目的とする。Ｎの有限時間長のブロック変調信号に対して、周波数軸上への議論の置き換えを行なうことなく、当該有限時間内で時間軸のままで等化することを特徴とする伝送システム、伝送方法、受信フィルタ及び復号方法を提供することを目的とする。 The present invention provides a transmission system, a transmission method, a reception filter, and a decoding method for realizing reception equalization that simultaneously performs an effect similar to inverse characteristic equalization and a path diversity effect on N block modulation signals having a finite time length. The purpose is to provide. A transmission system, a transmission method, and an equalization of N block modulation signals having a finite time length, with the time axis being equalized within the finite time without replacing the discussion on the frequency axis, An object is to provide a reception filter and a decoding method.

本発明の第１の観点は、長さＮの時系列信号であるブロック変調信号Ｘを生成する送信装置と、前記ブロック変調信号Ｘが歪伝送路を経由した後の長さＭの時系列信号である歪ブロック変調信号Ｙとし、当該歪ブロック変調信号Ｙを等化する受信フィルタを有する受信装置とを備えた伝送システムにおいて、前記受信フィルタが、並列配置されたＮ個の要素フィルタＥＦ＿ｉ（ｉ＝０〜Ｎ−１）からなり、入力された前記歪ブロック変調信号Ｙについて、前記各要素フィルタＥＦ＿ｉの出力シンボルｒ＿ｉがそれぞれ前記歪ブロック変調信号Ｙの等化後の長さＮの時系列信号シンボルＲ＝（ｒ＿０，ｒ＿１，・・・，ｒ＿Ｎ−１）となるように並列直列変換を行なうことを特徴とするものである。
A first aspect of the present invention is a transmitter that generates a block modulation signal X that is a time-series signal having a length N, and a time-series signal having a length M after the block modulation signal X has passed through a distorted transmission path. And a reception apparatus having a reception filter for equalizing the distortion block modulation signal Y, the reception filter includes N element filters EF_i (i) arranged in parallel. = 0 to N−1), and for the input distorted block modulated signal Y, the output symbol r_i of each element filter EF_i is a time-series signal of length N after equalization of the distorted block modulated signal Y, respectively. The parallel-to-serial conversion is performed such that the symbol R = (r_0, r_1,..., R_N−1).

本発明の第２の観点は、第１の観点において、前記各要素フィルタＥＦ＿ｉをそれぞれＭ次のＦＩＲフィルタで与え、当該ＦＩＲフィルタの各タップ係数をｆ＿ｉ，ｊ（ｊ＝０〜Ｍ−１）とし、ｆ＿ｉ，ｊをｉ行ｊ列の要素とする行列Ｈ_ＬＩ（Ｎ行Ｍ列の行列）を定め、前記歪伝送路のインパルス応答行列を行列Ｈ（Ｍ行Ｎ列の行列）とした場合に、行列Ｈ_ＬＩが下記第１式、第２式及び第３式から導かれる第４式の行列Ｈ_ＬＩによって設定された前記受信フィルタを有することを特徴とするものである。
Ｈ_ＬＩＨ＝Ｉ（１）
（但し、Ｉは単位行列）
Ｈ＝ＵΛ^１／２Ｖ^Ｈ（２）
（但し、Ｕ、ＶはそれぞれＨＨ^Ｈ、Ｈ^ＨＨの規格化された固有ベクトルを列ベクトルとするユニタリ行列、ΛはＨ^ＨＨの固有値を対角成分にもつ対角行列）
ＵΛ^１／２＝ＨＶ（３）
Ｈ_ＬＩ＝ＶΛ^−１Ｖ^ＨＨ^Ｈ（４）
According to a second aspect of the present invention , in the first aspect , each element filter EF_i is given by an M-th order FIR filter, and each tap coefficient of the FIR filter is set to f_i, j (j = 0 to M−1). And a matrix H _LI (N rows and M columns matrix) having f_i, j as elements of i rows and j columns is defined, and the impulse response matrix of the distortion transmission path is defined as a matrix H (M rows and N columns matrix). Further, the matrix H _LI includes the reception filter set by the matrix H _LI of the fourth formula derived from the following first, second, and third formulas.
H _LI H = I (1)
(Where I is the unit matrix)
H = UΛ ^1/2 V ^H (2)
(Where U and V are unitary matrices having normalized eigenvectors of HH ^H and H ^H H as column vectors, respectively, and Λ is a diagonal matrix having the eigenvalues of H ^H H as diagonal components)
UΛ ^1/2 = HV (3)
H _LI = VΛ ⁻¹ V ^H H ^H (4)

本発明の第３の観点は、第１の観点において、前記各要素フィルタＦ＿ｉをそれぞれＭ次のＦＩＲフィルタで与え、当該ＦＩＲフィルタの各タップ係数をｆ＿ｉ，ｊ（ｊ＝０〜Ｍ−１）とし、ｆ＿ｉ，ｊをｉ行ｊ列の要素とする行列Ｇ（Ｎ行Ｍ列の行列）を定め、前記ブロック変調信号Ｘとして受信側で既知の信号系列ＸＫを伝送し、前記時系列信号シンボルＲと前記信号系列ＸＫの誤差の自乗平均が最小となるように行列Ｇの各要素を制御することを特徴とするものである。
According to a third aspect of the present invention , in the first aspect , each element filter F_i is given by an M-th order FIR filter, and each tap coefficient of the FIR filter is set to f_i, j (j = 0 to M−1). And a matrix G (N rows and M columns matrix) having f_i, j as elements of i rows and j columns is defined, and a known signal sequence XK is transmitted as the block modulation signal X on the receiving side, and the time series signal symbols Each element of the matrix G is controlled so that the root mean square of errors between R and the signal sequence XK is minimized.

本発明の第４の観点は、第１の観点において、前記各要素フィルタＦ＿ｉをそれぞれＭ次のＦＩＲフィルタで与え、当該ＦＩＲフィルタの各タップ係数をｆ＿ｉ，ｊ（ｊ＝０〜Ｍ−１）とし、ｆ＿ｉ，ｊをｉ行ｊ列の要素とする行列Ｇ（Ｎ行Ｍ列の行列）を定め、等化後の前記時系列信号シンボルＲの系列を復号し、得られる復号信号系列Ｒ＿Ｈと前記時系列信号シンボルＲの系列との誤差の自乗平均が最小となるように行列Ｇの各要素を制御することを特徴とするものである。
According to a fourth aspect of the present invention , in the first aspect , each element filter F_i is given by an M-th order FIR filter, and each tap coefficient of the FIR filter is set to f_i, j (j = 0 to M−1). And a matrix G (a matrix of N rows and M columns) having f_i, j as elements of i rows and j columns is determined, a sequence of the time series signal symbols R after equalization is decoded, and a decoded signal sequence R_H obtained Each element of the matrix G is controlled so that the root mean square error with respect to the time series signal symbol R series is minimized.

本発明の第５の観点は、送信装置が長さＮの時系列信号であるブロック変調信号Ｘを生成して送信し、前記ブロック変調信号Ｘが歪伝送路を経由した後の長さＭの時系列信号である歪ブロック変調信号Ｙとし、受信装置が当該歪ブロック変調信号Ｙを等化する受信フィルタを有して受信する伝送方法において、前記受信フィルタが、並列配置されたＮ個の要素フィルタＥＦ＿ｉ（ｉ＝０〜Ｎ−１）からなり、入力された前記歪ブロック変調信号Ｙについて、前記各要素フィルタＥＦ＿ｉの出力シンボルｒ＿ｉがそれぞれ前記歪ブロック変調信号Ｙの等化後の長さＮの時系列信号シンボルＲ＝（ｒ＿０，ｒ＿１，・・・，ｒ＿Ｎ−１）となるように並列直列変換を行なうことを特徴とする。
According to a fifth aspect of the present invention , a transmission apparatus generates and transmits a block modulation signal X, which is a time-series signal having a length N, and the block modulation signal X has a length M after passing through a distortion transmission path. In the transmission method in which the reception device receives a distortion block modulation signal Y that is a time-series signal and the reception apparatus has a reception filter that equalizes the distortion block modulation signal Y, the reception filter includes N elements arranged in parallel. Consists of filters EF_i (i = 0 to N−1). For the input distorted block modulation signal Y, the output symbol r_i of each element filter EF_i is the length N after equalization of the distorted block modulation signal Y, respectively. The time-series signal symbol R = (r_0, r_1,..., R_N−1) is subjected to parallel / serial conversion.

本発明の第６の観点は、第５の観点において、前記歪伝送路のインパルス応答行列を行列Ｈ（Ｍ行Ｎ列の行列）とした場合に、行列Ｈ_ＬＩが下記第１式、第２式及び第３式から導かれる第４式の行列Ｈ_ＬＩによって設定された前記受信フィルタを有することを特徴とする。
Ｈ_ＬＩＨ＝Ｉ（１）
（但し、Ｉは単位行列）
Ｈ＝ＵΛ^１／２Ｖ^Ｈ（２）
（但し、Ｕ、ＶはそれぞれＨＨ^Ｈ、Ｈ^ＨＨの規格化された固有ベクトルを列ベクトルとするユニタリ行列、ΛはＨ^ＨＨの固有値を対角成分にもつ対角行列）
ＵΛ^１／２＝ＨＶ（３）
Ｈ_ＬＩ＝ＶΛ^−１Ｖ^ＨＨ^Ｈ（４）
According to a sixth aspect of the present invention , in the fifth aspect , when the impulse response matrix of the distorted transmission line is a matrix H (a matrix of M rows and N columns), the matrix H _LI is expressed by the following first and second formulas: The reception filter set by the matrix H _LI of the fourth equation derived from the equations and the third equation is provided.
H _LI H = I (1)
(Where I is the unit matrix)
H = UΛ ^1/2 V ^H (2)
(Where U and V are unitary matrices having normalized eigenvectors of HH ^H and H ^H H as column vectors, respectively, and Λ is a diagonal matrix having the eigenvalues of H ^H H as diagonal components)
UΛ ^1/2 = HV (3)
H _LI = VΛ ⁻¹ V ^H H ^H (4)

本発明の第７の観点は、長さＮＴ＊Ｎのブロック変調信号Ｘを長さＮの部分ブロック変調信号Ｘ＿ｉ（それぞれ長さＮの時系列信号、ｉ＝０〜ＮＴ−１）に分割し、当該部分ブロック変調信号Ｘ＿ｉをそれぞれ複数の送信アンテナから出力する送信装置と、前記部分ブロック変調信号Ｘ＿０〜Ｘ＿ＮＴ−１が歪伝送路を経由した後にＮＲ個の複数の受信アンテナによって受信される歪部分ブロック変調信号をＹ＿ｊ(それぞれ長さＭの時系列信号、ｊ＝０〜ＮＲ−１)とし、歪部分ブロック変調信号を従属に接続することで歪ブロック変調信号Ｙ＝（Ｙ＿０、Ｙ＿１、・・・、Ｙ＿ＮＲ−１）（長さＮＲ＊Ｍ）を生成し、当該歪ブロック変調信号Ｙを等化する受信フィルタを有する受信装置とを備えた伝送システムにおいて、前記受信フィルタが並列配置されたＮＴ＊Ｎ個の要素フィルタＥＦ＿ｋ（ｋ＝０〜ＮＴ＊Ｎ−１）からなり、前記歪ブロック変調信号Ｙが各要素フィルタＥＦ＿ｋに入力され、各要素フィルタＥＦ＿ｋの出力シンボルｒ＿ｋがそれぞれ前記歪ブロック変調信号Ｙの等化後のシンボルＲ＝（ｒ＿０，ｒ＿１，・・・，ｒ＿ＮＴ＊Ｎ−１）（長さＮＴ＊Ｎの時系列信号）となるように並列直列変換を行なうことを特徴とするものである。
According to a seventh aspect of the present invention, a block modulation signal X having a length NT * N is divided into partial block modulation signals X_i having a length N (time-series signals having a length N, i = 0 to NT-1 respectively). A transmission apparatus that outputs the partial block modulation signals X_i from a plurality of transmission antennas, and a distortion received by the NR reception antennas after the partial block modulation signals X_0 to X_NT-1 pass through a distortion transmission path. The partial block modulation signal is Y_j (each time-series signal of length M, j = 0 to NR-1), and the distortion partial block modulation signal is connected to the subordinates, thereby distorting block modulation signal Y = (Y_0, Y_1,. .., Y_NR-1) (length NR * M) and a reception apparatus having a reception filter for equalizing the distorted block modulation signal Y, the reception filter It consists of NT * N element filters EF_k (k = 0 to NT * N−1) arranged in parallel, and the distorted block modulation signal Y is input to each element filter EF_k, and the output symbol r_k of each element filter EF_k is Parallel-serial conversion is performed so that the equalized symbols R = (r_0, r_1,..., R_NT * N−1) (time-series signals of length NT * N) are obtained for the distorted block modulation signal Y, respectively. It is characterized by this.

本発明の第８の観点は、第７の観点において、前記各要素フィルタＥＦ＿ｉをそれぞれＮＲ＊Ｍ次のＦＩＲフィルタで与え、当該ＦＩＲフィルタの各タップ係数をｆ＿ｉ，ｊ（ｊ＝０〜ＮＲ＊Ｍ−１）とし、ｆ＿ｉ，ｊをｉ行ｊ列の要素とする行列Ｈ_Ｍ,ＬＩ（（ＮＴ＊Ｎ）行（ＮＲ＊Ｍ）列の行列）を定め、前記歪伝送路のインパルス応答行列を行列Ｈ_Ｍ（（ＮＲ＊Ｍ）行（ＮＴ＊Ｎ）列の行列）とした場合に、行列Ｈ_Ｍ,ＬＩが下記第１式、第２式及び第３式から導かれる第４式の行列Ｈ_Ｍ,ＬＩによって設定された前記受信フィルタを有することを特徴とする。
Ｈ_Ｍ,ＬＩＨ_Ｍ＝Ｉ（１）
（但し、Ｉは単位行列）
Ｈ_Ｍ＝ＵΛ^１／２Ｖ^Ｈ（２）
（但し、Ｕ、ＶはそれぞれＨ_ＭＨ_Ｍ ^Ｈ、Ｈ_Ｍ ^ＨＨ_Ｍの規格化された固有ベクトルを列ベクトルとするユニタリ行列、ΛはＨ_Ｍ ^ＨＨ_Ｍの固有値を対角成分にもつ対角行列）
ＵΛ^１／２＝Ｈ_ＭＶ（３）
Ｈ_Ｍ,ＬＩ＝ＶΛ^−１Ｖ^ＨＨ_Ｍ ^Ｈ（４）
According to an eighth aspect of the present invention , in the seventh aspect , each element filter EF_i is given by an NR * M-th order FIR filter, and each tap coefficient of the FIR filter is set to f_i, j (j = 0 to NR *). M-1), a matrix H _{M, LI} (matrix of (NT * N) rows (NR * M) columns) having f_i, j as elements of i rows and j columns is defined, and an impulse response matrix of the distortion transmission path Is a matrix H _M (a matrix of (NR * M) rows (NT * N) columns), the matrix H _{M, LI} is expressed by the following formulas (1), (2) and (3): It has the said reception filter set by matrix _{HM, LI} .
H _{M, LI} H _M = I (1)
(Where I is the unit matrix)
H _M = UΛ ^1/2 V ^H (2)
(Where U and V are unitary matrices whose column vectors are normalized eigenvectors of H _M H _M ^H and H _M ^H H _M , respectively, and Λ is a diagonal having the eigenvalues of H _M ^H H _M as diagonal components. line; queue; procession; parade)
UΛ ^1/2 = H _M V (3)
H _{M, LI} = VΛ ⁻¹ V ^H H _M ^H (4)

本発明の第９の観点は、第７の観点において、前記各要素フィルタＥＦ＿ｉをそれぞれＮＲ＊Ｍ次のＦＩＲフィルタで与え、当該ＦＩＲフィルタの各タップ係数をｆ＿ｉ，ｊ（ｊ＝０〜ＮＲ＊Ｍ−１）とし、ｆ＿ｉ，ｊをｉ行ｊ列の要素とする行列Ｇ（（ＮＴ＊Ｎ）行（ＮＲ＊Ｍ）列の行列）を定め、前記ブロック変調信号Ｘとして受信側で既知の信号系列ＸＫを伝送し、前記時系列信号シンボルＲと前記信号系列ＸＫの誤差の自乗平均が最小となるように行列Ｇの各要素を制御することを特徴とする。
According to a ninth aspect of the present invention , in the seventh aspect , each element filter EF_i is given by an NR * M-th order FIR filter, and each tap coefficient of the FIR filter is set to f_i, j (j = 0 to NR *). M-1), and a matrix G (matrix of (NT * N) rows (NR * M) columns) having f_i, j as elements of i rows and j columns is defined and known as the block modulation signal X on the receiving side. The signal sequence XK is transmitted, and each element of the matrix G is controlled so that the mean square of errors between the time series signal symbol R and the signal sequence XK is minimized.

本発明の第１０の観点は、第７の観点において、前記各要素フィルタＥＦ＿ｉをそれぞれＮＲ＊Ｍ次のＦＩＲフィルタで与え、当該ＦＩＲフィルタの各タップ係数をｆ＿ｉ，ｊ（ｊ＝０〜ＮＲ＊Ｍ−１）とし、ｆ＿ｉ，ｊをｉ行ｊ列の要素とする行列Ｇ（（ＮＴ＊Ｎ）行（ＮＲ＊Ｍ）列の行列）を定め、等化後の前記時系列信号シンボルＲの系列を復号し、得られる復号信号系列Ｒ＿Ｈと前記時系列信号シンボルＲの系列との誤差の自乗平均が最小となるように行列Ｇの各要素を制御することを特徴とする。
According to a tenth aspect of the present invention , in the seventh aspect , each element filter EF_i is given by an NR * M-th order FIR filter, and each tap coefficient of the FIR filter is set to f_i, j (j = 0 to NR *). M-1), a matrix G (matrix of (NT * N) rows (NR * M) columns) having f_i, j as elements of i rows and j columns is defined, and the time series signal symbol R after equalization is determined. Each element of the matrix G is controlled such that the mean square of errors between the decoded signal series R_H obtained and the time series signal symbol R series is minimized.

本発明の第１１の観点は、第１から第４と第７から第１０のいずれかの観点において、前記ブロック変調信号Ｘが、複数の時間ブロック符号にそれぞれに異なる情報信号を乗じて時間軸上で加算による多重により与えられることを特徴とする。
According to an eleventh aspect of the present invention, in any one of the first to fourth and seventh to tenth aspects , the block modulation signal X is obtained by multiplying a plurality of time block codes by different information signals, respectively. It is characterized by being given by multiplexing by addition above.

本発明の第１２の観点は、第１１の観点において、前記情報信号に誤り訂正符号を適用し、当該誤り訂正符号を適用した後の信号系列を新たな情報信号とすることを特徴とする。
According to a twelfth aspect of the present invention , in the eleventh aspect, an error correction code is applied to the information signal, and a signal sequence after the error correction code is applied is set as a new information signal.

本発明の第１３の観点は、第１から第４と第７から第１０のいずれかの伝送システムに用いられる受信フィルタである。
A thirteenth aspect of the present invention is a reception filter used in any one of the first to fourth and seventh to tenth transmission systems.

本発明の第１４の観点は、送信装置が長さＮの時系列信号であるブロック変調信号Ｘを生成して送信し、前記ブロック変調信号Ｘが歪伝送路を経由した後の長さＭの時系列信号である歪ブロック変調信号Ｙとし、受信装置が当該歪ブロック変調信号Ｙを等化する受信フィルタを有して受信する伝送方法における復号方法であって、前記受信フィルタが、並列配置されたＮ個の要素フィルタＥＦ＿ｉ（ｉ＝０〜Ｎ−１）からなり、入力された前記歪ブロック変調信号Ｙについて、前記各要素フィルタＥＦ＿ｉの出力シンボルｒ＿ｉがそれぞれ前記歪ブロック変調信号Ｙの等化後の長さＮの時系列信号シンボルＲ＝（ｒ＿０，ｒ＿１，・・・，ｒ＿Ｎ−１）となるように並列直列変換を行なうことを特徴とし、前記受信装置が、受信した歪ブロック変調信号Ｙを記録する受信変調ブロックバッファと、受信した各歪ブロック変調信号に対する復号結果を記録する復号ブロックバッファを有し、新たな歪ブロック変調信号Ａが受信されると、当該歪ブロック変調信号Ａを前記受信変調ブロックバッファに格納し、前記復号ブロックバッファに記録済みである前記歪ブロック変調信号Ａの直前に受信された歪ブロック変調信号Ｂの復号結果を再変調しさらに伝送路に相当するフィルタをかけることで前記歪ブロック変調信号Ａが歪ブロック変調信号Ｂから受ける前置干渉成分を求め、当該前置干渉成分を前記歪ブロック変調信号Ａから除去して復号した復号結果を前記復号ブロックバッファに格納し、前記受信変調ブロックバッファ内の歪ブロック変調信号のうち２番目に新しい歪ブロック変調信号を再復号対象歪ブロック変調信号に設定する第１ステップと、当該再復号対象歪ブロック変調信号の直前に受信された歪ブロック変調信号の最新の復号結果を前記復号ブロックバッファから読み出して再変調しさらに伝送路に相当するフィルタをかけることで前置干渉成分を求め、当該再復号対象歪ブロック変調信号の直後に受信された歪ブロック変調信号の最新の復号結果を前記復号ブロックバッファから読み出して再変調しさらに伝送路に相当するフィルタをかけることで後置干渉成分を求め、再復号対象歪ブロック変調信号から前記前置干渉成分と前記後置干渉線分とを除去して復号した復号結果を前記復号ブロックバッファに格納し、前記受信変調ブロックバッファに格納されている前記再復号対象歪ブロック変調信号の直前に受信された歪ブロック変調信号を新たな再復号対象歪ブロック変調信号に設定する第２ステップと、前記第２ステップを所望の回数繰り返して行う第３ステップと、を含むものである。
According to a fourteenth aspect of the present invention , a transmitter generates and transmits a block modulation signal X that is a time-series signal having a length N, and the block modulation signal X has a length M after passing through a distortion transmission path. A decoding method in a transmission method in which a distorted block modulated signal Y, which is a time-series signal, is received by a receiving apparatus having a reception filter that equalizes the distorted block modulated signal Y. The receiving filter is arranged in parallel N element filters EF_i (i = 0 to N−1), and for the input distorted block modulation signal Y, the output symbol r_i of each element filter EF_i is equalized to the distorted block modulation signal Y, respectively. A parallel-to-serial conversion is performed so that a time-series signal symbol R = (r_0, r_1,..., R_N−1) having a length of N is obtained later, and the receiving apparatus receives the received distortion block. When receiving a new distorted block modulation signal A, the received modulation block buffer has a reception modulation block buffer for recording the modulation signal Y and a decoding block buffer for recording a decoding result for each received distortion block modulation signal. A is stored in the reception modulation block buffer, and the decoding result of the distorted block modulation signal B received immediately before the distorted block modulation signal A recorded in the decoding block buffer is remodulated and further corresponds to a transmission path. A pre-interference component received by the distorted block modulated signal A from the distorted block modulated signal B is obtained by applying a filter, and the decoded result obtained by removing the pre-interference component from the distorted block modulated signal A and decoding the decoded block The second new distortion block stored in the buffer and the distortion block modulation signal in the reception modulation block buffer. A first step of setting a key signal as a re-decoding target distortion block modulation signal, and reading out the latest decoding result of the distortion block modulation signal received immediately before the re-decoding target distortion block modulation signal from the decoding block buffer and A pre-interference component is obtained by modulating and applying a filter corresponding to the transmission path, and the latest decoding result of the distorted block modulated signal received immediately after the distorted block modulated signal to be re-decoded is read from the decoded block buffer. Decoding is performed by obtaining a post-interference component by re-modulating and applying a filter corresponding to the transmission path, and removing the pre-interference component and the post-interference line segment from the distorted block modulation signal to be re-decoded. The result is stored in the decoding block buffer and immediately before the re-decoding target distorted block modulation signal stored in the reception modulation block buffer. A second step of setting the received distorted block modulated signal as a new re-decoding target distorted block modulated signal, and a third step of repeating the second step a desired number of times.

本発明の第１５の観点は、送信装置が、長さＮＴ＊Ｎのブロック変調信号Ｘを長さＮの部分ブロック変調信号Ｘ＿ｉ（それぞれ長さＮの時系列信号、ｉ＝０〜ＮＴ−１）に分割し、当該部分ブロック変調信号Ｘ＿ｉをそれぞれ複数の送信アンテナから出力し、前記部分ブロック変調信号Ｘ＿０〜Ｘ＿ＮＴ−１が歪伝送路を経由した後に、受信装置が、ＮＲ個の複数の受信アンテナによって受信される歪部分ブロック変調信号をＹ＿ｊ(それぞれ長さＭの時系列信号、ｊ＝０〜ＮＲ−１)とし、歪部分ブロック変調信号を従属に接続することで歪ブロック変調信号Ｙ＝（Ｙ＿０、Ｙ＿１、・・・、Ｙ＿ＮＲ−１）（長さＮＲ＊Ｍ）を生成し、当該歪ブロック変調信号Ｙを等化する受信フィルタを有して受信する伝送方法における復号方法であって、前記受信フィルタが並列配置されたＮＴ＊Ｎ個の要素フィルタＥＦ＿ｋ（ｋ＝０〜ＮＴ＊Ｎ−１）からなり、前記歪ブロック変調信号Ｙが各要素フィルタＥＦ＿ｋに入力され、各要素フィルタＥＦ＿ｋの出力シンボルｒ＿ｋがそれぞれ前記歪ブロック変調信号Ｙの等化後のシンボルＲ＝（ｒ＿０，ｒ＿１，・・・，ｒ＿ＮＴ＊Ｎ−１）（長さＮＴ＊Ｎの時系列信号）となるように並列直列変換を行なうことを特徴とし、前記受信装置が前記複数の受信アンテナにより並列に受信された各歪部分ブロック変調信号をそれぞれ記録するアンテナ毎の受信変調ブロックバッファと、前記歪ブロック変調信号に対する復号結果を記録する復号ブロックバッファとを有し、新たな歪部分ブロック変調信号Ａ＿ｉ(ｉはアンテナ番号)が受信されると、その直前に受信された歪部分ブロック変調信号Ｂ＿ｉに対応する歪ブロック変調信号の最新の復号結果を前記復号ブロックバッファから読み出して再変調しさらに伝送路に相当するフィルタをかけることでＡ＿ｉそれぞれに対する前置干渉成分を求め、前記歪部分ブロック変調信号Ａ＿ｉから当該前置干渉成分をそれぞれ除去し、前置干渉成分が除去された各Ａ＿ｉを従属に接続することで得られる歪ブロック変調信号に対して復号した復号結果を前記復号ブロックバッファに格納し、前記各受信アンテナ毎の受信変調ブロックバッファ内の歪部分ブロック変調信号のうち２番目に新しい歪部分ブロック変調信号を再復号対象歪部分ブロック変調信号に設定する第１ステップと、各受信アンテナ毎の再復号対象歪部分ブロック変調信号の直前に受信された歪部分ブロック変調信号に対応する歪ブロック変調信号の最新の復号結果を前記復号ブロックバッファから読み出して再変調しさらに伝送路に相当するフィルタをかけることで各受信アンテナ毎の再復号対象歪部分ブロック変調信号のそれぞれに対する前置干渉成分を求め、再復号対象歪部分ブロック変調信号の直後に受信された各受信アンテナ毎の歪部分ブロック変調信号に対応する歪ブロック変調信号の最新の復号結果を前記復号ブロックバッファから読み出して再変調しさらに伝送路に相当するフィルタをかけることで各アンテナ毎の再復号対象歪部分ブロック変調信号に対する後置干渉成分を求め、各再復号対象歪部分ブロック変調信号から前置干渉成分ならびに後置干渉成分をそれぞれ除去し、前置干渉ならびに後置干渉成分が除去された各再復号対象歪部分ブロック変調信号を従属に接続して得られる歪ブロック変調信号に対して復号した復号結果を前記復号ブロックバッファに格納し、前記各受信アンテナ毎の受信変調ブロックバッファ内の前記再復号対象歪部分ブロック変調信号の直前に受信された各歪部分ブロック変調信号を新たな再復号対象歪部分ブロック変調信号に設定する第２ステップと、前記第２ステップを所望の回数繰り返して行う第３ステップと、を含むものである。 According to a fifteenth aspect of the present invention , a transmitting apparatus converts a block modulation signal X having a length NT * N into a partial block modulation signal X_i having a length N (time-series signals each having a length N, i = 0 to NT−1). ), And the partial block modulation signals X_i are respectively output from a plurality of transmission antennas, and after the partial block modulation signals X_0 to X_NT-1 pass through the distorted transmission path, the reception apparatus receives a plurality of NR reception signals. The distortion partial block modulation signal received by the antenna is Y_j (time-series signal of length M, j = 0 to NR−1, respectively), and the distortion partial block modulation signal Y = (Y_0, Y_1,..., Y_NR-1) (length NR * M), and a decoding method in a transmission method for receiving with a reception filter that equalizes the distorted block modulation signal Y. The reception filter is composed of NT * N element filters EF_k (k = 0 to NT * N−1) arranged in parallel, and the distorted block modulation signal Y is input to each element filter EF_k. The output symbols r_k are parallel so that the equalized symbols R = (r_0, r_1,..., R_NT * N−1) (time-series signals of length NT * N) are respectively obtained after equalization of the distorted block modulation signal Y. A serial modulation, and a reception modulation block buffer for each antenna in which the reception apparatus records each distortion partial block modulation signal received in parallel by the plurality of reception antennas, and decoding for the distortion block modulation signal. A decoding block buffer for recording the result, and when a new distorted partial block modulation signal A_i (i is an antenna number) is received, The latest decoding result of the distorted block modulated signal corresponding to the previously received distorted partial block modulated signal B_i is read from the decoded block buffer, re-modulated, and further subjected to a filter corresponding to the transmission path, thereby applying a prefix to each A_i. An interference component is obtained, the pre-interference component is removed from the distorted partial block modulation signal A_i, and each A_i from which the pre-interference component has been removed is connected to the subordinate to decode the distorted block modulation signal. The decoded result is stored in the decoded block buffer, and the second newest distorted partial block modulated signal among the distorted partial block modulated signals in the received modulated block buffer for each receiving antenna is converted into a distorted partial block modulated signal to be re-decoded. The first step to be set is received immediately before the re-decoding target distorted partial block modulation signal for each receiving antenna. The latest decoding result of the distorted block modulated signal corresponding to the distorted partial block modulated signal is read out from the decoded block buffer, remodulated, and further subjected to a filter corresponding to the transmission path, so that the distortion to be recoded for each receiving antenna The latest decoding result of the distorted block modulated signal corresponding to the distorted partial block modulated signal for each receiving antenna received immediately after the distorted partial block modulated signal to be re-decoded is obtained for each of the partial block modulated signals. Is read out from the decoding block buffer, re-modulated, and a post-interference component for the re-decoding-target distortion partial block modulation signal for each antenna is obtained by applying a filter corresponding to the transmission path, and each re-decoding-target distortion partial block modulation is performed. Remove the front and rear interference components from the signal, respectively. The decoding result obtained by decoding the distorted block modulated signal obtained by subordinately connecting the distorted partial block modulated signals to be re-decoded with the components removed is stored in the decoded block buffer, and received modulation for each receiving antenna is performed. A second step of setting each distorted partial block modulation signal received immediately before the re-decoding target distorted partial block modulation signal in the block buffer as a new re-decoding target distorted partial block modulation signal; And a third step performed repeatedly.

本発明によれば、送信側で伝送路情報を有する必要がなく、回路構成が簡単で且つ変動する伝送路にも高い適合性がある受信等化が可能になる。有限時間Ｎの範囲において例えばＦＤＩＦと同様に元の送信波形に比例する等化後信号を得られると同時にパスダイバーシチの効果も得られる。 According to the present invention, it is not necessary to have transmission path information on the transmission side, and reception equalization with a simple circuit configuration and high adaptability to a changing transmission path is possible. In the range of finite time N, for example, an equalized signal proportional to the original transmission waveform can be obtained in the same manner as FDIF, and at the same time, the effect of path diversity can be obtained.

図１は本発明が開示する時間軸相似等化（ＦｉｎｉｔｅＷａｖｅｆｏｒｍＰｒｅｓｅｒｖａｔｉｏｎＦｉｌｔｅｒ, ＦＷＰＦ)と呼ぶ等化を実現するフィルタを説明するための図である。以下、まず、発明の原理について説明する。 FIG. 1 is a diagram for explaining a filter for realizing equalization called a time axis similarity equalization (Finite Waveform Preservation Filter, FWPF) disclosed by the present invention. Hereinafter, first, the principle of the invention will be described.

時間軸相似等化によるフィルタ１は、具体的には、タップ数Ｋの伝送路歪を受けた長さＮ＋Ｋ−１の歪変調ブロック波形が等化後に長さＮに渡り元の波形と相似な波形となるようなフィルタリングを行うものである。ここで元波形と等化後の波形が相似の関係にあるとは、等化後の波形が、元波形に固定の複素スカラー係数を掛けたもので与えられる関係を意味し、これを時間軸相似等化（ＦＷＰＦ）と呼ぶ。上記の周波数軸上での逆特性等化（ＦＤＩＦ）との違いは、逆特性等化が−∞〜∞の全ての時間において元波形と等化後の波形とが相似となることを要求するのに対して、時間軸相似等化では有限時間Ｎ内でのみこれを要求する点である。すなわち、時間軸相似等化における等化の要件は、逆特性等化に比べて緩い。以下、図１に示したフィルタについてさらに詳しく説明するとともに、時間軸相似等化と逆特性等化のそれぞれの等化要件の違いが両者の回路構成そのものに相違を生じさせることを明らかにし、さらに時間軸相似等化がパスダイバーシチの効果を生じさせることを示す。また、時間軸相似等化のこの特性を計算機シミュレーションによって明らかにする。 Specifically, the filter 1 by time axis similarity equalization is similar to the original waveform over the length N after the distortion modulation block waveform of length N + K−1 subjected to transmission line distortion of the number of taps K is equalized. Filtering is performed to obtain a waveform. Here, the relationship between the original waveform and the waveform after equalization means that the waveform after equalization is given by multiplying the original waveform by a fixed complex scalar coefficient. This is called similarity equalization (FWPF). The difference from the above-described inverse characteristic equalization (FDIF) on the frequency axis requires that the original waveform and the waveform after equalization are similar at all times where the inverse characteristic equalization is −∞ to ∞. On the other hand, time axis similarity equalization requires this only within a finite time N. That is, the requirement for equalization in time axis similarity equalization is looser than that for inverse characteristic equalization. Hereinafter, the filter shown in FIG. 1 will be described in more detail, and it will be clarified that the difference in the equalization requirements of the time axis similarity equalization and the inverse characteristic equalization causes a difference in the circuit configurations themselves. We show that time axis similarity equalization produces the effect of path diversity. Moreover, this characteristic of time axis similarity equalization is clarified by computer simulation.

以下、伝送路行列を定義し、時間軸相似等化を実現する図１のフィルタ１を導出する。導出された時間軸相似等化によりパスダイバーシチ効果が得られることを明らかにする。 Hereinafter, the transmission line matrix is defined, and the filter 1 of FIG. 1 that realizes time-axis similarity equalization is derived. It is clarified that the path diversity effect can be obtained by the derived time axis similarity equalization.

送信変調ブロックがＮ次の列ベクトルＸで与えられているものとする。タップ長Ｋの伝送路が与えられたとき、Ｍ行Ｎ列の伝送路行列Ｈは数１の（１）式のように定義される。 It is assumed that the transmission modulation block is given by an Nth-order column vector X. When a transmission line with a tap length K is given, a transmission line matrix H with M rows and N columns is defined as in equation (1).

ここで、ここでＭ＝Ｎ＋Ｋ−１である。変調信号ベクトルＸが伝送路Ｈを通過後に受信される信号ベクトルＹは、数２の（２）式で与えられる。 Here, M = N + K−1. The signal vector Y received after the modulated signal vector X passes through the transmission path H is given by the equation (2) in Equation 2.

ここで、ベクトルＹはＭ次の列ベクトルとなる。 Here, the vector Y is an M-th column vector.

各ブロック変調信号はガードインターバルによって変調ブロック間干渉（Inter Modulation Block Interference，ＩＢＩ）が十分抑制されているものとする。このようなブロック変調信号ベクトルＸが伝送路Ｈを通過した後の受信信号は（２）式で与えられるように有限時間長の信号ベクトルＹで与えられる。Ｈの左逆行列をＨ_ＬＩとすると、Ｈ_ＬＩは（３）式を満たせばよい。
Ｈ_ＬＩＨ＝Ｉ（３）
ここで、ＩはＮ行Ｎ列の単位行列を表す。 It is assumed that inter-modulation block interference (IBI) is sufficiently suppressed in each block modulation signal by the guard interval. The received signal after such a block modulation signal vector X has passed through the transmission path H is given by a signal vector Y having a finite time length as given by equation (2). Assuming that the left inverse matrix of H is H _LI , H _LI may satisfy equation (3).
H _LI H = I (3)
Here, I represents a unit matrix of N rows and N columns.

Ｈを特異値分解し、（４）式で表すとする。
Ｈ＝ＵΛ^１／２Ｖ^Ｈ（４）
ここで、Ｕ、ＶはそれぞれＨＨ^Ｈ、Ｈ^ＨＨの規格化された固有ベクトルを列ベクトルとするユニタリ行列、ΛはＨ^ＨＨ（或いはＨＨ^Ｈでも可）の固有値を対角成分にもつＮ行Ｎ列の対角行列である。 Let H be a singular value decomposition and be expressed by equation (4).
H = UΛ ^1/2 V ^H (4)
Here, U and V are unitary matrices whose column vectors are normalized eigenvectors of HH ^H and H ^H ^H , respectively, and Λ is N rows having eigenvalues of H ^H H (or HH ^H ) as diagonal components. It is a diagonal matrix with N columns.

また、ＵとＶは（５）式の関係があり、（３）式に（４）式及び（５）式を代入すれば、（６）式が導かれる。
ＵΛ^１／２＝ＨＶ（５）
Ｈ_ＬＩ＝ＶΛ^−１Ｖ^ＨＨ^Ｈ（６） U and V have the relationship of equation (5), and if equation (4) and equation (5) are substituted into equation (3), equation (6) is derived.
UΛ ^1/2 = HV (5)
H _LI = VΛ ⁻¹ V ^H H ^H (6)

図１のフィルタ１は、（６）式で与えられる等化回路がブロック図で示されたものであり、受信機初段に設置するＨ^Ｈ部３はＨ^Ｈが伝送路行列Ｈの随伴行列であるから伝送路最大比合成回路を意味し、パス捕捉漏洩のないパスダイバーシチを実現する（本願の発明者が既に提案している特願２００６−１６５２４９号（古川氏の論文”符号の直交分離とパスダイバーシチを同時に実現する符号分割多重伝送,” 信学技法, RCS2006-52, pp.101-106, June 2006.参照））。伝送路行列Ｈで表される伝送路２を通過した受信信号であるベクトルＹは、Ｈ^Ｈ部３によって伝送路最大比合成され、後続のＶ^Ｈ部５を通過することによりＶの直交する列ベクトルによって直交分解がなされ、その出力は分解後の分解係数を与える。各分解係数はさらにΛ^−１部７により固有値の逆数で重み付けが行なわれ、最終段のＶ部１１では重み付け後の係数より前記直交ベクトル群を再び重み付け加算し信号が再合成される。再合成された信号であるベクトルＲは、有限時間Ｎの範囲において歪を受ける前の送信信号（変調ブロック信号）であるベクトルＸと相似な波形（複素定数倍の波形）となる。なお、図１においてベクトルｎはノイズを表す。 The filter 1 in FIG. 1 is a block diagram of the equalization circuit given by equation (6). The H ^H unit 3 installed at the first stage of the receiver is an adjoint matrix of H ^H of the transmission line matrix H. Therefore, it means a transmission line maximum ratio combining circuit, and realizes path diversity without path capture leakage (Japanese Patent Application No. 2006-165249 already proposed by the inventors of the present application (Furukawa's paper) Code division multiplex transmission that realizes path diversity at the same time, "Science Technique, RCS2006-52, pp.101-106, June 2006.)). A vector Y, which is a received signal that has passed through the transmission path 2 represented by the transmission path matrix H, is combined in the transmission path maximum ratio by the H ^H section 3, and passes through the subsequent V ^H section 5, so that V is an orthogonal column. Orthogonal decomposition is performed by the vector, and its output gives a decomposition coefficient after decomposition. Each decomposition coefficient is further weighted by the inverse of the eigenvalue by the Λ ⁻¹ part 7, and the orthogonal vector group is weighted again by the weighted coefficient in the final V part 11 to re-synthesize the signal. The vector R, which is the recombined signal, has a waveform (a complex constant multiple waveform) similar to the vector X which is the transmission signal (modulation block signal) before being distorted in the range of the finite time N. In FIG. 1, vector n represents noise.

以上のように、（６）式で与えられる時間軸相似等化により、各変調ブロックの時間Ｎ内において逆特性等化と全く同一の等化作用が得られると同時に、パスダイバーシチの効果も得られる。送信側で事前の伝送路情報（ＣｈａｎｎｅｌＳｔａｔｅＩｎｆｏｒｍａｔｉｏｎ，ＣＳＩ）すなわち伝送路行列Ｈも必要としない。 As described above, the time axis similarity equalization given by the equation (6) provides the same equalization effect as the inverse characteristic equalization within the time N of each modulation block, and at the same time, the effect of path diversity. It is done. The transmission side information (Channel State Information, CSI), that is, the transmission line matrix H is not required on the transmission side.

図２は図１のフィルタを用いた伝送システムの概略を示す図である。図３は図１のフィルタ１の構成をさらに具体的に示すブロック図である。 FIG. 2 is a diagram showing an outline of a transmission system using the filter of FIG. FIG. 3 is a block diagram showing the configuration of the filter 1 of FIG. 1 more specifically.

伝送システム１３は、ＭＣ−ＤＳＳＳによるシステムであり、送信機１４と、受信機１５とを備え、送信機１４から送信された信号は伝送路２を介して受信機１５に受信される。なお、ＲＦ処理部は後述の図２０と異なり図示を省略した。情報信号Ｓ［ｎ］は、変調処理されるため、Ｓ／Ｐ変換部１６により各情報信号Ｓｉに変換され、乗算部１７により変調符号ベクトルｅｉが乗算される。そして、総和部１９により乗算後の総和が行われ、送信信号（変調ブロック信号）であるベクトルＸが送信される。送信信号は、伝送路２を介して歪後のベクトルＹとなり、ノイズ（ベクトルｎ）が加わった後、受信信号として受信機１５のフィルタ１に受信される。フィルタ１後の処理は復調処理部２１で復調され、Ｐ／Ｓ変換部２３によって復号信号Ｓ＾［ｎ］が得られる。ここで、拡散符号は、Ｗａｌｓｈ符号等の直交符号が適用される。直交符号の一種であるインパルス符号も適用可能であるが、パスダイバーシチ効果が最も発揮される変調ブロック内の両端付近のシンボルではＩＢＩに対する耐性が低く、時間軸拡散されないインパルス符号は、そのままでは時間軸相似等化の本来の性能を発揮できない。一方で、インパルス符号の低ＰＡＰＲ（ＰｅａｋｔｏＡｖｅｒａｇｅＰｏｗｅｒＲａｔｉｏ、送信信号の平均電力とピーク電力の比で与えられる）特性は送信機に要求される送信アンプの設計基準を緩和させるため、その点において有効である。そこで、後述するように誤り訂正符号の適用を行なうことにより時間軸上で情報信号を分散させて伝送することによってインパルス符号適用時にも上記の問題を解決可能となる。 The transmission system 13 is a system based on MC-DSSS, and includes a transmitter 14 and a receiver 15, and a signal transmitted from the transmitter 14 is received by the receiver 15 via the transmission path 2. Note that the RF processing unit is not shown unlike FIG. 20 described later. Since the information signal S [n] is modulated, it is converted into each information signal Si by the S / P converter 16 and multiplied by the modulation code vector ei by the multiplier 17. Then, the summation unit 19 performs the summation after multiplication, and transmits a vector X which is a transmission signal (modulated block signal). The transmission signal becomes a distorted vector Y through the transmission line 2, and after noise (vector n) is added, it is received by the filter 1 of the receiver 15 as a reception signal. The processing after the filter 1 is demodulated by the demodulation processing unit 21, and the decoded signal S ^ [n] is obtained by the P / S conversion unit 23. Here, an orthogonal code such as a Walsh code is applied as the spreading code. An impulse code which is a kind of orthogonal code is also applicable, but the symbol near the both ends in the modulation block where the path diversity effect is most exerted has low resistance to IBI. The original performance of similarity equalization cannot be demonstrated. On the other hand, the low PAPR (Peak to Average Power Ratio, given by the ratio of the average power and peak power of the transmission signal) characteristic of the impulse code relaxes the design criteria of the transmission amplifier required for the transmitter. It is valid. Therefore, by applying an error correction code as will be described later, the above-mentioned problem can be solved even when an impulse code is applied by distributing and transmitting information signals on the time axis.

図３を用いて、フィルタ１について説明する。時間軸相似等化を表す行列Ｈ_ＬＩは（１）式で与えられる伝送路行列ＨのようないわゆるToeplitz型のマトリクスとはならない。Toeplitz型のマトリクスならば図２９に示すようなひとつのトランスバーサルフィルタによって回路を実現できるが、Ｈ_ＬＩのような非Toeplitz型のマトリクスの場合、チップ毎に異なるトランスバーサルフィルタを用意せねばならない。すなわち、フィルタ１は、チップ毎に異なるトランスバーサルフィルタＥＦ_０〜ＥＦ_Ｎ-１を有し、Ｈ_ＬＩ＝（eij）とすると、左逆行列Ｈ_ＬＩによるフィルタ１は図３に示すようなフィルタバンク、具体的にはＦＩＲアレイフィルタによって構成される。図３に示すフィルタでは受信信号ベクトルＹとし、図２９に示すフィルタ２００ではベクトルＸとし、その点での相違は図１に示しているが、図２９に示す逆特性等化と図３に示す時間軸相似等化から、両者の構成は大きく相違し、時間軸相似等化回路における直並列変換器は時間Ｎ毎に実行され、出力ベクトルＲが出力される。繰り返しになるが、逆特性等化ではいずれのチップにおいても等しいＦＩＲフィルタが適用されるが、時間軸相似等化ではチップ毎に異なるＦＩＲフィルタが適用され、時間軸相似等化はＦＩＲアレイフィルタ構成をとる。直並列変換器は変調ブロック時間毎に出力され、r_0,r_1,..,r_N-1の順に時間軸上でシリーズに出力される。当該ＦＩＲアレイフィルタの各係数は逆拡散後の出力シンボル（図２におけるhat(S_i)のこと）の判定前と判定後の誤差成分の自乗を最小化するように適応制御することが可能である。これにより特異値分解等のマトリクス処理が不要となり回路構成が大幅に簡素化できる。 The filter 1 will be described with reference to FIG. The matrix H _LI representing time axis similarity equalization is not a so-called Toeplitz type matrix like the transmission line matrix H given by the equation (1). If Toeplitz-type matrix can be realized circuit by one of the transversal filter as shown in FIG. 29, in the case of a non-Toeplitz-type matrix, such as H _LI, it must be prepared with different transversal filter for each chip. That is, the filter 1 has different transversal filters EF _{0 to} EF _N-1 for each chip. When H _LI = (eij), the filter 1 based on the left inverse matrix H _LI is a filter bank as shown in FIG. Specifically, it is constituted by an FIR array filter. In the filter shown in FIG. 3, the received signal vector Y is used, and in the filter 200 shown in FIG. 29, the vector X is used. The difference in this respect is shown in FIG. 1, but the inverse characteristic equalization shown in FIG. Due to the time axis similarity equalization, the configurations of the two are greatly different. The serial-parallel converter in the time axis similarity equalization circuit is executed every time N, and an output vector R is output. To repeat, equal FIR filters are applied to all chips in inverse characteristic equalization, but different FIR filters are applied to each chip in time axis similarity equalization, and time axis similarity equalization is an FIR array filter configuration. Take. The serial-to-parallel converter is output every modulation block time, and is output to the series on the time axis in the order of r_0, r_1,..., R_N-1. Each coefficient of the FIR array filter can be adaptively controlled so as to minimize the square of the error component before and after the determination of the output symbol after despreading (hat (S_i) in FIG. 2). . This eliminates the need for matrix processing such as singular value decomposition, and can greatly simplify the circuit configuration.

図４は計算機シミュレーションの結果を示す図であって、図３及び図２９の各場合の各ＦＩＲフィルタのエネルギー伝達関数を示した図である。図４（Ａ）は伝送路のエネルギー伝達関数を示し、図４（Ｂ）は図２９の場合の逆特性等化フィルタのエネルギー伝達関数を示し、図４（Ｃ）は図３の場合の時間軸相似等化における各ＦＩＲフィルタのエネルギー伝達関数を示す。なお、伝送路は３２パスの指数減衰型遅延プロファイルモデルを採用した。１次変調はＱＰＳＫ、受信機における伝送路推定は誤差なく実施できるものとした。Ｎ＝３２とした。 FIG. 4 is a diagram showing a result of computer simulation, and is a diagram showing an energy transfer function of each FIR filter in each case of FIGS. 3 and 29. 4A shows the energy transfer function of the transmission line, FIG. 4B shows the energy transfer function of the inverse characteristic equalization filter in the case of FIG. 29, and FIG. 4C shows the time in the case of FIG. The energy transfer function of each FIR filter in axial similarity equalization is shown. The transmission path employs a 32-path exponential decay type delay profile model. The primary modulation is assumed to be QPSK, and the transmission path estimation at the receiver can be performed without error. N = 32.

図４において、逆特性等化では伝送路の逆特性となるフィルタが構成されるのに対して、時間軸相似等化の各ＦＩＲフィルタはチップ位置によって異なっている。なお、厳密にはチップｋとチップＮ−１−ｋのエネルギー伝達関数は等しい。変調ブロックの前後端付近のチップに対応するＦＩＲフィルタでは、逆特性等化のように伝送路の深い落ち込みを補正するようにゲインを持たせるのではなく、整合フィルタのように伝送路伝達特性に整合するフィルタの伝達特性を示している。変調ブロックの中央付近で、前後端から離れるシンボルに対するフィルタほどＺＦ等化のエネルギー伝達特性に近いフィルタとなっている。さらに説明すると、逆特性等化では伝送路の周波数特性上のノッチを等化するため、当該周波数成分に高いゲインを与えている様子が分かる。これは通常雑音強調を起こし、ＢＥＲ特性を劣化させる要因となる。これに対して本発明によるアレイフィルタの各ＦＩＲフィルタは、ノッチ部分を無理に等化しようとせず、整合フィルタ的な伝達特性を与えることでパスダイバーシチを実現する。 In FIG. 4, a filter having a reverse characteristic of the transmission path is configured in the inverse characteristic equalization, whereas each FIR filter for time axis similarity equalization differs depending on the chip position. Strictly speaking, the energy transfer functions of the chip k and the chip N-1-k are equal. The FIR filter corresponding to the chips near the front and rear ends of the modulation block does not give a gain so as to correct a deep drop in the transmission path as in the case of inverse characteristic equalization, but has a transmission path transmission characteristic like a matched filter. The transfer characteristics of the matched filter are shown. In the vicinity of the center of the modulation block, the filter for a symbol far from the front and rear ends is closer to the energy transfer characteristic of ZF equalization. To explain further, it can be seen that the inverse characteristic equalization equalizes notches on the frequency characteristic of the transmission line, so that a high gain is given to the frequency component. This usually causes noise enhancement and causes deterioration of the BER characteristics. On the other hand, each FIR filter of the array filter according to the present invention realizes path diversity by giving a transfer characteristic like a matching filter without forcibly equalizing the notch portion.

次に、時間軸相似等化の拡散符号種による特性差について検討する。拡散符号にはＷａｌｓｈ符号とインパルス符号を適用する。後者のインパルス符号とは、拡散符号ベクトルｓｉを図５で与える符号と定義し、当該符号を用いた拡散後の信号は無拡散変調のそれと一致する。チップ周期は50nsecとし、誤り訂正符号は適用しない。 Next, the characteristic difference depending on the spreading code type of time axis similarity equalization will be discussed. A Walsh code and an impulse code are applied to the spreading code. The latter impulse code is defined as a code that gives the spreading code vector si in FIG. 5, and a signal after spreading using the code matches that of non-spread modulation. The chip period is 50 nsec and no error correction code is applied.

図６は遅延スプレッド５０ｎｓｅｃとした場合、図７は遅延スプレッド２５０ｎｓｅｃとした場合の結果を示した図である。符号長Ｎ＝８とし、ＧＩ＝∞すなわちＩＢＩが無視できる場合とＧＩ＝１の場合を示した。いずれの図においてもＧＩ＝∞の場合はインパルス符号がＷａｌｓｈ符号に比べて若干良い特性を示しているが、ＧＩ＝１としＩＢＩが無視できない場合には両者の関係は逆転する。遅延スプレッドが大きいほどＩＢＩの影響が顕著となるため、ＧＩ＝∞とＧＩ＝１との特性差が大きく表れる。 FIG. 6 shows the results when the delay spread is 50 nsec, and FIG. 7 shows the results when the delay spread is 250 nsec. The code length N = 8, GI = ∞, that is, the case where IBI can be ignored and the case where GI = 1 are shown. In either figure, when GI = ∞, the impulse code shows slightly better characteristics than the Walsh code, but when GI = 1 and IBI cannot be ignored, the relationship between the two is reversed. Since the influence of IBI becomes more significant as the delay spread is larger, a characteristic difference between GI = ∞ and GI = 1 appears greatly.

図８は、Ｎ＝８のＷａｌｓｈ符号、ＧＩ＝∞とする場合に、遅延スプレッドを変化させて平均ＢＥＲ特性を調べた結果を示した図である。遅延スプレッドが大きくなるほどパスダイバーシチ効果により特性が向上する様子が分かる。さらに拡散符号長Ｎを変化させた場合の特性を評価する。 FIG. 8 is a diagram showing the result of examining the average BER characteristic by changing the delay spread when N = 8 Walsh code and GI = ∞. It can be seen that the characteristics are improved by the path diversity effect as the delay spread is increased. Further, the characteristics when the spreading code length N is changed are evaluated.

図９は、ＧＩ＝∞の場合と、伝送効率８／９となるようにＧＩを設定した場合の結果を示した図である。Ｎ＝８、１６、３２の場合について調べ、ＧＩ＝∞の場合のみＮ＝２と４の場合も加えて表示した。Ｗａｌｓｈ符号を用いた。ＧＩ＝∞の場合、拡散符号長Ｎを小さくするほどパスダイバーシチ効果が顕著となり特性が向上する。ところが、ＧＩを有限とする場合、逆にＮが小さいほど特性は劣化する。伝送効率を一定に保つ場合、Ｎが小さいほどＧＩ長は短くなるため、ＩＢＩマージンが減り、変調ブロック間の激しいＩＢＩにより特性が劣化するのである。以上の見てきたように、Ｎが小さいほどパスダイバーシチの効果が顕著になることが分かった。しかし、ＩＢＩが無視できない実環境ではその効果が損なわれてしまうことが分かった。 FIG. 9 is a diagram showing results when GI = ∞ and when GI is set so that the transmission efficiency is 8/9. The cases of N = 8, 16, 32 were examined, and the cases of N = 2 and 4 were also added only when GI = ∞. Walsh code was used. In the case of GI = ∞, as the spreading code length N is reduced, the path diversity effect becomes more prominent and the characteristics are improved. However, when the GI is finite, the characteristic is deteriorated as N is smaller. When the transmission efficiency is kept constant, the smaller the N is, the shorter the GI length is, so that the IBI margin is reduced and the characteristics deteriorate due to the intense IBI between the modulation blocks. As has been seen above, it has been found that the smaller the N is, the more prominent the path diversity effect is. However, it has been found that the effect is impaired in an actual environment where IBI cannot be ignored.

上記したように、Ｎが小さいほどパスダイバーシチの効果が顕著となるため、ＩＢＩを除去することができればパスダイバーシチ効果により高い伝送特性を達成可能であると考えられる。以下、本願の発明者が既に提案している特願２００６−１６５２４９号に記載のＩＢＩ除去を可能にする干渉キャンセラを示す（本願の発明者である古川氏の論文”符号の直交分離とパスダイバーシチを同時に実現する符号分割多重伝送,” 信学技法, RCS2006-52, pp.101-106, June 2006.参照）。 As described above, the effect of path diversity becomes more pronounced as N is smaller. Therefore, if IBI can be removed, it is considered that high transmission characteristics can be achieved by the path diversity effect. The interference canceller described in Japanese Patent Application No. 2006-165249 already proposed by the inventor of the present application will be described below (the paper of Dr. Furukawa, the inventor of the present application) “Code orthogonal separation and path diversity. Code division multiplex transmission, "Science Technique, RCS2006-52, pp.101-106, June 2006."

以下に、提案している手法を説明する。これまではＮ次の孤立した変調ブロックを前提にして説明を進めてきた。しかし、一つの変調ブロックは、通常、伝送路のひずみにより隣接変調ブロックへ干渉を及ぼす。ここで、ひとつの変調ブロックから発生する伝送路歪による当該変調ブロック区間外への遅延波成分を干渉ｔａｉｌと定義する。干渉ｔａｉｌの長さは伝送路の歪の程度に依存し、符号長Ｎには依存しない。すなわち、Ｎが長ければ長いほど相対的な変調ブロック間の干渉は減らすことができる。そこで、Ｎを干渉ｔａｉｌが次隣接ブロックまでのみ影響し、次々隣接ブロック以降は影響を及ぼさない程度に長く設定するものとすれば、図１０に示すように、ＩＢＩは、各変調ブロックに関し、ひとつ前の変調ブロックからの前置干渉（ＰｒｅＩＦ）とひとつ後の変調ブロックからの後置干渉（ＰｏｓｔＩＦ）のみを受ける。 The proposed method is described below. So far, the description has been made on the assumption of an Nth-order isolated modulation block. However, one modulation block usually interferes with adjacent modulation blocks due to transmission path distortion. Here, a delayed wave component outside the modulation block section due to transmission path distortion generated from one modulation block is defined as an interference tail. The length of the interference tail depends on the degree of distortion of the transmission path, and does not depend on the code length N. That is, the longer N is, the more interference between relative modulation blocks can be reduced. Therefore, if N is set long enough that the interference tail affects only the next adjacent block and does not affect the subsequent adjacent blocks, as shown in FIG. 10, the IBI is one for each modulation block. Only pre-interference (Pre IF) from the previous modulation block and post-interference (Post IF) from the next modulation block are received.

これらの干渉を軽減するためには、変調ブロック間にＯＦＤＭのようなガードインターバルを緩衝区間として設ける手法が考えられるが、ガードインターバルの区間は情報信号を伝送できないためデータ伝送効率の低下が生じる。そこで、ガードインターバルを設けず、復号前信号からその前後の変調ブロックの復号結果を再変調して干渉成分を差し引く手法を説明する。干渉除去の性能を高めるため、変調ブロックを受信する都度、数ブロック前に遡って復号を繰り返し行なうことで、より高い干渉除去性能を引き出す復号法となる。 In order to reduce such interference, a method of providing a guard interval such as OFDM as a buffer interval between modulation blocks can be considered. However, since an information signal cannot be transmitted in the guard interval interval, data transmission efficiency is reduced. Therefore, a method will be described in which a guard interval is not provided and the decoding result of the preceding and subsequent modulation blocks is remodulated from the pre-decoding signal to subtract the interference component. In order to improve the performance of interference cancellation, every time a modulation block is received, the decoding is repeated retroactively several blocks before, thereby providing a decoding method that draws out higher interference cancellation performance.

図１１は、復調処理における干渉キャンセラの原理を説明するための図である。 FIG. 11 is a diagram for explaining the principle of the interference canceller in the demodulation process.

図１１を参照して、時刻ｔ1において信号ブロックＳＣが受信されたとする。受信機は直ちにＳＣの復号を行ない、復号結果ＳＣ＿１を得る。このとき、ＳＣがその直前の受信ブロックＳＢから受けるＰｒｅＩＦ成分をすでに復号済みの結果ＳＢ＿１を用いて除去する。具体的にはＳＢ＿１を再変調し、受信機側で把握している伝送路行列を用いてＰｒｅＩＦ成分を推定し、ＳＣの復号前信号から差し引く。ＳＣ＿１の復号後、ただちにＳＢの再復号を行なう。その際、ＳＢに隣接するブロックの復号結果ＳＡ＿２及びＳＣ＿１を用いる。具体的には、ＳＡ＿２からＰｒｅＩＦ成分、ＳＣ＿１からＰｏｓｔＩＦ成分を求め、ＳＢの復号前信号からそれぞれ差し引いた後、復号しＳＢ＿２を得る。同様にＳＡの再々復号をその隣接するブロックに対する復号結果から実施しＳＡ＿３を得る。図１１では、以上の処理を４つ前の変調ブロックまで遡って実施した様子を示している。 Referring to FIG. 11, assume that signal block SC is received at time t1. The receiver immediately decodes the SC and obtains a decoding result SC_1. At this time, the Pre IF component received by the SC from the immediately preceding reception block SB is removed using the already decoded result SB_1. Specifically, SB_1 is remodulated, the Pre IF component is estimated using the transmission path matrix known on the receiver side, and is subtracted from the signal before decoding of SC. Immediately after decoding SC_1, SB is re-decoded. At that time, the decoding results SA_2 and SC_1 of the blocks adjacent to the SB are used. Specifically, the Pre IF component is obtained from SA_2, and the Post IF component is obtained from SC_1. After subtracting from the pre-decoding signal of SB, decoding is performed to obtain SB_2. Similarly, SA re-decoding is performed from the decoding result for the adjacent block to obtain SA_3. FIG. 11 shows a state in which the above processing is performed retroactively to the previous modulation block.

ＳＣ＿１復号時にはその直前の変調ブロックＳＢからのＰｒｅＩＦ成分のみが除去されるが、ＳＣの２回目の復号時ＳＣ＿２では、前後の変調ブロックからのＰｒｅＩＦ成分ならびにＰｏｓｔＩＦ成分が除去された上で復号が行なわれるため、ＳＣ＿１復号時よりも干渉が低減される。さらにＳＣ＿３復号時には、ＳＣ＿２復号時に用いられた復号結果（ＳＤ＿１及びＳＢ＿２）よりも誤りが低減された復号結果（ＳＤ＿２及びＳＢ＿３）が干渉除去に用いられるため、ＳＣ＿２復号時よりさらに正確な干渉除去が可能になる。これを繰り返すことで復号誤りが低減すると考えられる。 At the time of SC_1 decoding, only the Pre IF component from the immediately preceding modulation block SB is removed, but at the time of SC_2 decoding SC_2, the Pre IF component and the Post IF component from the preceding and succeeding modulation blocks are removed. Since decoding is performed, interference is reduced compared to SC_1 decoding. Furthermore, at the time of SC_3 decoding, the decoding results (SD_2 and SB_3) in which errors are reduced compared to the decoding results (SD_1 and SB_2) used at the time of SC_2 decoding are used for interference cancellation. It becomes possible. It is considered that decoding errors are reduced by repeating this.

なお、上述のようなガードインターバル（ＧＩ）を挿入せずに繰り返し復号による干渉除去をＯＦＤＭへ適用することは困難である。なぜなら、ＯＦＤＭにとってＧＩは、ＩＢＩを抑制するためだけでなく、ＧＩ内へＣＰを充填することで同一ブロック内のサブキャリア間の直交性を維持するために必須であり、ＣＰ充填を行わないと同一ブロック内でもサブキャリア間干渉（ＩｎｔｅｒＣｈａｎｎｅｌＩｎｔｅｒｆｅｒｅｎｃｅ，ＩＣＩ）が発生するからである。 It is difficult to apply interference cancellation by iterative decoding to OFDM without inserting a guard interval (GI) as described above. This is because, for OFDM, GI is essential not only for suppressing IBI but also for maintaining orthogonality between subcarriers in the same block by filling CP in GI. This is because inter-subcarrier interference (ICI) occurs even within the same block.

図１２は拡散符号に対して平均ＢＥＲ特性がどのように変化するかを調べた結果を示す図である。 FIG. 12 is a diagram showing the results of examining how the average BER characteristic changes with respect to the spreading code.

拡散符号としてＷａｌｓｈ符号及びインパルス符号を用い、Ｎ＝８、ＧＩ＝１とした。図６及び図７で示したように、ＧＩ＝∞としＩＢＩが無視できる場合、インパルス符号のほうがＷａｌｓｈ符号より若干低いＢＥＲ特性が得られていた。それに対し、図１２に示したＩＢＩが発生する下で干渉キャンセラを適用する場合、Ｗａｌｓｈ符号を適用したほうが良い特性が得られている。ＩＢＩの影響は変調ブロックの前後端に集中する。Ｗａｌｓｈ符号を用いる場合、逆拡散処理時に全ての符号に均等にＩＢＩが分散されるが、インパルス符号の場合、特定の符号、すなわち変調ブロックの前後端付近の符号にＩＢＩが集中する。悪しくもこれらの符号は、ＩＢＩを最も受ける符号であると同時に隣接する変調ブロックへＩＢＩを最も与える符号でもある。干渉キャンセラが良好に機能するためにはもっとも正確に生成されなければならない符号群である。以上の理由によりインパルス符号を適用する場合には干渉キャンセラの効果が小さくなると考えられる。 Walsh code and impulse code were used as spreading codes, and N = 8 and GI = 1. As shown in FIGS. 6 and 7, when GI = ∞ and IBI can be ignored, the BER characteristic of the impulse code is slightly lower than that of the Walsh code. On the other hand, when the interference canceller is applied under the occurrence of the IBI shown in FIG. 12, a better characteristic is obtained when the Walsh code is applied. The influence of IBI is concentrated on the front and rear ends of the modulation block. When the Walsh code is used, the IBI is uniformly distributed to all codes during the despreading process. However, in the case of the impulse code, the IBI is concentrated on a specific code, that is, a code near the front and rear ends of the modulation block. Worse, these codes are the codes that receive the most IBI and the codes that give the most IBI to the adjacent modulation block. It is a code group that must be generated most accurately in order for the interference canceller to function well. For the above reason, it is considered that the effect of the interference canceller is reduced when the impulse code is applied.

図１３〜図１５は、Ｗａｌｓｈ符号を適用した場合に拡散符号長を変化させて干渉キャンセラの特性を調べた結果を示す図である。図１３はＮ＝８、Ｗａｌｓｈｃｏｄｅの場合の干渉キャンセラの平均ＢＥＲ特性を示した図であり、図１４はＮ＝１６、Ｗａｌｓｈｃｏｄｅの場合の干渉キャンセラの平均ＢＥＲ特性を示した図であり、図１５はＮ＝３２、Ｗａｌｓｈｃｏｄｅの場合の干渉キャンセラの平均ＢＥＲ特性を示した図である。ここで、ＧＩ長が８／９となるように適宜設定した。遅延スプレッドはいずれも５０ｎｓｅｃである。各図には、比較のためＩＢＩが無視できる場合とＩＢＩが無視できずかつ干渉キャンセラを適用しない場合とを示した。 FIGS. 13 to 15 are diagrams illustrating the results of examining the characteristics of the interference canceller by changing the spreading code length when the Walsh code is applied. FIG. 13 is a diagram showing an average BER characteristic of an interference canceller when N = 8 and Walsh code, and FIG. 14 is a diagram showing an average BER characteristic of the interference canceller when N = 16 and Walsh code. FIG. 15 is a diagram showing an average BER characteristic of the interference canceller in the case of N = 32 and Walsh code. Here, the GI length was appropriately set to be 8/9. Both delay spreads are 50 nsec. In each figure, the case where IBI can be ignored and the case where IBI cannot be ignored and no interference canceller is applied are shown for comparison.

図１３〜図１５に示すように、干渉キャンセラを適用することによっていずれの拡散符号長Ｎの場合にもＩＢＩが無視できる場合の特性に近い特性が達成されていることが分かる。Ｎが大きいほど干渉キャンセラの特性はＩＢＩが無視できる場合に近い特性が達成されている。干渉キャンセラを適用しない場合の平均ＢＥＲ特性はＮが小さいほど高いが、これは伝送効率を８／９と一定にしたためにＮが小さいほどＧＩ長が短くなりＩＢＩの影響が顕著となったためである。 As shown in FIGS. 13 to 15, it can be seen that by applying the interference canceller, a characteristic close to the characteristic when IBI can be ignored is achieved for any spreading code length N. The larger N is, the closer the interference canceller characteristic is to the case where IBI can be ignored. When the interference canceller is not applied, the average BER characteristic is higher as N is smaller. This is because the transmission efficiency is constant at 8/9, so that the smaller the N is, the shorter the GI length becomes and the influence of IBI becomes remarkable. .

図１６は遅延スプレッドを２５０ｎｓｅｃとした場合の干渉キャンセラの特性を調べた結果を示す図である。なお、Ｎ＝３２、ＧＩ＝４のＷａｌｓｈ符号を適用した。 FIG. 16 is a diagram showing the results of examining the characteristics of the interference canceller when the delay spread is 250 nsec. A Walsh code with N = 32 and GI = 4 was applied.

干渉キャンセラを適用することでＩＢＩが無視できる場合の特性に近い特性が達成されていることが分かる。図１３〜図１５の結果と図１６の結果を比較すると、後者のほうがより低い平均ＢＥＲ特性を達成していることが分かる。これは遅延スプレッドが長くなったことによりパスダイバーシチの効果がより高く表れた結果を表している。 It can be seen that by applying the interference canceller, a characteristic close to the characteristic when IBI can be ignored is achieved. Comparing the results of FIGS. 13 to 15 with the results of FIG. 16, it can be seen that the latter achieves a lower average BER characteristic. This represents the result that the effect of the path diversity is more enhanced due to the longer delay spread.

図１７は、ＯＦＤＭとの比較のため、拡散符号Ｎを変化させた場合のＷａｌｓｈ符号を適用した時間軸相似等化とＮ＝３２、ＧＩ＝４のＯＦＤＭとを比較した結果を示した図である。なお、時間軸相似等化はいずれも干渉キャンセラを適用し、ＧＩは伝送効率が８／９となるように各Ｎに対して適宜設定を行なった。遅延スプレッドは５０ｎｓｅｃである。 FIG. 17 is a diagram showing a result of comparison between time-axis similarity equalization using a Walsh code when the spreading code N is changed and OFDM with N = 32 and GI = 4 for comparison with OFDM. is there. Note that in all the time-axis similarity equalization, an interference canceller was applied, and GI was appropriately set for each N so that the transmission efficiency was 8/9. The delay spread is 50 nsec.

図１７のように、Ｎ＝８の時間軸相似等化が最もよい特性を示している。ＩＢＩが無視できないにもかかわらず、このような結果が得られたことは、干渉キャンセラによってＩＢＩがよく抑制され、Ｎが小さい場合のパスダイバーシチの効果が発揮された結果と考えられる。ＯＦＤＭとの比較においては、平均ＢＥＲ＝１ｅ−３の所要Ｅｂ／Ｎ０を比較すると、Ｎ＝８の提案方式はＯＦＤＭに比べておよそ９ｄＢの所要Ｅｂ／Ｎ０の低減を達成している。 As shown in FIG. 17, the time axis similarity equalization with N = 8 shows the best characteristic. The fact that such a result was obtained even though IBI cannot be ignored is considered to be a result of the IBI being well suppressed by the interference canceller and the effect of path diversity when N is small. In comparison with OFDM, comparing the required Eb / N0 with an average BER = 1e-3, the proposed scheme with N = 8 achieves a reduction in required Eb / N0 of approximately 9 dB compared to OFDM.

図１８は上記動作を行うキャンセラの内部構成を示した図である。図１９は図１８のキャンセラの動作を説明するためのフロー図である。以下、図１１を用いて説明した原理についてハード及びソフトの面から説明をさらに行う。 FIG. 18 is a diagram showing an internal configuration of a canceller that performs the above operation. FIG. 19 is a flowchart for explaining the operation of the canceller of FIG. Hereinafter, the principle described with reference to FIG. 11 will be further described in terms of hardware and software.

キャンセラ２９は、図１におけるフィルタ１よりも伝送路（Ｈ）２側に配置され、Ｍシンボルサンプリング部からの受信信号を受信する受信部２９ａと、再復号部２９ｂと、制御部２９ｃと、ブロックバッファ部２９ｄとを備える。ブロックバッファ部２９ｄは、受信した変調ブロック信号を記録する受信変調ブロックバッファと、受信した各変調ブロック信号に対する復号結果を記録する復号ブロックバッファを有する。受信変調ブロックバッファは復号前の受信したアナログ信号をバッファリングする。復号ブロックバッファは受信変調ブロックからディジタル復号ビットを格納する。以下、図１９を用いて、動作を説明する。 The canceller 29 is arranged on the transmission line (H) 2 side of the filter 1 in FIG. 1, and receives a reception unit 29a that receives a reception signal from the M symbol sampling unit, a re-decoding unit 29b, a control unit 29c, and a block And a buffer unit 29d. The block buffer unit 29d has a reception modulation block buffer for recording the received modulation block signal and a decoding block buffer for recording a decoding result for each received modulation block signal. The reception modulation block buffer buffers the received analog signal before decoding. The decoding block buffer stores digital decoded bits from the received modulation block. Hereinafter, the operation will be described with reference to FIG.

ステップＳＴ１において、新たな変調ブロックが受信されたことが受信部２９ａにより判定される。受信されると、ステップＳＴ２において、当該受信された変調ブロックＡが受信変調ブロックバッファに格納され、復号ブロックバッファに記録済みである変調ブロックＡの直前に受信された変調ブロックＢの最新の復号結果が再変調されさらに伝送路に相当するフィルタをかけることで変調ブロックＡが変調ブロックＢから受ける前置干渉成分が求められ、当該前置干渉成分が変調ブロックＡから除去されて復号された復号結果が復号ブロックバッファに格納され、受信変調ブロックバッファ内の変調ブロックのうち２番目に新しい変調ブロックが再復号対象変調ブロックに設定される。ステップＳＴ３において、再復号対象変調ブロックに対して、当該再復号対象変調ブロックの直前に受信された変調ブロックの最新の復号結果が復号ブロックバッファから読み出されて再変調されさらに伝送路に相当するフィルタをかけることで得られる前置干渉成分及び当該再復号対象変調ブロックの直後に受信された変調ブロックの最新の復号結果が復号ブロックバッファから読み出されて再変調されさらに伝送路に相当するフィルタをかけることで得られる後置干渉成分が除去されて復号された復号結果が復号ブロックバッファに格納される。そして、ステップＳＴ５において、受信変調ブロックバッファに格納されている再復号対象変調ブロックの直前に受信された変調ブロックが新たな再復号対象変調ブロックに設定される更新処理が行われる。このステップＳＴ４及びステップＳＴ６の処理は、所望の回数繰り返して行われる。 In step ST1, the receiving unit 29a determines that a new modulation block has been received. When received, in step ST2, the received modulation block A is stored in the reception modulation block buffer, and the latest decoding result of the modulation block B received immediately before the modulation block A recorded in the decoding block buffer. Is decoded, and a filter corresponding to the transmission path is applied to obtain a pre-interference component received by the modulation block A from the modulation block B, and the pre-interference component is removed from the modulation block A and decoded. Is stored in the decoding block buffer, and the second newest modulation block among the modulation blocks in the reception modulation block buffer is set as the modulation block to be re-decoded. In step ST3, with respect to the re-decoding target modulation block, the latest decoding result of the modulation block received immediately before the re-decoding target modulation block is read from the decoding block buffer, re-modulated, and further corresponds to the transmission path. A filter corresponding to a transmission path, in which the pre-interference component obtained by applying the filter and the latest decoding result of the modulation block received immediately after the modulation block to be re-decoded are read out from the decoding block buffer and remodulated. The decoding result obtained by removing the post-interference component obtained by applying is applied to the decoding block buffer. In step ST5, an update process is performed in which the modulation block received immediately before the re-decoding target modulation block stored in the reception modulation block buffer is set as a new re-decoding target modulation block. The processes in step ST4 and step ST6 are repeated a desired number of times.

図２０は本発明の実施の形態に係るＭＩＭＯ形式の符号分割多重伝送システムを示す図である。 FIG. 20 is a diagram showing a MIMO code division multiplexing transmission system according to an embodiment of the present invention.

このＭＩＭＯ形式の符号分割多重伝送システム３０は、送信機３０ａに、変調処理部３１と、アップコンバータ３３−１〜３３−ｎと、ＲＦ３５−１〜３５−ｎと、送信アンテナＴＸ＿ＡＮＴ−１〜ＴＸ＿ＡＮＴ−ＮＴとを備える。また、符号分割多重伝送システム３０は、受信機３０ｂに、受信アンテナＲＸ＿ＡＮＴ−１〜ＲＸ＿ＡＮＴ−ＮＲと、ＬＮＡ＆受信フィルタ３７−１〜３７−ｎと、ダウンコンバータ３９−１〜３９−ｎと、復調処理部４０とを備える。送信機３０ａから送信された信号は、伝送路４１を介して受信機３０ｂにより受信される。 The MIMO code division multiplexing transmission system 30 includes a transmitter 30a, a modulation processing unit 31, up-converters 33-1 to 33-n, RFs 35-1 to 35-n, and transmission antennas TX_ANT-1 to TX_ANT. -NT. Further, the code division multiplexing transmission system 30 includes a receiver 30b, reception antennas RX_ANT-1 to RX_ANT-NR, LNA & reception filters 37-1 to 37-n, down converters 39-1 to 39-n, and demodulation. And a processing unit 40. The signal transmitted from the transmitter 30a is received by the receiver 30b via the transmission path 41.

図２１は図２０の伝送路４１の伝送路行列Ｈを説明するための図である。 FIG. 21 is a diagram for explaining the transmission path matrix H of the transmission path 41 of FIG.

伝送路４１は歪ＭＩＭＯ伝送路であり、全体の伝送路行列は送信アンテナＴＸ＿ＡＮＴ−１〜ＴＸ＿ＡＮＴ−ＮＴと受信アンテナＲＸ＿ＡＮＴ−１〜ＲＸ＿ＡＮＴ−ＮＲ間のそれぞれの伝送路行列Ｈijにより形成され、伝送路行列ＨijはＭ行Ｎ列の伝送路行列である。（１）式の伝送路行列Ｈに対応するＭＩＭＯの全体の伝送路行列Ｈ_Ｍは図２２に示すようになる。 The transmission line 41 is a distorted MIMO transmission line, and the entire transmission line matrix is formed by transmission line matrices Hij between the transmission antennas TX_ANT-1 to TX_ANT-NT and the reception antennas RX_ANT-1 to RX_ANT-NR. The matrix Hij is a transmission line matrix of M rows and N columns. (1) channel matrix H _M of the overall MIMO corresponding to the channel matrix H of the expression is as shown in FIG. 22.

図２３は図２０の変調処理部３１の内部構成を示した図である。 FIG. 23 is a diagram showing an internal configuration of the modulation processing unit 31 of FIG.

変調処理部３１は、Ｓ／Ｐ変換部３１１と、乗算部３１３と、総和部３１５と、波形整形フィルタ３１７−１〜３１７−ｎとを備える。乗算部３１３では、Ｓ［ｉ］にＷａｌｓｈ符号などの直交符号が拡散符号Ｅ_Ｍ＿ｉ(ただしｉ＝０〜ＮＴ＊Ｎ−１)として選択されて乗算される。ここで、Ｅ_Ｍ＿ｉはＤ＝Ｈ_Ｍ ^ＨＨ_Ｍの固有ベクトルであり、Ｄは（Ｎ＊ＮＴ）行（Ｎ＊ＮＴ）列の正方行列である。総和部３１５により得られるＸ_Ｍ［ｎ］は下記の（７）式で与えられる。 The modulation processing unit 31 includes an S / P conversion unit 311, a multiplication unit 313, a summation unit 315, and waveform shaping filters 317-1 to 317-n. In the multiplication unit 313, an orthogonal code such as a Walsh code is selected as the spreading code E _{M —} i (where i = 0 to NT * N−1) and multiplied by S [i]. Here, E _{M —} i is an eigenvector of D = H _M ^H H _M , and D is a square matrix of (N * NT) rows (N * NT) columns. X _M [n] obtained by the summation unit 315 is given by the following equation (7).

Ｘ_Ｍ＿ｉは、ｉ番目アンテナの出力シンボル系列を表すＮ次の行ベクトルである。この変調処理部３１により、情報信号に時間ブロック符号を乗じて送信アンテナのそれぞれに割り当てが行われ、並列送信が可能になっている。 X _{M —} i is an N-th row vector representing an output symbol sequence of the i-th antenna. The modulation processing unit 31 multiplies the information signal by a time block code and assigns it to each of the transmission antennas, thereby enabling parallel transmission.

図２４は図２０の復調処理部４０の内部構成を示した図である。 FIG. 24 is a diagram showing an internal configuration of the demodulation processing unit 40 of FIG.

復調処理部４０は、Ｍシンボルサンプリング部４０１−１〜４０１−ｎと、結合部（Ｃｏｍｂｉｎｅｒ）４０３と、フィルタ４０５と、内積部４０７とを備える。ここで、Ｒ＿ｉ（ｉ＝１〜ＮＲ）はＭ次の行ベクトルであり、Ｒ_ＭはＭ＊ＮＲ次の行ベクトルである。Ｒ_ＭはＨ_Ｍ＊Ｘ_Ｍで与えられる。ここでＸ_Ｍは（Ｘ_Ｍ＿１^Ｔ，ＸＭ＿２^Ｔ，・・・，Ｘ_Ｍ＿ＮＴ^Ｔ）^Ｔである。Ｒ_Ｍ＿ｆは、図２５に示す随伴行列Ｈ_Ｍ ^Ｈを用い、さらに（６）式による左逆行列としたＨ_Ｍ，ＬＩによるフィルタ４０５により得られ、Ｎ＊ＮＴ次の行ベクトルである。そして、内積部４０７において、入力されるＩＮベクトルは、すべてＮ＊ＮＴ次ベクトルである。ここで、結合部４０３は、各送信アンテナが送信した情報を受信アンテナが別々に受信しており、その並列に受信した受信信号を順番並べて統合している。 The demodulation processing unit 40 includes M symbol sampling units 401-1 to 401-n, a combining unit 403, a filter 405, and an inner product unit 407. Here, R_i (i = 1~NR) is M next row vector, _{R M} is M * NR next row vector. R _M is given by H _M * X _M. Here, X _M is (X _{M —} 1 ^T , XM — 2 ^T ,..., X _M —NT ^T ) ^T. R _M _f uses the adjoint matrix _H ^{M H} shown in FIG. 25, further (6) _{H M} was left inverse matrix by _{an equation} obtained by the filter 405 by _LI, is N * NT following row vector. In the inner product unit 407, all input IN vectors are N * NT-order vectors. Here, combining section 403 receives information transmitted by each transmitting antenna separately by the receiving antenna, and integrates the received signals received in parallel.

なお、上記したキャンセラがＭＩＭＯの場合にも導入されたときは、図２４のＭシンボルサンプリング部４０１−１〜４０１−ｎと、結合部（Ｃｏｍｂｉｎｅｒ）４０３との間に、各Ｍシンボルサンプリング部４０１−１〜４０１−ｎに対応して配置される。すなわち、各アンテナが受信した受信変調ブロックから前置干渉成分ならびに後置干渉成分をキャンセラにより除去する。前置干渉成分ならびに後置干渉成分の推定による干渉除去は、受信ブロックバッファに対して行われる。そのため、受信変調ブロックバッファは各アンテナに必要とされる。干渉が除去された後に、結合部４０３によって統合され、復号が行われる。 When the canceller described above is also introduced in the case of MIMO, each M symbol sampling unit 401 is interposed between the M symbol sampling units 401-1 to 401-n and the combining unit (combiner) 403 in FIG. It arrange | positions corresponding to -1 to 401-n. That is, the front interference component and the rear interference component are removed from the reception modulation block received by each antenna by the canceller. Interference cancellation by estimating the front interference component and the rear interference component is performed on the reception block buffer. Therefore, a reception modulation block buffer is required for each antenna. After the interference is removed, integration is performed by the combining unit 403 and decoding is performed.

図２６は上記実施の形態に対して自乗誤差最小規範に基づく適応フィルタ係数制御を行う発明を説明するためのブロック図である。当該適応フィルタ制御は、図１〜図１９で説明した非ＭＩＭＯ型の本願発明（キャンセラの有無を問わない）、また図２０〜図２５で説明したＭＩＭＯ型の本願発明（キャンセラの有無を問わない）、いずれにも適用される。 FIG. 26 is a block diagram for explaining the invention for performing adaptive filter coefficient control based on the minimum square error criterion for the above embodiment. The adaptive filter control is the non-MIMO type invention of the present invention described with reference to FIGS. 1 to 19 (with or without canceller), and the MIMO type of present invention described with reference to FIGS. 20 to 25 (with or without canceller). ), Both apply.

ここでは、自乗誤差最小規範に基づく適応フィルタ係数制御を行うべく、制御部５０は、例えば図３に示したフィルタ１に対応するＦＩＲアレイフィルタ５１の出力の値を判定する判定部５３と、ＦＩＲアレイフィルタ５１の出力と判定部５３の判定結果との誤差を演算する自乗誤差部５５と、係数制御部５７とを備える。ＦＩＲアレイフィルタ出力の判定結果と判定前結果との自乗誤差が自乗誤差部５５により計算され、係数制御部５７が当該自乗誤差を最小とするようにＦＩＲフィルタアレイの各フィルタ重みを制御する。当該適応制御機構により、上記したＨ、Ｖ、Λの事前把握の必要なくＦＩＲアレイフィルタの係数を設定可能となり、動的伝送路環境にも適用可能となり、ハードウェア規模を大幅に削減できる。 Here, in order to perform adaptive filter coefficient control based on the minimum square error criterion, the control unit 50 includes, for example, a determination unit 53 that determines an output value of the FIR array filter 51 corresponding to the filter 1 illustrated in FIG. A square error unit 55 for calculating an error between the output of the array filter 51 and the determination result of the determination unit 53 and a coefficient control unit 57 are provided. A square error between the determination result of the FIR array filter output and the result before determination is calculated by the square error unit 55, and the coefficient control unit 57 controls each filter weight of the FIR filter array so as to minimize the square error. With this adaptive control mechanism, the coefficients of the FIR array filter can be set without the need for prior understanding of the above-described H, V, and Λ, which can be applied to a dynamic transmission line environment, and the hardware scale can be greatly reduced.

さらに説明する。制御部５０は、各要素フィルタＥＦ＿ｉをそれぞれＮＲ＊Ｍ次のＦＩＲフィルタで与え、当該ＦＩＲフィルタの各タップ係数をｆ＿ｉ，ｊ（ｊ＝０〜ＮＲ＊Ｍ−１）とし、ｆ＿ｉ，ｊをｉ行ｊ列の要素とする行列Ｇ（（ＮＴ＊Ｎ）行（ＮＲ＊Ｍ）列の行列）を定め、ブロック変調信号Ｘとして受信側で既知の信号系列ＸＫを伝送し、時系列信号シンボルＲと信号系列ＸＫの誤差の自乗平均が最小となるように行列Ｇの各要素を制御する。或いは、制御部５０は、各要素フィルタＥＦ＿ｉをそれぞれＮＲ＊Ｍ次のＦＩＲフィルタで与え、当該ＦＩＲフィルタの各タップ係数をｆ＿ｉ，ｊ（ｊ＝０〜ＮＲ＊Ｍ−１）とし、ｆ＿ｉ，ｊをｉ行ｊ列の要素とする行列Ｇ（（ＮＴ＊Ｎ）行（ＮＲ＊Ｍ）列の行列）を定め、等化後の時系列信号シンボルＲの系列を復号し、得られる復号信号系列Ｒ＿Ｈと時系列信号シンボルＲの系列との誤差の自乗平均が最小となるように行列Ｇの各要素を制御する。 Further explanation will be given. The control unit 50 gives each element filter EF_i by an NR * M-th order FIR filter, sets each tap coefficient of the FIR filter to f_i, j (j = 0 to NR * M−1), and sets f_i, j to i A matrix G (matrix of (NT * N) rows (NR * M) columns) is defined as an element of row j column, a known signal sequence XK is transmitted as a block modulation signal X on the receiving side, and a time series signal symbol R And each element of the matrix G is controlled so that the mean square of the error of the signal series XK is minimized. Alternatively, the control unit 50 gives each element filter EF_i by an NR * M-th order FIR filter, sets each tap coefficient of the FIR filter to f_i, j (j = 0 to NR * M−1), and sets f_i, j Is a matrix G (a matrix of (NT * N) rows (NR * M) columns), and a sequence of time-series signal symbols R after equalization is decoded and a decoded signal sequence obtained Each element of the matrix G is controlled so that the mean square of the error between R_H and the sequence of the time series signal symbol R is minimized.

図２７は符号化と復号化の処理を含む伝送の全体を説明する図であって誤り訂正符号の適用も説明するための図である。 FIG. 27 is a diagram for explaining the entire transmission including encoding and decoding processes, and also for explaining the application of error correction codes.

時間軸相似等化において、直交符号の一種であるインパルス符号を適用する場合、パスダイバーシチ効果が最も発揮される変調ブロック内両端付近のシンボルではＩＢＩに対する耐性が低く、時間軸拡散されないインパルス符号は、そのままでは時間軸相似等化の本来の性能を発揮できない。一方で、インパルス符号の低ＰＡＰＲ（Peak to Average Power Ratio)特性は送信機に要求される送信アンプの設計基準を緩和させるため、その点において有効である。そこで、誤り訂正符号の適用を行なうことにより時間軸上で情報信号を分散させて伝送することでインパルス符号適用時にも上記の問題を解決可能となる。図２７では、この場合の伝送系のブロック図が示され、符号化された情報信号はＭＣ−ＤＳＳＳ変調が行なわれた後送信され、伝送路を経由し受信した信号は時間軸相似等化を経て、復号処理がなされる。ＭＣ−ＤＳＳＳにおける拡散符号には、インパルス符号のみならずＷａｌｓｈ符号等、他の拡散符号を用いることも可能である。 When applying an impulse code which is a kind of orthogonal code in time-axis similarity equalization, symbols near both ends in the modulation block where the path diversity effect is most exerted have low resistance to IBI, and the impulse code that is not time-axis spread is As it is, the original performance of time axis similarity equalization cannot be demonstrated. On the other hand, the low PAPR (Peak to Average Power Ratio) characteristic of the impulse code is effective in that respect because it relaxes the design criteria of the transmission amplifier required for the transmitter. Therefore, by applying the error correction code, the above problem can be solved even when the impulse code is applied by transmitting the information signal in a distributed manner on the time axis. FIG. 27 shows a block diagram of the transmission system in this case, and the encoded information signal is transmitted after MC-DSSS modulation is performed, and the signal received via the transmission path is time-axis-similarized. After that, decryption processing is performed. As the spreading code in MC-DSSS, it is possible to use not only an impulse code but also other spreading codes such as a Walsh code.

最後に、簡単にまとめる。通常、有限時間信号に対する等化問題は、周期関数近似やゼロ充填近似によって無限時間信号に置き換えることで検討がなされるが、本発明では時間軸上のマトリクス展開によって議論を進め、その結果、変調ブロックの各チップ毎に異なるフィルタ回路を設置するバンクフィルタによってこの目的が達せられることを示した。この当該フィルタバンクによってパスダイバーシチの効果が得られる。計算機シミュレーションによりパスダイバーシチの効果も確認された。ブロック変調信号が時間的に連続して伝送される場合、伝送路歪による干渉ｔａｉｌによってブロック間干渉が発生するが、これを除去するための干渉キャンセラを提案し、その有効性も明らかにした。また、ＭＩＭＯへの適用や適応フィルタ係数制御、インパルス符号の場合の誤り符合を適用することによる工夫も示した。 Finally, let me summarize it briefly. Normally, the equalization problem for a finite-time signal is studied by replacing it with an infinite time signal by periodic function approximation or zero-filling approximation, but in the present invention, discussion proceeds by matrix expansion on the time axis, and as a result, modulation is performed. It has been shown that this goal can be achieved by bank filters with different filter circuits for each chip of the block. This filter bank provides a path diversity effect. The effect of path diversity was also confirmed by computer simulation. When block modulated signals are transmitted continuously in time, inter-block interference occurs due to interference tails caused by transmission path distortion. We proposed an interference canceller to remove this interference and clarified its effectiveness. Moreover, the device by applying the error code in the case of application to MIMO, adaptive filter coefficient control, and impulse code was also shown.

時間軸相似等化（ＦｉｎｉｔｅＷａｖｅｆｏｒｍＰｒｅｓｅｒｖａｔｉｏｎＦｉｌｔｅｒ, ＦＷＰＦ)と呼ぶ等化を実現するフィルタを説明するための図である。It is a figure for demonstrating the filter which implement | achieves equalization called a time-axis similarity equalization (Finite Waveform Preservation Filter, FWPF). 図１のフィルタを用いた伝送システムの概略を示す図である。It is a figure which shows the outline of the transmission system using the filter of FIG. 図１のフィルタ１の構成をさらに具体的に示すブロック図である。It is a block diagram which shows more concretely the structure of the filter 1 of FIG. 計算機シミュレーションの結果を示す図であって、図３及び図２９の各場合の各ＦＩＲフィルタのエネルギー伝達関数を示した図である。It is a figure which shows the result of a computer simulation, Comprising: It is the figure which showed the energy transfer function of each FIR filter in each case of FIG.3 and FIG.29. 拡散符号ベクトルｓｉの定義を示した図である。It is the figure which showed the definition of the spreading code vector si. 計算機シミュレーションの結果を示す図であって、遅延スプレッド５０ｎｓｅｃとした場合の平均ＢＥＲを示した図である。It is a figure which shows the result of computer simulation, Comprising: It is the figure which showed the average BER at the time of setting it as delay spread 50nsec. 計算機シミュレーションの結果を示す図であって、遅延スプレッド２５０ｎｓｅｃとした場合の平均ＢＥＲを示した図である。It is a figure which shows the result of computer simulation, Comprising: It is the figure which showed the average BER at the time of setting delay spread to 250 nsec. 計算機シミュレーションの結果を示す図であって、Ｎ＝８のＷａｌｓｈ符号、ＧＩ＝∞とする場合に、遅延スプレッドを変化させて平均ＢＥＲ特性を調べた結果を示した図である。It is a figure which shows the result of computer simulation, Comprising: When N = 8 Walsh code | cord and GI = infinity, it is the figure which showed the result of having investigated delay average and changing the average BER characteristic. 計算機シミュレーションの結果を示す図であって、ＧＩ＝∞の場合と、伝送効率８／９となるようにＧＩを設定した場合の結果を示した図である。It is a figure which shows the result of a computer simulation, Comprising: It is the figure which showed the result at the time of setting GI so that it may become transmission efficiency 8/9 when GI = infinity. ＩＢＩが、各変調ブロックに関し、ひとつ前の変調ブロックからの前置干渉（ＰｒｅＩＦ）とひとつ後の変調ブロックからの後置干渉（ＰｏｓｔＩＦ）のみを受ける状態について説明する図である。It is a figure explaining the state where IBI receives only the pre-interference (Pre IF) from the previous modulation block and the post-interference (Post IF) from the next modulation block for each modulation block. 復調処理における干渉キャンセラの原理を説明するための図である。It is a figure for demonstrating the principle of the interference canceller in a demodulation process. 拡散符号に対して平均ＢＥＲ特性がどのように変化するかを調べた結果を示す図である。It is a figure which shows the result of having investigated how the average BER characteristic changes with respect to a spreading code. Ｗａｌｓｈ符号を適用した場合に拡散符号長を変化させて干渉キャンセラの特性を調べた結果を示す図であって、Ｎ＝８、Ｗａｌｓｈｃｏｄｅの場合の干渉キャンセラの平均ＢＥＲ特性を示した図である。It is a figure which shows the result of having investigated the characteristic of the interference canceller by changing the spreading | diffusion code length when applying a Walsh code | symbol, Comprising: It is the figure which showed the average BER characteristic of the interference canceller in the case of N = 8 and Walsh code. . Ｗａｌｓｈ符号を適用した場合に拡散符号長を変化させて干渉キャンセラの特性を調べた結果を示す図であって、Ｎ＝１６、Ｗａｌｓｈｃｏｄｅの場合の干渉キャンセラの平均ＢＥＲ特性を示した図である。It is a figure which shows the result of having investigated the characteristic of the interference canceller by changing the spreading | diffusion code length when applying a Walsh code | symbol, Comprising: It is the figure which showed the average BER characteristic of the interference canceller in case of N = 16 and Walsh code. . Ｗａｌｓｈ符号を適用した場合に拡散符号長を変化させて干渉キャンセラの特性を調べた結果を示す図であって、Ｎ＝３２、Ｗａｌｓｈｃｏｄｅの場合の干渉キャンセラの平均ＢＥＲ特性を示した図である。It is a figure which shows the result of having investigated the characteristic of the interference canceller by changing the spreading code length when applying the Walsh code, and showing the average BER characteristic of the interference canceller in the case of N = 32 and Walsh code. . 遅延スプレッドを２５０ｎｓｅｃとした場合の干渉キャンセラの特性を調べた結果を示す図である。It is a figure which shows the result of having investigated the characteristic of the interference canceller when a delay spread is 250 nsec. ＯＦＤＭとの比較のため、拡散符号Ｎを変化させた場合のＷａｌｓｈ符号を適用した時間軸相似等化とＮ＝３２、ＧＩ＝４のＯＦＤＭとを比較した結果を示した図である。It is the figure which showed the result of having compared the time-axis similarity equalization which applied the Walsh code at the time of changing spreading code N, and OFDM of N = 32 and GI = 4 for comparison with OFDM. 図１１に示した原理の動作を行うキャンセラの内部構成を示した図である。It is the figure which showed the internal structure of the canceller which performs the operation | movement of the principle shown in FIG. 図１８のキャンセラの動作を説明するためのフロー図である。It is a flowchart for demonstrating operation | movement of the canceller of FIG. 本発明の実施の形態に係るＭＩＭＯ形式の符号分割多重伝送システムを示す図である。It is a figure which shows the code division multiplexing transmission system of the MIMO format which concerns on embodiment of this invention. 図２０の伝送路４１の伝送路行列Ｈを説明するための図である。It is a figure for demonstrating the transmission line matrix H of the transmission line 41 of FIG. 伝送路行列Ｈに対応するＭＩＭＯの全体の伝送路行列Ｈ_Ｍを示した図である。Is a diagram showing a channel matrix H _M of the overall MIMO corresponding to the channel matrix H. 図２０の変調処理部３１の内部構成を示した図である。It is the figure which showed the internal structure of the modulation process part 31 of FIG. 図２０の復調処理部４０の内部構成を示した図である。It is the figure which showed the internal structure of the demodulation process part 40 of FIG. 図２０に示す伝送行列Ｈ_Ｍに対する随伴行列Ｈ_Ｍ ^Ｈを示す図である。Is a diagram showing an adjoint matrix _H ^{M H} for the transmission matrix _{H M} shown in FIG. 20. 自乗誤差最小規範に基づく適応フィルタ係数制御を行う発明を説明するためのブロック図である。It is a block diagram for demonstrating the invention which performs adaptive filter coefficient control based on the square error minimum norm. 符号化と復号化の処理を含む伝送の全体を説明する図であって誤り訂正符号の適用も説明するための図である。It is a figure explaining the whole transmission including the process of an encoding and a decoding, Comprising: It is a figure for demonstrating application of an error correction code. 一般的な符号分割多重伝送システムの様子を示した図である。It is the figure which showed the mode of the general code division multiplexing transmission system. 図２８のＬＮＡ＆受信フィルタ１１３等において用いられる通常の周波数軸上での逆特性等化（Frequency Domain Inverse Filter, ＦＤＩＦ）を説明するための図である。It is a figure for demonstrating the reverse characteristic equalization (Frequency Domain Inverse Filter, FDIF) on the normal frequency axis used in LNA & reception filter 113 grade | etc., Of FIG.

Explanation of symbols

１、４０５フィルタ
１３、３０伝送システム
５０制御部 1,405 Filter 13, 30 Transmission system 50 Control unit

Claims

A transmission apparatus that generates a block modulation signal X that is a time-series signal having a length N, a distortion block modulation signal Y that is a time-series signal having a length M after the block modulation signal X has passed through a distortion transmission path, etc. In a transmission system including a receiving device having a receiving filter to be converted,
The reception filter includes N element filters EF_i (i = 0 to N−1) arranged in parallel and performing parallel-serial conversion.
Each of the element filters EF_i is
The characteristic is set according to the i-th symbol on the time axis in the block,
The entered the distortion block modulation signal Y, by performing a filtering process by setting properties, the output symbols r_i of each element filters EF_i, the strain block modulation signal Y length N after equalization of A transmission system characterized in that time-series signal symbols R = (r_0, r_1,..., R_N−1).

Each element filter EF_i has a characteristic similar to that of the block modulation signal X over the length N after equalization of the distorted block modulation signal Y according to the i-th symbol on the time axis in the block. those that have been set Te, the transmission system according to claim 1, wherein.

Each element filter EF_i is given by an M-th order FIR filter, each tap coefficient of the FIR filter is set to f_i, j (j = 0 to M−1), and f_i, j is an element of i rows and j columns. When H _LI (N rows and M columns matrix) is defined and the impulse response matrix of the distorted transmission path is defined as a matrix H (M rows and N columns matrix), the matrix H _LI is represented by the following first, second, and The transmission system according to claim 1, further comprising the reception filter set by a matrix H _{LI of a} fourth equation derived from the third equation.
H _LI H = I (1)
(Where I is the unit matrix)
H = UΛ ^1/2 V ^H (2)
(Where U and V are unitary matrices having normalized eigenvectors of HH ^H and H ^H H as column vectors, respectively, and Λ is a diagonal matrix having the eigenvalues of H ^H H as diagonal components)
UΛ ^1/2 = HV (3)
H _LI = VΛ ⁻¹ V ^H H ^H (4)

Each element filter EF_i is given by an M-th order FIR filter, each tap coefficient of the FIR filter is set to f_i, j (j = 0 to M−1), and f_i, j is an element of i rows and j columns. G (a matrix of N rows and M columns) is determined, and a known signal sequence XK is transmitted as the block modulation signal X on the receiving side, and the mean square of errors between the time series signal symbol R and the signal sequence XK is minimized. The transmission system according to claim 1 or 2, wherein each element of the matrix G is controlled as described above.

Each element filter EF_i is given by an M-th order FIR filter, each tap coefficient of the FIR filter is set to f_i, j (j = 0 to M−1), and f_i, j is an element of i rows and j columns. G (a matrix of N rows and M columns) is determined, and the sequence of the time-series signal symbol R after equalization is decoded. The root mean square error between the obtained decoded signal sequence R_H and the sequence of the time-series signal symbol R is 3. The transmission system according to claim 1, wherein each element of the matrix G is controlled so as to be minimized.

When the transmission device generates and transmits a block modulation signal X that is a time-series signal of length N, and the reception filter of the reception device has a length M after the block modulation signal X passes through the distortion transmission path In a transmission method for equalizing a distortion block modulation signal Y that is a series signal,
The reception filter includes N element filters EF_i (i = 0 to N−1) arranged in parallel and performing parallel-serial conversion.
Each of the element filters EF_i is
The characteristic is set according to the i-th symbol on the time axis in the block,
When the inputted distortion block modulation signal Y is subjected to filtering processing according to the set characteristics, the output symbol r_i of each element filter EF_i is equal to the length N after equalization of the distortion block modulation signal Y. A sequence signal symbol R = (r_0, r_1,..., R_N−1).

When the impulse response matrix of said strain transmission path matrix H (a matrix of M rows and N columns), the first equation matrix H _LI is below a fourth equation of the matrix H _LI derived from the two equations and third expression The transmission method according to claim 6, further comprising: the reception filter set by:
H _LI H = I (1)
(Where I is the unit matrix)
H = UΛ ^1/2 V ^H (2)
(Where U and V are unitary matrices having normalized eigenvectors of HH ^H and H ^H H as column vectors, respectively, and Λ is a diagonal matrix having the eigenvalues of H ^H H as diagonal components)
UΛ ^1/2 = HV (3)
H _LI = VΛ ⁻¹ V ^H H ^H (4)

A block modulation signal X having a length NT * N is divided into partial block modulation signals X_i having a length N (time-series signals having a length N, i = 0 to NT−1, respectively), and the partial block modulation signals X_i are respectively obtained. A transmitting apparatus that outputs from a plurality of transmitting antennas, and a distorted partial block modulated signal that is received by NR multiple receiving antennas after the partial block modulated signals X_0 to X_NT-1 pass through a distorted transmission path are represented by Y_j (each long And a distortion partial block modulation signal Y = (Y_0, Y_1,..., Y_NR-1) (long) NR * M), and a reception system having a reception filter that equalizes the distortion block modulation signal Y,
The reception filter includes NT * N element filters EF_k (k = 0 to NT * N−1) arranged in parallel and performing parallel-serial conversion.
Each element filter EF_k is
The characteristic is set according to the kth symbol on the time axis in the block,
The entered the distortion block modulation signal Y, by performing a filtering process by setting properties, the output symbols r_k of each element filters EF_k, the strain block modulation signal Y length N after equalization of A transmission system characterized in that time-series signal symbols R = (r_0, r_1,..., R_N−1).

Each element filter EF_k is given by an NR * M-th order FIR filter, each tap coefficient of the FIR filter is set to f_i, j (j = 0 to NR * M−1), and f_i, j is set to i rows and j columns. A matrix H _{M, LI} (matrix of (NT * N) rows (NR * M) columns) is defined as an element, and the impulse response matrix of the distortion transmission path is defined as a matrix H _M ((NR * M) rows (NT * N ) Column matrix), the matrix H _{M, LI} has the reception filter set by the matrix H _{M, LI} of the fourth formula derived from the following first, second and third formulas: The transmission system according to claim 8 .
H _{M, LI} H _M = I (1)
(Where I is the unit matrix)
H _M = UΛ ^1/2 V ^H (2)
(Where U and V are unitary matrices whose column vectors are normalized eigenvectors of H _M H _M ^H and H _M ^H H _M , respectively, and Λ is a diagonal having the eigenvalues of H _M ^H H _M as diagonal components. line; queue; procession; parade)
UΛ ^1/2 = H _M V (3)
H _{M, LI} = VΛ ⁻¹ V ^H H _M ^H (4)

Each element filter EF_k is given by an NR * M-th order FIR filter, each tap coefficient of the FIR filter is set to f_i, j (j = 0 to NR * M−1), and f_i, j is set to i rows and j columns. A matrix G (matrix of (NT * N) rows (NR * M) columns) is defined as an element, a known signal sequence XK is transmitted as the block modulation signal X on the receiving side, and the time-series signal symbol R and the 9. The transmission system according to claim 8 , wherein each element of the matrix G is controlled so that a mean square error of the signal sequence XK is minimized.

Each element filter EF_k is given by an NR * M-th order FIR filter, each tap coefficient of the FIR filter is set to f_i, j (j = 0 to NR * M−1), and f_i, j is set to i rows and j columns. A matrix G (matrix of (NT * N) rows (NR * M) columns) is defined as an element, the sequence of the time series signal symbols R after equalization is decoded, and the obtained decoded signal series R_H and the time series are obtained. 9. The transmission system according to claim 8 , wherein each element of the matrix G is controlled so that a mean square of an error with respect to the sequence of signal symbols R is minimized.

Said block encoded signal X is, in any one of claims 1 to 4 and 8, characterized in that provided by multiplexing by summing over time multiplied by a different information signal to each of a plurality of time block codes axis 1 1 of The described transmission system.

Transmission system according to claim 1 2, wherein the applying the error correction code information signal, characterized in that with the error correction code new information signal a signal sequence after applying.

Receiving filter used claims 1 to 4 and 8 to 1 3 or the transmission system.

A transmitting apparatus generates and transmits a block modulation signal X that is a time-series signal having a length N, and the block modulation signal X is a time-series signal having a length M after passing through a distortion transmission path. Y is a decoding method in the transmission method in which the receiving apparatus has a reception filter for equalizing the distorted block modulated signal Y,
The reception filter includes N element filters EF_i (i = 0 to N−1) arranged in parallel and performing parallel-serial conversion.
Each of the element filters EF_i is
The characteristic is set according to the i-th symbol on the time axis in the block,
The entered the distortion block modulation signal Y, by performing a filtering process by setting properties, the output symbols r_i of each element filters EF_i, the strain block modulation signal Y length N after equalization of Time-series signal symbol R = (r_0, r_1,..., R_N−1),
The receiving apparatus has a reception modulation block buffer that records the received distortion block modulation signal Y, and a decoding block buffer that records a decoding result for each received distortion block modulation signal,
When a new distorted block modulated signal A is received, the distorted block modulated signal A is stored in the received modulated block buffer and received immediately before the distorted block modulated signal A recorded in the decoded block buffer. By remodulating the decoding result of the distorted block modulation signal B and applying a filter corresponding to the transmission path, a pre-interference component received by the distorted block modulation signal A from the distorted block modulation signal B is obtained, and the pre-interference component is obtained. A decoding result obtained by removing and decoding from the distorted block modulated signal A is stored in the decoded block buffer, and the second new distorted block modulated signal among the distorted block modulated signals in the received modulated block buffer is re-decoded distorted block A first step of setting the modulation signal;
The latest decoding result of the distorted block modulated signal received immediately before the distorted block modulated signal to be re-decoded is read out from the decoded block buffer, re-modulated, and further subjected to a filter corresponding to the transmission path to reduce the pre-interference component. The latest decoding result of the distorted block modulation signal received immediately after the distorted block modulation signal to be re-decoded is read out from the decoding block buffer, re-modulated, and further subjected to a filter corresponding to the transmission path to perform post-interference A decoding result obtained by obtaining a component, removing the pre-interference component and the post-interference line segment from the re-decoding target distortion block modulation signal, and storing the decoded result in the decoding block buffer and stored in the reception modulation block buffer The distorted block modulated signal received immediately before the re-decoded distorted block modulated signal A second step of setting No.,
A third step in which the second step is repeated a desired number of times;
Including a decoding method.

The transmitter divides the block modulation signal X having a length NT * N into a partial block modulation signal X_i having a length N (time-series signals having a length N, i = 0 to NT-1 respectively), and the partial block modulation is performed. Each of the signal X_i is output from a plurality of transmission antennas, and after the partial block modulation signals X_0 to X_NT-1 pass through the distortion transmission path, the reception apparatus receives the distortion partial block modulation received by the plurality of NR reception antennas. The signal is Y_j (each time-series signal of length M, j = 0 to NR-1), and the distorted block modulation signal Y = (Y_0, Y_1,... Y_NR-1) (length NR * M), and a decoding method in the transmission method for receiving the received distortion block modulated signal Y with a reception filter,
The reception filter includes NT * N element filters EF_k (k = 0 to NT * N−1) that are arranged in parallel and perform parallel-serial conversion, and each element filter EF_k is i on the time axis in the block. th is set characteristics in accordance with the symbol for the distortion block modulation signal Y inputted by performing a filtering process by setting properties, the output symbol r_k of each element filter EF_k, the strain block modulation A time-series signal symbol of length N after equalization of the signal Y is R = (r_0, r_1,..., R_N−1),
A reception modulation block buffer for each antenna that records the respective distortion partial block modulation signals received in parallel by the plurality of reception antennas by the reception apparatus; and a decoding block buffer that records a decoding result for the distortion block modulation signals. Have
When a new distortion partial block modulation signal A_i (i is an antenna number) is received, the latest decoding result of the distortion block modulation signal corresponding to the distortion partial block modulation signal B_i received immediately before is received from the decoding block buffer. Read, remodulate, and apply a filter corresponding to the transmission path to obtain a pre-interference component for each A_i, remove each pre-interference component from the distorted partial block modulation signal A_i, and remove the pre-interference component A decoded result obtained by decoding the distorted block modulated signal obtained by connecting each A_i to the subordinates is stored in the decoded block buffer, and the distorted partial block modulated signal in the received modulated block buffer for each receiving antenna is stored. The first new distortion partial block modulation signal is set as the re-decoding target distortion partial block modulation signal. And the step,
The latest decoding result of the distorted block modulation signal corresponding to the distorted partial block modulation signal received immediately before the re-decoding-target distorted partial block modulation signal for each receiving antenna is read from the decoded block buffer, remodulated, and further transmitted. To obtain a pre-interference component for each re-decoding target distorted partial block modulated signal for each receiving antenna, and for each receiving antenna received immediately after the re-decoding target distorted partial block modulated signal. The latest decoding result of the distorted block modulated signal corresponding to the distorted partial block modulated signal is read out from the decoded block buffer, re-modulated, and further subjected to a filter corresponding to the transmission path to re-decode the distorted partial block modulated for each antenna. The post-interference component for the signal is obtained, and the pre-interference component such as Decoding result obtained by decoding each of the distorted block modulation signals obtained by subordinately connecting the re-decoding target distorted partial block modulation signals from which the post-interference components and the post-interference components have been removed. Is stored in the decoding block buffer, and each distortion partial block modulation signal received immediately before the re-decoding-target distortion partial block modulation signal in the reception modulation block buffer for each reception antenna is replaced with a new re-decoding-target distortion part. A second step of setting the block modulation signal;
A third step in which the second step is repeated a desired number of times;
Including a decoding method.