JP3880370B2 - Multi-carrier CDMA receiver - Google Patents

Description

ここで、［・］^Tは転置行列を表す。
拡散符号は、ユーザごとに異なる符号を用いる。拡散符号のチップ数（拡散符号ベクトルの要素数）は、サブキャリヤのチャネル数に等しい。サブキャリヤの周波数は、OFDMと同様に、直交配置されたものである。
図１に示した逆離散フーリエ変換部５は、各サブキャリヤの複素振幅を入力して、逆離散フーリエ変換して、時間軸上の波形信号を出力する。並列デジタル信号で出力されるので、並直列変換器６で直列デジタル信号に変換して出力する。
ガードインターバル挿入部７において、ガードインターバルΔを挿入して、マルチキャリヤＣＤＭＡ方式の送信信号が生成される。
【００２７】
なお、図１に示した構成では、ＱＰＳＫ変調信号を、同時に１シンボルしか送信していない。しかし、１シンボル長を長くして準同期を容易にするためには、ＱＰＳＫ変調器２を複数個設けて、多重化データを分配する。加えて、数倍長いチップ数の拡散符号と数倍のサブキャリヤ数を使用することにより、１シンボル長において複数のＱＰＳＫ変調器から出力される複数のシンボルを同時に送信する。
このような場合も含む概念で表現するため、本明細書では、１シンボル区間における送信信号波形の全体を改めて１シンボル波形とみなすことにより、１シンボル区間において１シンボルが送信されるとする。しかし、この場合でも、既知のパイロットシンボルを用いてパイロットシンボル区間を構成する。
【００２８】
図３は、マルチキャリヤＣＤＭＡ方式の送信信号波形を示す波形図である。
サブキャリヤは、OFDMと同様に直交に配置されているため、送信波形はOFDMのそれと同様になる。
ガードインターバルは、遅延波による符号間干渉を除去するために挿入される。図示のガードインターバルは、逆離散フーリエ変換された原送信信号波形の終わり部分をコピーして、原送信信号波形の前の部分に挿入したものである。
受信装置において、図示の送信信号波形を直接波とし、遅延波がガードインターバルの長さΔ以内の遅延時間で到来するものとすれば、このような遅延波によるシンボル間干渉の影響を抑えることが可能となる。
すなわち、受信側では、１シンボル区間よりもガードインターバルの長さΔだけ短い区間（原送信信号波形と同じ長さである）を切り出して利用すれば、切り出された波形に含まれる遅延波は、この切り出された波形に含まれる直接波を単に遅延させただけの波形であるから、遅延波による干渉は生じない。
【００２９】
なお、受信装置側で、一番目に到来した波形を利用するとは限らない場合に、この波形に先行する到来波による符号間干渉が生じる。この場合は、周知のように、原送信信号波形の始めの部分（長さΔ）をコピーしたものを、そのまま原送信信号波形の後ろの部分に挿入することにより、後ろの部分にもガードインターバルを挿入する。受信側では、１シンボル区間よりもガードインターバルの合計長さ（２Δ）だけ短い区間を切り出して利用する。
以上の構成によって、周波数領域で送信データが拡散された送信信号波形が生成される。この送信信号波形は、メインキャリアを用いて無線周波数帯に変換されて送信される。
【００３０】
送信信号ｘ（q）は、時間領域で次式で表される。
【数２】

ここで、Ｆは離散フーリエ変換、Ｆ^-1は逆離散フーリエ変換を表す。
時間軸上のL個のサンプル点x₁ ，x₂，x₃ ，……，x_Lは、１シンボル期間のうち、ガードインターバルを除いた区間（すなわち、ガードインターバルを挿入する前の原送信信号波形）を等分したサンプル点1〜Lにおける送信信号の複素振幅である。
【００３１】
離散フーリエ変換Ｆは、次式の行列で表される。
【数３】

L個のサンプル点に対応して、サブキャリヤの周波数は、サンプル時間をT_sとするとベースバンドでは、-(L/2-1)/(LT_s),-(L/2-2)/(LT_s) ,-(L/2-3)/(LT_s) ,……,0,1/(LT_s) ,2/(LT_s) ,3/(LT_s) ,……,(L/2)/(LT_s)となる。
【００３２】
図４は、本発明のマルチキャリヤＣＤＭＡ受信装置のブロック構成図である。図中、１１は多重分離部（デマルチプレクサ）、１２はガードインターバル除去部、１３はチャネル推定部、１４は直並列変換部、１５は離散フーリエ変換（ＤＦＴ）部、１６₁〜１６_Lは乗算器、１７は加算器、１８はＱＰＳＫ復調器、１９は離散フーリエ変換（ＤＦＴ）部、２０は重みベクトル計算部である。
受信信号は、多重分離部１１において、同期タイミングに基づいて、データ信号区間とパイロット信号区間とに分離する。ガードインターバル除去部１２は、受信信号中のガードインターバルを除去した上で、直並列変換部１４で並列信号に変換して、離散フーリエ変換（ＤＦＴ）部１５に出力される。
【００３３】
希望するある単一のユーザの移動局から基地局までの伝搬路の特性は、次式に示す伝搬遅延プロファイルh(t)で表される。
【数４】

ここで、a_lは伝搬ベクトルａのl（Lの小文字）番目のパスの成分、T_sはサンプル時間、またδ(t)はディラックのデルタ関数である。したがって、l番目のパスの遅延時間は、（l-1）T_sとなる。
【００３４】
受信信号y(q)（時間領域）は、送信信号ｘ（q）が伝搬路で変動を受けたものであるから、式(2)，(4)を用いて次式で表される。ただし、ガードインターバルが除去された後の受信信号である。
【数５】

ここで、ｘ（q；l）は、送信信号ｘ（q）が（l-１）T_sだけ遅延した信号、ｎは干渉雑音成分を表す。
【００３５】
上述した式における、遅延した信号の特徴を補足説明する。
図２に示したように、ガードインターバルにおいて、送信信号波形は、原送信信号波形の後ろの部分をコピーした信号である。したがって、ガードインターバルの最後から逆順にサンプル点の複素振幅を示すと、x_L ，x_L-1 ，x_L-2 ，x_L-3 ，……となる。
全ての遅延した信号ｘ（q；1），ｘ（q；2），……が、ガードインターバルの長さΔ内に収まると仮定すると、遅延した信号は、ガードインターバルにおけるサンプル点の複素振幅の特徴から、離散フーリエ変換を行う区間において、直接波ｘ（q；1）を巡回シフトしたものとなる。
ただし、基地局に実際に到来する遅延した信号としては、l＝１〜LまでのL波が常に全て存在するわけではない。例えば、２番目に到来する遅延した信号がないときは、a₂＝0となる。
【００３６】
離散フーリエ変換（ＤＦＴ）部１５は、離散フーリエ変換を行い、直交する複数のサブキャリヤの複素振幅を、離散フーリエ変換された受信信号として出力する。
離散フーリエ変換された受信信号、すなわち、周波数領域の受信信号Ｙ（q）は、次式で与えられる。
【数６】

ここで、Ｊ^l-1は対角行列の各要素を（l-1）乗した対角行列である。丸印を付した記号「×」は２つのベクトルの要素ごとに乗算をしたものを各要素とするベクトルを生成する演算である。
なお、上述した式(6)内の変形において、次式の関係が成り立つことを用いた。
【数７】

【００３７】
各サブキャリヤに対応した乗算器１６₁〜１６_Lにおいて、重みベクトル計算部２１から出力される重みが乗算される。この重みは、希望する単一ユーザに割り当てられて送信時に使用された拡散符号のチップ１〜Ｌの値に重み付けが施されたものである。
全ての乗算器１６₁〜１６_Lの出力は、加算器１７において加算合成され、複素ベースバンド信号で表されるＱＰＳＫ変調されたシンボルとなり、ＱＰＳＫ復調器１８において、同相成分、直交成分をレベル判定され、復調されて、受信データが出力される。
【００３８】
MC-CDMA受信信号のベクトルＹに対して、適切な重みベクトル
【数８】

に基づき、サブキャリヤ間信号合成を行う。
加算器１７から出力される最終的な合成出力は次式で表される。
【数９】

ここで、［・］^†は行列の各要素の複素共役を要素とした行列を転置（共役転置）した行列（エルミート行列）を示す。
【００３９】
この重みベクトルｗは、信号電力が大きく干渉雑音電力が小さくなるように決定されることが望ましい。このような要求を満たす方法として、従来技術で触れたMMSE合成基準を用いる。重みベクトルｗは、次式によって求める。
【数１０】

ここで、Ｒは相関行列、ｖは相関ベクトルと呼ばれるものである。また、Ｒ^-1はＲの逆行列、Ｅ［・］はアンサンブル平均、（・）^*は複素共役を表している。
相関行列Ｒは、受信信号を離散フーリエ変換した受信信号ベクトルＹ（q）の要素（サブキャリヤの複素振幅）の相互間における相関値のアンサンブル平均である。希望ユーザの移動局からの到来波だけでなく、全ユーザの移動局からの到来波を含めた相関のアンサンブル平均となる。
一方、相関ベクトルｖは、受信信号を離散フーリエ変換した受信信号ベクトルＹ（q）と希望するユーザの移動局の送信シンボルデータd(q)との相関のアンサンブル平均である。
【００４０】
上述した(9)式における、Ｒの第１番目の関係式にしたがって、離散フーリエ変換された受信信号Ｙ（q）に関する第１の相関式のアンサンブル平均をとり、かつ、ｖの第１番目の関係式にしたがって、離散フーリエ変換した受信信号Ｙ（q）に関する第２の相関式のアンサンブル平均をとることにより、重みベクトルｗを求めることができる。
しかし、本発明のマルチキャリヤＣＤＭＡ受信装置では、パイロットシンボルのインパルス応答に基づいて伝搬ベクトルａを求めてから、上述した(9)式における、Ｒ，ｖの第２番目の関係式にしたがって、相関行列Ｒと相関ベクトルｖを算出する方法をとる。
図４に示したチャネル推定部１３、離散フーリエ変換部１９、重みベクトル計算部２０で実現されるのであるが、その処理の詳細は、図５を参照して説明する。
【００４１】
図５は、パイロット信号から重みベクトルを算出するための処理を示すフローチャートである。
図６は、インパルス応答の推定方法を説明する模式図である。
Ｓ３１において、パイロット信号を用いてインパルス応答の推定値から伝搬ベクトルａを推定する。
なお、全てのユーザの移動局からの伝搬路の伝搬ベクトルを推定する。しかし、以下の数式を用いた説明では、説明を簡単にするため、途中まで、希望ユーザの移動局からの伝搬路の伝搬ベクトルで代表させ、これをａとして説明する。最終的には、ユーザk=1〜Kについて、同様な方法で伝搬ベクトルａ_kを計算する。
【００４２】
パイロット信号のサンプルqにおいて、既知の送信信号Ｘ（q）と受信信号y(q)の相関φ_s（q）は、サンプル差をｓとして、
【数１１】

なお、式(10)内の変換において次式の関係を用いた。
【数１２】

【００４３】
式(10)において、ｘ（q；s）とｎとは無相関であるので、ｘ（q;s）^†ｎは、シンボルごとに異なる値をとる。したがって、φ_s（q）のシンボル値qについてのアンサンブル平均をとることにより、ｘ^†（q;s）ｎが消えて、伝搬係数a_sが求まる。
図６に示すように、sの値を1サンプルずつシフトして、インパルス応答の遅延した信号ｘ（q;s）について、相関演算を行うことにより、伝搬ベクトルａ＝｛a₁，a₂ ，……，a_L ｝を推定できる。
全ての遅延した信号がガードインターバル内に収まると仮定すると、ガードインターバルの区間においてだけ、インパルス応答を計算すればよい。
なお、相関演算は、マッチドフィルタを用いて行うことができる。
【００４４】
次のＳ３２においては、インパルス応答の補正を行うことにより特性改善をする。
図７は、インパルス応答の補正方法を説明する模式図である。
図７（ａ）において、太い矢印がインパルス応答の電力値を表している。
まず、主要な遅延した信号（遅延パスとして表される）が、全てガードインターバル内に入るようにシステム設計を行うので、ガードインターバル外の遅延パスは、全て干渉雑音（干渉または雑音）とみなしてカットする。
次に、電力が最大のパス（図示の例では、直接波に相当する直接パス）よりもT_h[dB]以上低いパス（図示の例では２番目のパス）も干渉雑音とみなしてカットする。図７では、カットされたパスには、Ｘ印を付している。
以上の操作により、図７（ａ）で示したインパルス応答は、図７（ｂ）で示したインパルス応答に補正される。
【００４５】
再び図５に戻って説明を続ける。
これより以下の説明では、ユーザkごとの伝搬ベクトルａ_kを用いて説明する。
Ｓ３３において、希望ユーザ（k=1とする）を含む全てのユーザの伝搬路について、それぞれ伝搬路の推定結果であるインパルス応答ａ_kをフーリエ変換してＦａ_kを得る。
Ｓ３４において、式(9)の相関ベクトルｖの第２行目の式にしたがって、Ｆａ₁と拡散符号ベクトルＣ₁（ユーザk=1に割り当てられた拡散符号）とから相関ベクトルｖを算出する。
一方、Ｓ３５においては、全ユーザ（k=1〜K）について、上述したＳ３２におけるインパルス応答の補正処理でカットしたパスの電力から雑音電力Ｐ_Nを推定する。
【００４６】
雑音電力Ｐ_Nの推定について詳述する。
希望ユーザ以外の他ユーザの移動局からの受信信号は、希望ユーザの移動局からの受信信号とは、必ずしもシンボル同期がとれていない。そのため、希望ユーザの移動局からの受信信号の、１シンボル内において、他ユーザの移動局からの受信信号は、隣接する２つのシンボルを部分的に含んでいる状態となり得る。
しかし、ここでは計算を簡単にするため、他ユーザも希望ユーザと同じタイミングでシンボルを伝送するものとする。
基地局が下り回線で供給する下りパイロット信号に同期して、移動局が送信信号を送信することとし、かつ、移動局が基地局から所定の距離内にあるようにシステム設計しており、他ユーザの移動局からの受信信号を、ガードインターバル内に収めることができるので、離散フーリエ変換する範囲内では、他ユーザも希望ユーザと同じタイミングでシンボルを伝送しているに等しい。
【００４７】
上述した式(9)のＲの２行目の式において、Ｆｎｎ^†Ｆ^†の項には、他のユーザとの干渉成分と雑音成分とが混じっているので、評価が難しい。
そこで、式(9)を参考にして、他ユーザの伝搬ベクトルも考慮した相関行列を、次式の通りとする。
【数１３】

ここで、ａ_kはユーザkの伝搬ベクトル、Ｃ_kはユーザkの拡散符号ベクトル、Ｐ_Nは雑音電力、Ｉは単位行列を表す。
Ｐ_Nは、図５のＳ３２において、各ユーザkについて伝搬ベクトルａ_kを求める際に、カットしたパスの総電力を求めて、この総電力を全てのユーザkについて平均した値を雑音電力Ｐ_Nとする。
上述した式によって、マルチユーザ干渉と雑音とを、完全ではないが、ほぼ分離して評価した上で、相関行列Ｒが得られる。
Ｓ３６においては、相関行列Ｒの計算を行う。
Ｓ３７において、重みベクトルを計算する。全てのユーザkのフーリエ変換された伝搬路の推定結果（伝搬ベクトルａ_k）と、全てのユーザkに割り当てられた拡散符号ベクトルＣ_kの情報（レプリカ）と、雑音電力Ｐ_Nの推定値とに基づいて、式(9)に示した希望ユーザ（k=1）の重みベクトルｗを計算する。
【００４８】
最後に、図８〜図１０を参照して、本発明のマルチキャリヤCDMA受信装置を計算機シミュレーションした結果を説明する。
図８は、単局ユーザのみで他局ユーザが存在しない（マルチユーザ干渉がない）場合において、インパルス応答推定値の補正を行わないときのBER（ビット誤り率）特性を示す線図である。
E_b/N_o（１ビットあたりの信号エネルギー／片側雑音電力スペクトル密度）に対するBERを、パイロットシンボル数（q₀）をパラメータとして示している。
拡散符号ベクトルＣには符号長（チップ数）32の直交ゴールド符号を用いている。干渉移動局はなく、付加的白色ガウス雑音（AWGN）伝送路とする。
このとき、ビット誤り率特性は、パイロットシンボル数（q₀）が少ないほど特性が劣化しており、これはインパルス応答推定値の推定誤差が原因と考えられる。
【００４９】
図９は、単局ユーザのみで他局ユーザが存在しない（マルチユーザ干渉がない）場合において、インパルス応答推定値の補正を行ったときのBER（ビット誤り率）特性を示す線図である。
図中、比較のため、インパルス応答推定値の補正を行わないときのBER特性を一部、図８からコピーして黒のプロットで示している。
伝搬路等の条件は図８の場合と同様である。図７に示したインパルス応答推定値の補正を行うことでビット誤り率特性が改善されることがわかる。なお、ここではスレッショルドT_hを６dBとしたが、この値は伝搬路によって最適な値を選択する必要がある。
【００５０】
図１０は、多元接続を行うために他局ユーザが存在する（マルチユーザ干渉がある）場合における、ユーザ数に対する平均SINR（信号対干渉雑音比）特性を示す線図である。
E_b/N_o＝30[dB]、パイロットシンボル数q₀=8、伝搬路は２波モデルとし、フェージング変動は十分遅く、１フレーム（パイロットとデータシンボルのセット）内の伝搬ベクトルａ_kは一定であるとする。また、簡単化のため同期システムで評価した。
比較対象として、単一ユーザ受信で用いられるフェージング補償方式を用いた場合と、収束が速いことで知られるRLSアルゴリズムによるMMSE合成方式を用いた場合についても図示している。
【００５１】
単一ユーザ受信のフェージング補償方式では、他局の伝搬路特性等を考慮していないため、２ユーザ多重でSINRが急激に低下している。また、RLSではパイロットシンボル数が足りないため、重みベクトルが収束せず、多重数が多いときのSINRの低下が大きくなっている。一方、本発明の方式は、多重数の増加に伴ってSINRが低下するものの急激な劣化はみられず、比較した２つの方式よりも良い特性が得られていることがわかる。
【００５２】
上述した説明では、インパルス応答による伝搬ベクトルを用いて重みベクトルを算出している。しかし、このようにして算出された重みベクトルを初期値として、従来の重みベクトル算出用のMMSE型適応アルゴリズムを用いて重みベクトルを逐次更新させてもよい。逐次更新されるにつれて、重みベクトルの推定精度が向上し、また、重みベクトルを伝搬路変動に追従させることができる。
【００５３】
その際、同じ１フレームの最初のパイロットシンボルからこのフレームの最後のデータシンボルまで、先に算出された重みベクトルを初期値として、MMSE型適応アルゴリズムを用いて重みベクトルを逐次更新させる。あるいは、同じ１フレームのデータシンボルの最初からの最後までを、先に算出された重みベクトルを初期値として、MMSE型適応アルゴリズムを用いて重みベクトルを逐次更新させる。
いずれの場合も、データシンボル区間では、この重みベクトルを用いて次々に合成されたデータシンボルを、送信されたデータシンボルの仮判定データシンボルとして、MMSE型適応アルゴリズムを適用する。
あるいは、先に算出された重みベクトルを初期値として、同じ１フレームのパイロットシンボル区間においてMMSE型適応アルゴリズムを用いて重みベクトルを逐次更新させてもよい。更新された所定値の重みベクトルを用いて、同じ１フレームのデータシンボル区間において、送信されたデータシンボルを合成してもよい。
【００５４】
この他、１フレーム内のパイロットシンボル区間において生成された重みベクトルを用いて合成されたデータシンボルを、仮判定されたデータシンボルとして、パイロットシンボルとともに新たにパイロットシンボルと見なすことができる。したがって、再びインパルス応答推定に基づいて重みベクトルを更新することにより、送信されたデータシンボルを合成できる。
また、新たなパイロットシンボル区間において、従来のMMSE型適応アルゴリズムにより重みベクトルを逐次更新させ、更新された重みベクトルを用いて送信されたデータシンボルを合成できる。
以下、仮判定したデータシンボルを用いて重みベクトルの精度を向上させる実施の形態について説明する。
【００５５】
図１１は、本発明のマルチキャリヤＣＤＭＡ受信装置の、他の実施の形態を示すブロック構成図である。図中、図４と同様な部分には同じ符号を付して説明を省略する。
４１はチャネル推定部、４２はデータシンボルレプリカ生成部である。
データシンボルレプリカ生成部４２は、ＱＰＳＫ復調された受信データを送信されたデータであると仮定して、まず、送信されたデータシンボルを仮判定する。この仮判定データシンボルの情報に基づいて、この仮判定データシンボルの送信信号（波形）、すなわち、データシンボルレプリカを生成する。
【００５６】
チャネル推定部４１は、データシンボルレプリカ生成部４２が出力するデータシンボルレプリカを、希望ユーザのデータシンボルレプリカとして入力する。同時に、他の全てのユーザの各受信装置からも、上述した希望ユーザの受信装置と同様に、受信データを送信されたデータと仮判定して、各ユーザの仮判定データシンボルの情報に基づいて生成されたデータシンボルレプリカを入力する。
希望ユーザおよび他ユーザからなる全てのユーザの各データシンボルレプリカは、図４において使用していた全ユーザの各パイロットシンボルレプリカと同様な既知のものとなる。
【００５７】
したがって、各ユーザの、パイロットシンボルレプリカとデータシンボルレプリカとを合わせた１フレーム中のシンボルのレプリカは、図４において使用した各ユーザのパイロットシンボルのレプリカと同様に用いることができる。
その結果、チャネル推定部４１は、データシンボル区間の受信信号も入力し、パイロットシンボル区間に加えてデータシンボル区間において、チャネル推定を行うことができる。
ここで、データシンボル区間の全てのデータシンボルについて仮判定することは必須でない。データシンボル区間の始めから少なくとも１シンボルまでのデータシンボルについて仮判定すれば、既知のシンボルを増加させて、より長い期間にわたって重みベクトルの計算を行うことができるので、重みベクトルの精度が向上する。
ただし、希望ユーザだけでなく、他ユーザの仮判定データシンボルも必要となる場合がある。しかし、基地局における上り回線受信のように、「マルチユーザ受信」においては、他ユーザの仮判定データを利用することが簡単にできるので、処理量は増加するものの特に問題はない。
【００５８】
なお、データシンボルレプリカ生成部４２について説明を補足しておく。
受信データを仮判定データとして、これに基づいて、データシンボルレプリカを受信装置側で生成する。この機能を実現するブロックをデータシンボルレプリカ生成部４２としている。
より具体的には、図１の送信装置構成と同様に、仮判定データをＱＰＳＫ変調して仮判定データシンボルを生成し、分岐、拡散符号の乗算、逆拡散フーリエ変換、並直列変換、ガードインターバル挿入を行う。
したがって、あたかも、図１１に示した受信装置の内部に送信装置を併せ持つことになり、装置規模が複雑になる。しかし、ＱＰＳＫシンボルでは送信信号（データシンボルレプリカ）が４パターンしかないので、メモリ空間に４パターンの送信信号を保存しておき、仮判定したデータシンボルの情報に対応するパターンに応じて、メモリ空間から送信信号を読み出せば、上述した逆拡散フーリエ変換等の複雑な処理が不要になる。
【００５９】
以下の説明では、ある１フレームについて、そのデータシンボル区間の始めから所定のデータシンボル（少なくとも１シンボル）までを仮判定する。これをパイロットシンボルとともに用いて、同じ１フレームのパイロットシンボル区間およびデータシンボル区間の始めから所定のデータシンボルまで、に対して重みベクトルを更新する計算を行い、この更新された重みベクトルを用いて、同じフレームの全データシンボル区間において、送信されたデータシンボルを再び合成する。
ただし、処理遅延が生じるため、１フレーム長のパケットデータである場合や、ファイル転送等に適する。
ここで、上述した「所定のデータシンボル」とは、１シンボルでもよいし、データシンボル区間の全データシンボルでもよい。
なお、図１１では、図４と対比するため、パイロットシンボルのレプリカとデータシンボルのレプリカを別々に生成するものとして図示している。しかし、１フレーム中のパイロットシンボルとデータシンボル区間の始めから所定のデータシンボルまでを合わせたものに対して、レプリカを一括生成してもよい。
【００６０】
図１２は、仮判定データシンボルを用いて行う重みベクトルの第１の更新方法を説明する模式図である。図１１に示した受信機構成に対応する。
ステップ（Ｓ１）では、１フレーム中のパイロットシンボル区間において、図６と同様の処理を行う。ただし、希望ユーザだけでなく、全ユーザk（k =1〜K）の受信信号についてそれぞれ行う。すなわち、全ユーザkのパイロットシンボルのレプリカを用いて、全ユーザkの重みベクトル（以後、これをｗ_kと表記する）をそれぞれ計算する。
【００６１】
ステップ（Ｓ２）では、全ユーザk（k=1〜K）の受信信号と全ユーザkの重みベクトルｗ_kを用いて、全ユーザkのデータシンボルをそれぞれ仮判定する。すなわち、パイロットシンボルを用いて求めた全ユーザkの重みベクトルｗ_kによって全ユーザのデータシンボルを仮判定する。
処理区間は、データシンボル区間の始めから所定のデータシンボルまでである。なお、既に説明したように、データシンボル区間の全てのデータシンボルでもよい。
１フレーム中のデータシンボル区間の始めから所定のデータシンボルまでにおいて、データシンボルをパイロットシンボルと同じと見なすことができる。全ユーザkに関して、既知のパイロットシンボルと仮判定されたデータシンボルを基にして、１フレーム中の送信信号（パイロットシンボル区間およびデータシンボル区間の始めから所定のデータシンボルまで）について、レプリカ（時間軸信号）を再構成する。
【００６２】
ステップ（Ｓ３）では、受信信号（パイロットシンボル区間およびデータシンボル区間の始めから所定のデータシンボルまで）と、既に得られている全ユーザkの［パイロットシンボル＋仮判定データシンボル］のレプリカを新たなパイロットシンボルと見なしたものを用いて、再びインパルス応答値を推定し、希望ユーザの重みベクトルｗを計算し直す。
ステップ（Ｓ４）において、データシンボル区間において、受信信号と希望ユーザの重みベクトルｗとを用いて、希望ユーザのデータシンボルを合成（判定）する。
【００６３】
図１３は、仮判定データシンボルを用いて行う重みベクトルの第２の更新方法を説明する模式図である。
基本的には、図１１に示した受信機構成をとるが、重みベクトルを更新するための構成は異なる。インパルス応答の推定に関しては、図１２に示した上半分の説明図と同様であるので説明を省略する。
先に説明した図１２においては、ステップ（Ｓ３）において、重みベクトルを更新する再計算を、インパルス応答の推定で行っていたのに対し、この第２の更新方法では、従来のMMSE（最小２乗平均誤差）型適応アルゴリズムによって更新させるというものである。この更新計算の相違によって、仮判定データシンボルの生成までのステップ（Ｓ１），（Ｓ２）の処理が変更され、希望ユーザのデータシンボルだけを仮判定すればよい。
また、仮判定データシンボルの情報が得られればよく、そのレプリカは必要としない。ステップ（Ｓ４）については同じである。
フレーム内のシンボル数が十分多い場合に、この重みベクトルの逐次更新によるMMSE型適応アルゴリズムが有効である。
【００６４】
ステップ（Ｓ１１）では、１フレーム中のパイロットシンボル区間において、図６と同様の処理を行う。全ユーザのパイロットシンボルのレプリカを用いて、希望ユーザの重みベクトルｗを計算する。
ステップ（Ｓ１２）では、１フレーム中のデータシンボル区間の始めから所定のシンボル（少なくとも１シンボル、データシンボル区間の全てのデータシンボルでもよい）までにおいて、受信信号と希望ユーザの重みベクトルｗを用いて、希望ユーザのデータシンボルを仮判定する。
希望ユーザに関して、既知のパイロットシンボルと仮判定されたデータシンボルとを基にして、１フレーム中の送信シンボル（パイロットシンボルおよび仮判定データシンボル）情報を再構成する。
【００６５】
ステップ（Ｓ１３）において、１フレーム中の受信信号の、それぞれのパイロット信号およびデータ信号について、希望ユーザの重みベクトルｗをMMSE型適応アルゴリズムにより計算し直す。
具体的には、ステップ（Ｓ１１）のパイロットシンボル区間において得られた重みベクトルｗを初期値とし、受信信号（パイロットシンボル区間およびデータシンボル区間の始めから所定のデータシンボルまで）と、ステップ（Ｓ１２）で得られた希望ユーザの［パイロットシンボル＋仮判定データシンボル］情報をパイロットシンボル情報と見なしたものを用いて、希望ユーザの重みベクトルｗを計算し直すことによって、重みベクトルの精度を向上させる。
【００６６】
MMSE型適応アルゴリズムでは、既に説明したように、(9)式における、Ｒの第１番目の関係式にしたがって、離散フーリエ変換された受信信号Ｙ（q）に関する第１の相関式のアンサンブル平均をとる。
一方、ｖの第１番目の関係式にしたがって、離散フーリエ変換した受信信号Ｙ（q）に関する第２の相関式のアンサンブル平均をとる。ここで、d^*(q)に、上述した希望ユーザの送信シンボル（パイロットシンボル＋仮判定データシンボル）の値が代入される。その結果、最終的には、１フレームの終了時に、(9)式のｗ＝Ｒ^-1ｖにより、重みベクトルｗが更新される。
ステップ（Ｓ４）については、図１２と同様に、データシンボル区間において、希望ユーザのデータシンボルを判定する。
なお、この更新方法において、（Ｓ１２）を省略して仮判定データシンボル数を０とすれば、既に説明したような、先に算出された重みベクトルを初期値として、同じ１フレームのパイロットシンボル区間においてMMSE型適応アルゴリズムを用いて重みベクトルを逐次更新させ、更新された所定値の重みベクトルを用いて、同じ１フレームのデータシンボル区間において、送信されたデータシンボルを合成する場合に一致する。
【００６７】
先に図１２を参照して説明した重みベクトルの更新方法では、「パイロット信号による伝搬路の推定→仮判定→パイロット信号とデータ信号による伝搬路の推定→判定」というように、伝搬路の推定とデータの判定を繰り返している。
したがって、受信における処理時間の遅延を許すならば、すなわち、電話のようなリアルタイム通信ではなく、パケット通信や、静止画像等のファイル転送であれば、伝搬路の推定を繰り返し行うことにより、重みベクトルの精度、すなわち、送信されたデータシンボルの推定精度を上げることが可能である。
【００６８】
図１４は、仮判定データシンボルを用いて行う重みベクトルの第３の更新方法を説明する模式図である。
図中、図１２と同じ処理を行うステップについては、同じステップ番号を付している。図１２のステップ（Ｓ２）とステップ（Ｓ３）との間に、パイロット信号とデータ信号とによる伝搬路の推定を行うステップ（Ｓ２１），（Ｓ２２）を挿入することにより、全ユーザk（k＝１〜K）について、データシンボルの仮判定を２回行っている。
【００６９】
同様にして、「パイロット信号による伝搬路の推定→仮判定→パイロット信号とデータ信号とによる伝搬路の推定→仮判定→パイロットとデータ信号による伝搬路の推定→仮判定→・・・」とし、繰り返し回数を１またはそれ以上として、複数回行うこともできる。
繰り返しのいずれの過程においても、データシンボル区間の始めからの所定のデータシンボル、すなわち、仮判定データシンボルは、最初の１シンボルだけでもよいし、データシンボル区間の全てのデータシンボルでもよい。
更新毎に仮判定データシンボルの数を変化させてもよい。例えば、重みベクトルの更新動作をする毎に、仮判定データシンボルの個数を逐次増加させる。より具体的には、仮判定データを最初はデータシンボル区間の最初のiシンボル（iは正整数）までとし、次回は、これを２iシンボルまでとし、更新回数が１回増える毎に、仮判定データの数を＋iずつ増加させる。
【００７０】
仮判定データシンボルの数が多くなるほど、理論的には伝搬路推定の精度が向上する。しかし、仮判定データシンボルの数が多くなるほど、送信データシンボルの仮判定誤りの影響が大きくなるし、また、処理に要する時間が長くなってしまう。したがって、このように、徐々に仮判定データシンボルの数を増やすことにより、更新動作に要する処理遅延を抑制しながら、重みベクトルの精度を、更新回数に応じて向上させることができる。
なお、重みベクトルの更新時における重みベクトルの計算方法は、本発明のようなインパルス応答による伝搬路の推定と、従来のMMSE型適応アルゴリズムとを、交互に使用したり、混在させたりすることが可能である。このほか、重みベクトルを計算できる方法が他にあれば、他の計算方法を用いてもよい。ただし、計算に使用するのに必要な情報やデータ等については、それぞれによって異なる。
【００７１】
図１５は、仮判定データシンボルを用いて行う重みベクトルの第４の更新方法を説明する模式図である。インパルス応答による伝搬路の推定と、従来のMMSE型適応アルゴリズムの両者を使用した例である。
図中、図１２，図１４と同じ処理を行うステップについては、同じステップ番号を付している。図１２のステップ（Ｓ２）とステップ（Ｓ３）との間に、従来のMMSE型適応アルゴリズムを用い、全ユーザk（k＝１〜K）に関して、パイロット信号とデータ信号とによる伝搬路の推定を行うステップ（Ｓ３１），（Ｓ２２）を挿入することにより、データシンボルの仮判定を２回行っている。この例においても、仮判定データシンボルは、最初の１シンボルだけでもよいし、データシンボル区間の全てのデータシンボルでもよいし、更新する毎に仮判定データシンボルの数を変化させてもよい。
【００７２】
上述した説明では、１フレームを単位として、重みベクトルの計算と、送信されたデータシンボルの合成処理を実行するものを説明した。しかし、伝搬路特性の変動など、周囲環境の変化が影響を及ぼさない限りにおいて、以前の１ないし数フレームにおける伝搬路推定データを、以後のフレームにおいても用いてアンサンブル平均を行ったり、以前の１ないし数フレームにおける重みベクトルを初期値として重みベクトルを更新したりしてもよい。
【００７３】
【発明の効果】
上述した説明から明らかように、本発明によれば、データ伝送用の送受信装置にパイロット信号を挿入し、受信したパイロット信号を時間軸上で処理してインパルス応答値を求めるようにしたため、インパルス応答推定用のDS-CDMA系の送受信設備が不要であるので、装置規模が小さくなるという効果がある。
上り回線にマルチキャリヤＣＤＭＡを用いる場合のように他局ユーザの干渉がある場合でも、この干渉除去が可能であり、また、従来のRLS法によるMMSE合成では重みベクトルが収束せずSINR等の向上が図れないような、少ないパイロットシンボル数でもMMSE合成が可能となるという効果がある。
さらに、仮判定データシンボルを用いれば、重みベクトルを更新して重みベクトルの精度が向上するという効果がある。
【図面の簡単な説明】
【図１】本発明のマルチキャリヤＣＤＭＡ受信装置とともに用いるマルチキャリヤＣＤＭＡ送信装置のブロック構成図である。
【図２】データシンボルとパイロットシンボルを時間多重したマルチキャリヤＣＤＭＡ方式のフレーム構成図である。
【図３】マルチキャリヤＣＤＭＡ方式の送信信号波形を示す波形図である。
【図４】本発明のマルチキャリヤＣＤＭＡ受信装置のブロック構成図である。
【図５】パイロット信号から重みベクトルを算出するための処理を示すフローチャートである。
【図６】インパルス応答の推定方法を説明する模式図である。
【図７】インパルス応答の補正方法を説明する模式図である。
【図８】単局ユーザのみで他局ユーザが存在しない場合において、インパルス応答の補正を行わないときのBER（ビット誤り率）特性を示す線図である。
【図９】単局ユーザのみで他局ユーザが存在しない場合において、インパルス応答値の補正を行ったときのBER（ビット誤り率）特性を示す線図である。
【図１０】多元接続を行うために他局ユーザが存在する場合における、ユーザ数に対する平均SINR（信号対干渉雑音比）特性を示す線図である。
【図１１】本発明のマルチキャリヤＣＤＭＡ受信装置の、他の実施の形態を示すブロック構成図である。
【図１２】図１１に示した受信機構成に対応し、仮判定データシンボルを用いて行う重みベクトルの第１の更新方法を説明する模式図である。
【図１３】仮判定データシンボルを用いて行う重みベクトルの第２の更新方法を説明する模式図である。
【図１４】仮判定データシンボルを用いて行う重みベクトルの第３の更新方法を説明する模式図である。
【図１５】仮判定データシンボルを用いて行う重みベクトルの第４の更新方法を説明する模式図である。
【符号の説明】
１…多重化部、２…ＱＰＳＫ変調器、３…分岐回路、４₁〜４_L…乗算器、５…逆離散フーリエ変換部、６…並直列変換器、７…ガードインターバル挿入部、１１…多重分離部、１２…ガードインターバル除去部、１３…チャネル推定部、１４…直並列変換部、１５…離散フーリエ変換部、１６₁〜１６_L…乗算器、１７…加算器、１８…ＱＰＳＫ復調器、１９…離散フーリエ変換部、２０…重みベクトル計算部[0001]
BACKGROUND OF THE INVENTION
The present invention multi-codes code division multiple access (MC) that spreads transmission data to a plurality of subcarriers by multiplying a symbol of transmission data by the value of each chip of the spreading code and modulating a plurality of orthogonal subcarriers. -CDMA: Multi-carrier-Code Division Multiple Access (Receiver) In particular, it is suitable for an uplink receiving apparatus in a mobile communication base station.
[0002]
[Prior art]
The 3rd generation mobile communication test service has begun, and multimedia communication in mobile communication is finally becoming a reality. Under such circumstances, studies on the next fourth generation mobile communication system have begun, and higher capacity transmission is required so that the Internet and multimedia communication can be comfortably performed similarly to fixed communication. .
The direct access code division multiple access (DS-CDMA) method used in IMT-2000 as a multiple access method for 4th generation mobile communication systems, and multicarrier code division that assigns each chip of a spread code on the frequency axis Multiple access (MC-CDMA) systems are being actively studied. MC-CDMA is known, for example, in S. Haraand R. Prasad, “Overview of multicarrier CDMA” IEEE Communications Magazine, vol. 35, no. 12, pp. 126-133, Dec. 1997.
In view of the current Web browsing and the like, asymmetric transmission is considered in which the capacity of the downlink is larger than that of the uplink.
In particular, MC-CDMA systems have been reported to be capable of higher capacity communication than DS-CDMA systems in the downlink from base stations to mobile stations, and are expected as candidates for fourth-generation mobile communication systems. Has been.
[0003]
However, in future communications, it is considered that there will be an increase in usage methods such as transmitting information such as images freely at a favorite place when an individual likes it. Therefore, in such a situation, it is necessary to increase the capacity of the uplink if not as much as the downlink.
In the uplink, since the propagation path from the mobile station to the base station is different for each mobile station (user), the orthogonal relationship of spreading codes assigned to each user is broken. Further, since each mobile station accesses the base station at random, the code synchronization is an asynchronous system.
Therefore, in the uplink, DS-CDMA that facilitates building an asynchronous system or DS-CDMA system that directly spreads each multi-carrier transmission data sequence is more promising than MC-CDMA.
[0004]
Recently, however, studies have started to use MC-CDMA in the uplink. If the symbol length is increased by transmitting a plurality of symbols in parallel, a quasi-synchronous system with a large margin on the time axis can be realized.
Since the multi-user reception is performed on the uplink, as described above, the orthogonality of the spread codes of each user is lost, interference of other station users (multi-user interference) occurs, and the characteristics are greatly deteriorated. Under such circumstances, if single-user reception is performed using only information on the local station without using information on other stations, multi-user interference cannot be removed, and thus reception performance cannot be improved.
[0005]
Therefore, by using a canceller technique that creates a replica of a received signal from another station and removes the received signal from the other station, interference from the other station can be suppressed and the system capacity can be increased. However, since a multi-stage configuration is used, processing delay becomes a problem.
On the other hand, the system capacity can be increased by using MMSE (Minimum Mean Square Error) combining that performs weighting when performing chip combination of spreading codes and removes multi-user interference. .
When chip synthesis is performed by MMSE synthesis, since it can be configured in a single stage, there is an advantage that a large processing delay does not occur. However, in order to obtain good characteristics in the MMSE synthesis, it is necessary to converge the weight vector.
[0006]
As an algorithm for realizing MMSE synthesis, LMS (Least Mean Square) method, SMI (Sample Matrix Inversion), RLS (Recursive Least Square) method, which is also known as an adaptive equalizer tap gain update algorithm, is used. it can.
The LMS method has a feature that can be realized with a simple configuration, but requires a lot of time for convergence of the weight vector.
The SMI method is a direct solution method using sample values, and it takes about twice as long as the spreading code length for convergence of the weight vector, and high-speed convergence is possible, but an inverse matrix operation is required. The device scale increases.
The RLS method also converges in about twice as long as the spreading code length, but it is possible to suppress an increase in device scale as compared with SMI.
[0007]
In this way, the LMS method requires a very large convergence time, and even when the RLS method is used, a convergence time that is about twice the spreading code length is required. During this convergence time, a pilot signal is transmitted. Is required.
In the future, when packet transmission is assumed, or when the spreading code length is increased as the number of multiplexing increases, it is necessary to improve the characteristics with a smaller number of pilot symbols. However, in the algorithm such as the RLS method described above, if the number of pilot symbols is small, a correct weight vector cannot be output, and the characteristics cannot be improved.
[0008]
The reason why the orthogonality between codes cannot be maintained is that the received signal is distorted by frequency selective fading of an independent channel for each user. Therefore, the method of estimating the characteristics of the propagation path by superimposing the direct spread pilot signal is, for example, Keisuke Hara, “Applicability of multi-carrier CDMA”, 2001 IEICE General Conference (Basic / Boundary) SB-4-1, March 2001, pp. Proposed in 779-780.
This proposal estimates the impulse response of the propagation path using the pilot signal directly spread in the time domain, calculates the frequency response from its Fourier transform, and calculates the received signal from the product of this frequency response and the spreading code. The correlation matrix and the distorted spreading code matrix are estimated.
[0009]
Here, the impulse response estimation method is, for example, Daiichi Imamura, Keisuke Hara, Norihiko Morinaga, “Subcarrier recovery method in OFDM using pilot signals”, IEICE Transactions B, Vol. J82-B, No. .3, March 1999, pp.393-401 etc.
In OFDM (Orthogonal Frequency Division Multiplexing), a direct spread pilot signal with a guard interval is inserted into a symbol sequence at regular intervals, and the direct spread pilot signal is input to a matched filter to estimate the impulse response of the propagation path. ing.
However, the above-described method for estimating a propagation path by superimposing a pilot signal by direct spreading requires two transmission / reception apparatuses, namely, an MC-CDMA system for data transmission and a DS-CDMA system for impulse response estimation. Therefore, there arises a problem that the scale of the transmission / reception apparatus increases.
[0010]
[Problems to be solved by the invention]
The present invention has been made to solve the above-described problems, and can suppress multi-user interference with a small number of pilot symbols to increase system capacity, and can reduce multi-carrier CDMA reception with a small apparatus scale. The object is to provide an apparatus.
It is another object of the present invention to provide a multi-carrier CDMA receiver that improves the accuracy of the weight vector by updating the weight vector.
[0011]
[Means for Solving the Problems]
According to the first aspect of the present invention, a symbol of a transmission symbol sequence in which a pilot symbol is time-multiplexed with a data symbol is multiplied by a value of each chip of a spreading code, and a plurality of orthogonal values are obtained by each of the multiplication values. A multi-carrier CDMA receiver for receiving a multi-carrier CDMA signal, wherein a sub-carrier is modulated and a guard interval is inserted, receiving the multi-carrier CDMA signal, extracting a pilot symbol period, Impulse response estimation means for outputting an impulse response estimation value of a propagation path from each transmission apparatus of all users based on the multi-carrier CDMA signal in the pilot symbol period and pilot symbol information of all users; Impulse response of the propagation path from the transmitter A vector obtained by discrete Fourier transform of a constant value, correlation vector generation means for generating a correlation vector of the desired user according to the spread code vector information of the desired user, and an impulse of a propagation path from each transmission device of all the users Interference noise removed from the impulse response estimation value when the impulse response estimation means outputs the impulse response estimation values of the propagation paths from the transmission devices of all the users, and the vector obtained by discrete Fourier transform of the response estimation value Correlation matrix generating means for generating a correlation matrix of the received multicarrier CDMA signal according to the component and spread code vector information of all users, and based on the correlation vector and the correlation matrix, A weight vector generating means for generating a weight vector, and the received multi-carrier A composite signal generating unit configured to multiply a vector obtained by discrete Fourier transform of a data symbol section of a CDMA signal and a conjugate transpose of the weight vector to synthesize a data symbol transmitted from the transmission device of the desired user; is there.
Therefore, multiuser interference can be suppressed with a small number of pilot symbols, the system capacity can be increased, and the apparatus scale can be reduced.
[0012]
According to a second aspect of the present invention, in the multicarrier CDMA receiver according to the first aspect, the correlation vector generating means performs a discrete Fourier transform on the impulse response estimation value of the propagation path from the transmission device of the desired user. And the spread code vector information of the desired user for each vector element, and the correlation matrix generation means, for all the users, the impulse response estimated value of the propagation path from each transmitting device A matrix obtained by multiplying a vector obtained by discrete Fourier transform of the vector and each spreading code vector information for each vector element and a conjugate transposed vector of each vector element, and adding the interference noise component to the user. A matrix obtained by multiplying the average value for all the users by a unit matrix is added.
Therefore, the correlation vector generation unit and the correlation matrix generation unit can be easily realized. In particular, the correlation matrix generation means calculates the interference component and the noise component separately. As a result, the weight vector estimation accuracy is improved.
[0013]
According to a third aspect of the present invention, in the multicarrier CDMA receiver according to the first aspect, the impulse response estimating means averages the impulse response estimated values calculated for a plurality of symbols of the pilot symbol. Output.
Therefore, the estimation accuracy of the impulse response is improved. As a result, the weight vector estimation accuracy is improved.
[0014]
According to a fourth aspect of the present invention, in the multicarrier CDMA receiver according to any one of the first to third aspects, the impulse response estimation means is a path within a guard interval and has a maximum value. The impulse response estimation value is output based only on a path exceeding a threshold set in accordance with the power path.
Therefore, the estimation accuracy of the impulse response is improved. As a result, the weight vector estimation accuracy is improved.
[0015]
According to a fifth aspect of the present invention, in the multicarrier CDMA receiver according to any one of the first to fourth aspects of the present invention, the weight vector generated by the weight vector generating means is used as an initial value, and the least square is used. There is provided weight vector updating means for sequentially updating the weight vector using an average error type adaptive algorithm.
Therefore, the weight vector estimation accuracy improves as it is sequentially updated. Moreover, the weight vector can be made to follow the fluctuation of the propagation path characteristic.
[0016]
According to a sixth aspect of the present invention, a multicarrier CDMA receiver includes the multicarrier CDMA receiver according to any one of the first to fourth aspects and weight vector update means, and the weight vector update The means causes the multicarrier CDMA receiver according to any one of claims 1 to 4 to output a weight vector for the desired user, and uses the weight vector for the desired user as an initial value. The multi-carrier CDMA signal according to any one of claims 1 to 4, wherein the weight vector is sequentially updated by a least mean square error adaptive algorithm for the multi-carrier CDMA signal in the pilot symbol period. The updated weight vector is given to the combined signal generating means in the carrier CDMA receiver. It is intended to combine the data symbols of the desired user.
Therefore, the weight vector estimation accuracy can be improved.
[0017]
In a seventh aspect of the present invention, a symbol of a transmission symbol sequence in which a pilot symbol is time-multiplexed with a data symbol is multiplied by a value of each chip of a spread code, and a plurality of orthogonal subcarriers are obtained by each of the multiplication values. A multicarrier CDMA receiver for receiving a multicarrier CDMA signal, which is modulated and inserted with a guard interval, receives the multicarrier CDMA signal, extracts a pilot symbol period, and extracts the pilot symbol period. Based on the multi-carrier CDMA signal and pilot symbol information of all users, impulse response estimation means for outputting impulse response estimation values of propagation paths from the transmission apparatuses of all users, and the transmission apparatus of desired users The impulse response estimate of the propagation path from Correlation vector generation means for generating a correlation vector of the desired user according to the Fourier transformed vector and the spread code vector information of the desired user, and an impulse response estimated value of the propagation path from each transmission device of all the users A discrete Fourier transform vector, and when the impulse response estimation means outputs an impulse response estimation value of a propagation path from each of the transmission devices of all users, an interference noise component removed from the impulse response estimation value, Correlation matrix generating means for generating a correlation matrix of the received multicarrier CDMA signal according to the spread code vector information of all users, and a weight vector for the desired user based on the correlation vector and the correlation matrix A weight vector generating means for generating the received multi-carrier CDMA; Data of at least one symbol transmitted from the transmitting device of the desired user by multiplying a vector obtained by discrete Fourier transform of at least one symbol from the beginning of the data symbol interval of the signal and a conjugate transposed of the weight vector. It has synthetic signal generation means for synthesizing symbols.
Therefore, it is possible to increase the system capacity by suppressing multiuser interference with respect to the data signal from the beginning of the data symbol period of the received multicarrier CDMA signal to at least one symbol with a small number of pilot symbols, and the scale of the apparatus. Can be reduced.
[0018]
According to an eighth aspect of the present invention, in a multi-carrier CDMA receiver, the multi-carrier CDMA receiver includes the multi-carrier CDMA receiver according to claim 7, weight vector updating means, and the weight vector updating means. 8. The multi-carrier CDMA receiver according to claim 7, wherein at least one desired data symbol of at least the desired user among all users is provisionally determined data symbols from the beginning of the data symbol period. And based on the pilot symbol information and the tentative decision data symbol of at least the desired user and the multicarrier CDMA signal from the beginning of the pilot symbol period and the data symbol period to at least one symbol. 8. The weight vector of a desired user is updated, and the updated weight vector is given to the combined signal generating means in the multi-carrier CDMA receiver according to claim 7, so that at least the data symbol of the desired user is provided. Are synthesized.
Therefore, the accuracy of the weight vector can be improved by updating the weight vector with the pilot signal and the data signal from the beginning of the data symbol interval to at least one symbol.
The update of the weight vector can be realized by executing the impulse response estimation of the propagation path or executing the conventional least mean square error type adaptive algorithm.
[0019]
In a ninth aspect of the present invention, in the multicarrier CDMA receiver according to the eighth aspect, the weight vector updating means repeats the updating operation a plurality of times.
Therefore, the estimation accuracy of the weight vector can be further improved.
For example, the multi-carrier CDMA receiver according to claim 7 is reused a plurality of times, the conventional least mean square error type adaptive algorithm is used once or a plurality of times, and both are mixed. Can be used.
[0020]
According to a tenth aspect of the present invention, in the multicarrier CDMA receiver according to the ninth aspect, the multicarrier CDMA receiver according to the seventh aspect is provided each time the weight vector updating means performs the updating operation. The number of the data symbols to be combined as the provisional determination data symbol is sequentially increased.
Therefore, it is possible to improve the accuracy of the weight vector according to the number of updates while suppressing the processing delay required for the update operation.
[0021]
According to an eleventh aspect of the present invention, the multicarrier CDMA receiving apparatus includes the multicarrier CDMA receiving apparatus according to the seventh aspect and weight vector updating means, and the weight vector updating means corresponds to the seventh aspect. For the multicarrier CDMA receiver described above, each of all the users is set as the desired user, and data symbols of all the users from the beginning of the data symbol period to at least one symbol are combined as provisional decision data symbols. The pilot symbol information and the provisional decision data symbol information of all the users are regarded as new pilot symbol information, and at least one symbol from the beginning of the pilot symbol period and the data symbol period is renewed. Pilot symbol It considers, for multi-carrier CDMA receiving apparatus according to claim 7, by updating the weight vector, is intended to combine the data symbols of the desired user.
Therefore, the process of updating the weight vector with the pilot signal and the data signal can be efficiently performed by re-activating the multicarrier CDMA receiver according to claim 7.
[0022]
According to a twelfth aspect of the present invention, a multicarrier CDMA receiver includes the multicarrier CDMA receiver according to the seventh aspect and weight vector updating means, and the weight vector updating means is provided in the seventh aspect. The multicarrier CDMA receiver described above is configured to synthesize at least one symbol of the desired user from the beginning of the data symbol period as a provisional determination data symbol, and output a weight vector for the desired user, The pilot symbol information and the provisional decision data symbol information are regarded as new pilot symbol information, and the weight vector for the desired user is used as an initial value from the beginning of the pilot symbol period and the data symbol period. At least 1 8. The weight vector is sequentially updated by a least mean square error type adaptive algorithm for the multi-carrier CDMA signal up to a symbol, and the combined signal generating means in the multi-carrier CDMA receiver according to claim 7 The updated weight vector is given to synthesize the data symbol of the desired user.
Therefore, the conventional least mean square error type adaptive algorithm can be used for the process of updating the weight vector with the pilot signal and the data signal.
[0023]
DETAILED DESCRIPTION OF THE INVENTION
FIG. 1 is a block diagram of a multicarrier CDMA transmitter used with the multicarrier CDMA receiver of the present invention.
In the figure, 1 is a multiplexing unit (multiplexer), 2 is a QPSK (Quadrature Phase Shift Keying) modulator, 3 is a branch circuit, 4 ₁ ~ 4 _L Is a multiplier, 5 is an inverse discrete Fourier transform (IDFT), 6 is a parallel / serial converter, and 7 is a guard interval insertion unit.
Data (configured in units of one symbol) and pilot (configured of known data in units of one symbol) are time-multiplexed by the multiplexing unit 1.
[0024]
The time-multiplexed signal is QPSK modulated in the QPSK modulator 2 to generate a symbol represented by a narrow band complex baseband signal (in-phase and quadrature signal). The q-th symbol data is represented as d (q).
This complex baseband signal is output in the branch circuit 3 by the number of subcarriers, and is multiplied by the multiplier 4 of each subcarrier. ₁ ~ 4 _L 2 is multiplied by the value of each chip of the spread code to obtain the complex amplitude of the subcarrier.
Note that the modulation method is not limited to QPSK.
[0025]
FIG. 2 is a frame configuration diagram of a multicarrier CDMA system in which data symbols and pilot symbols are time-multiplexed. Since data and pilot are multiplexed in the multiplexing unit 1 in units of one symbol, the QPSK modulator 2 generates pilot symbols and data symbols.
In this example, one frame is constituted by one or a plurality of pilot symbols and a plurality of subsequent data symbols. When packet transmission is assumed, transmission is performed once with this one frame as a packet unit. When continuous transmission is assumed, a plurality of frames are repeatedly transmitted.
The propagation path is estimated using the pilot symbols, and multi-user interference is removed. The smaller the number of pilot symbols, the smaller the load on the header part and the better the transmission efficiency.
[0026]
A spreading code having a chip length L is represented as a spreading code vector C, and is represented by the following equation.
[Expression 1]

here,[·] ^T Represents a transposed matrix.
A spreading code uses a different code for each user. The number of chips of the spreading code (number of elements of the spreading code vector) is equal to the number of subcarrier channels. The subcarrier frequencies are orthogonally arranged as in OFDM.
The inverse discrete Fourier transform unit 5 shown in FIG. 1 inputs the complex amplitude of each subcarrier, performs inverse discrete Fourier transform, and outputs a waveform signal on the time axis. Since it is output as a parallel digital signal, the parallel-serial converter 6 converts it into a serial digital signal and outputs it.
The guard interval insertion unit 7 inserts the guard interval Δ to generate a multicarrier CDMA transmission signal.
[0027]
In the configuration shown in FIG. 1, only one symbol is transmitted at the same time for the QPSK modulated signal. However, in order to increase the length of one symbol to facilitate quasi-synchronization, a plurality of QPSK modulators 2 are provided to distribute multiplexed data. In addition, a plurality of symbols output from a plurality of QPSK modulators are simultaneously transmitted in one symbol length by using a spreading code having a chip number several times longer and a subcarrier number several times.
In order to express the concept including such a case, in this specification, it is assumed that one symbol is transmitted in one symbol section by regarding the entire transmission signal waveform in one symbol section as a single symbol waveform. However, even in this case, a pilot symbol section is configured using known pilot symbols.
[0028]
FIG. 3 is a waveform diagram showing a transmission signal waveform of the multicarrier CDMA system.
Since the subcarriers are arranged orthogonally like OFDM, the transmission waveform is similar to that of OFDM.
The guard interval is inserted in order to eliminate intersymbol interference caused by delayed waves. The illustrated guard interval is obtained by copying the end portion of the original transmission signal waveform subjected to the inverse discrete Fourier transform and inserting it in the previous portion of the original transmission signal waveform.
In the receiving apparatus, if the illustrated transmission signal waveform is a direct wave and the delayed wave arrives with a delay time within the guard interval length Δ, the influence of intersymbol interference caused by such a delayed wave can be suppressed. It becomes possible.
That is, on the receiving side, if a section (having the same length as the original transmission signal waveform) that is shorter than the one symbol section by the guard interval length Δ is cut out and used, the delayed wave included in the extracted waveform is Since this is a waveform obtained by simply delaying the direct wave included in the cut out waveform, interference due to the delayed wave does not occur.
[0029]
When the receiving device does not always use the waveform that arrives first, intersymbol interference occurs due to the arriving wave that precedes this waveform. In this case, as is well known, a copy of the original transmission signal waveform (length Δ) is copied as it is into the rear part of the original transmission signal waveform, so that the guard interval is also added to the rear part. Insert. On the receiving side, a section shorter than the total length (2Δ) of the guard interval than one symbol section is cut out and used.
With the above configuration, a transmission signal waveform in which transmission data is spread in the frequency domain is generated. This transmission signal waveform is converted into a radio frequency band using the main carrier and transmitted.
[0030]
The transmission signal x (q) is expressed by the following equation in the time domain.
[Expression 2]

Where F is the discrete Fourier transform, F ^-1 Represents the inverse discrete Fourier transform.
L sample points on the time axis x ₁ , X ₂ , X _Three , ..., x _L Is a complex amplitude of the transmission signal at sample points 1 to L obtained by equally dividing a section excluding the guard interval (that is, the original transmission signal waveform before inserting the guard interval) in one symbol period.
[0031]
The discrete Fourier transform F is represented by the following matrix.
[Equation 3]

Corresponding to the L sample points, the frequency of the subcarrier is the sample time T _s Then, in baseband,-(L / 2-1) / (LT _s ),-(L / 2-2) / (LT _s ),-(L / 2-3) / (LT _s ), ……, 0,1 / (LT _s ), 2 / (LT _s ), 3 / (LT _s ), ……, (L / 2) / (LT _s ).
[0032]
FIG. 4 is a block diagram of the multi-carrier CDMA receiver according to the present invention. In the figure, 11 is a demultiplexing unit (demultiplexer), 12 is a guard interval removal unit, 13 is a channel estimation unit, 14 is a serial-parallel conversion unit, 15 is a discrete Fourier transform (DFT) unit, 16 ₁ ~ 16 _L Is a multiplier, 17 is an adder, 18 is a QPSK demodulator, 19 is a discrete Fourier transform (DFT) unit, and 20 is a weight vector calculation unit.
The received signal is demultiplexed into a data signal section and a pilot signal section in the demultiplexing unit 11 based on the synchronization timing. The guard interval removal unit 12 removes the guard interval from the received signal, converts the guard interval into a parallel signal by the serial-parallel conversion unit 14, and outputs the parallel signal to the discrete Fourier transform (DFT) unit 15.
[0033]
A desired propagation path characteristic from a mobile station to a base station of a single user is represented by a propagation delay profile h (t) expressed by the following equation.
[Expression 4]

Where a _l Is the component of the l (lowercase L) path of the propagation vector a, T _s Is the sample time and δ (t) is the Dirac delta function. Therefore, the delay time of the lth path is (l-1) T _s It becomes.
[0034]
The received signal y (q) (time domain) is expressed by the following equation using Equations (2) and (4) because the transmission signal x (q) is subject to fluctuations in the propagation path. However, it is a received signal after the guard interval is removed.
[Equation 5]

Here, x (q; l) indicates that the transmission signal x (q) is (l-1) T _s The signal delayed by n, n represents an interference noise component.
[0035]
A supplementary description will be given of the characteristics of the delayed signal in the above-described equation.
As shown in FIG. 2, in the guard interval, the transmission signal waveform is a signal obtained by copying the rear part of the original transmission signal waveform. Therefore, if we show the complex amplitude of the sample points in reverse order from the end of the guard interval, x _L , X _L-1 , X _L-2 , X _L-3 …….
Assuming that all the delayed signals x (q; 1), x (q; 2),... Fall within the guard interval length Δ, the delayed signal has the complex amplitude of the sample points in the guard interval. From the feature, the direct wave x (q; 1) is cyclically shifted in the section where the discrete Fourier transform is performed.
However, not all L waves from l = 1 to L are always present as delayed signals that actually arrive at the base station. For example, if there is no second incoming delayed signal, a ₂ = 0.
[0036]
The discrete Fourier transform (DFT) unit 15 performs discrete Fourier transform, and outputs the complex amplitudes of a plurality of orthogonal subcarriers as a reception signal subjected to discrete Fourier transform.
A reception signal subjected to discrete Fourier transform, that is, a reception signal Y (q) in the frequency domain is given by the following equation.
[Formula 6]

Where J ^l-1 Is a diagonal matrix in which each element of the diagonal matrix is raised to the power of (l-1). The symbol “x” with a circle is an operation for generating a vector having each element obtained by multiplying each element of two vectors.
It should be noted that in the above-described modification in the equation (6), the fact that the relationship of the following equation holds is used.
[Expression 7]

[0037]
Multiplier 16 corresponding to each subcarrier ₁ ~ 16 _L , The weights output from the weight vector calculator 21 are multiplied. This weight is obtained by weighting the values of the spreading code chips 1 to L assigned to a desired single user and used at the time of transmission.
All multipliers 16 ₁ ~ 16 _L Are added and synthesized by the adder 17 to become a QPSK-modulated symbol represented by a complex baseband signal. The QPSK demodulator 18 determines the level of the in-phase component and the quadrature component and demodulates the received data. Is output.
[0038]
Appropriate weight vector for MC-CDMA received signal vector Y
[Equation 8]

Based on the above, intersubcarrier signal synthesis is performed.
The final combined output output from the adder 17 is expressed by the following equation.
[Equation 9]

here,[·] ^† Indicates a matrix (Hermitian matrix) obtained by transposing (conjugate transposing) a matrix having the complex conjugate of each element of the matrix as an element.
[0039]
This weight vector w is preferably determined so that the signal power is large and the interference noise power is small. As a method for satisfying such a requirement, the MMSE synthesis standard mentioned in the prior art is used. The weight vector w is obtained by the following equation.
[Expression 10]

Here, R is a correlation matrix, and v is a correlation vector. R ^-1 Is the inverse of R, E [•] is the ensemble average, (•) ^* Represents a complex conjugate.
Correlation matrix R is an ensemble average of correlation values between elements (complex amplitudes of subcarriers) of reception signal vector Y (q) obtained by discrete Fourier transform of the reception signal. It is an ensemble average of correlations including not only incoming waves from mobile stations of desired users but also incoming waves from mobile stations of all users.
On the other hand, the correlation vector v is an ensemble average of the correlation between the received signal vector Y (q) obtained by discrete Fourier transform of the received signal and the transmission symbol data d (q) of the desired user's mobile station.
[0040]
In accordance with the first relational expression of R in the above equation (9), the ensemble average of the first correlation expression regarding the received signal Y (q) subjected to the discrete Fourier transform is taken, and the first of v The weight vector w can be obtained by taking the ensemble average of the second correlation equation for the received signal Y (q) subjected to discrete Fourier transform according to the relational expression.
However, in the multicarrier CDMA receiver of the present invention, after obtaining the propagation vector a based on the impulse response of the pilot symbol, the correlation is performed according to the second relational expression of R and v in the above-described equation (9). A method of calculating the matrix R and the correlation vector v is taken.
The process is realized by the channel estimation unit 13, the discrete Fourier transform unit 19, and the weight vector calculation unit 20 shown in FIG. 4, and details of the processing will be described with reference to FIG.
[0041]
FIG. 5 is a flowchart showing a process for calculating a weight vector from a pilot signal.
FIG. 6 is a schematic diagram for explaining an impulse response estimation method.
In S31, the propagation vector a is estimated from the estimated value of the impulse response using the pilot signal.
Note that propagation vectors of propagation paths from mobile stations of all users are estimated. However, in the description using the following mathematical expressions, for the sake of simplicity, the propagation vector of the propagation path from the mobile station of the desired user is represented to the middle, and this will be described as a. Finally, for the users k = 1 to K, the propagation vector a _k Calculate
[0042]
In pilot signal sample q, correlation φ between known transmission signal X (q) and reception signal y (q) _s (Q) is the sample difference s,
[Expression 11]

It should be noted that the relationship of the following equation was used in the conversion in equation (10).
[Expression 12]

[0043]
In equation (10), x (q; s) and n are uncorrelated, so x (q; s) ^† n takes a different value for each symbol. Therefore, φ _s By taking an ensemble average over the symbol value q of (q), x ^† (Q; s) n disappears and the propagation coefficient a _s Is obtained.
As shown in FIG. 6, the propagation vector a = {a is obtained by shifting the value of s by one sample and performing a correlation operation on the delayed signal x (q; s) of the impulse response. ₁ , A ₂ , ……, a _L } Can be estimated.
Assuming that all delayed signals fall within the guard interval, it is only necessary to calculate the impulse response during the guard interval.
The correlation calculation can be performed using a matched filter.
[0044]
In the next S32, the characteristics are improved by correcting the impulse response.
FIG. 7 is a schematic diagram illustrating a method for correcting an impulse response.
In FIG. 7A, the thick arrow represents the power value of the impulse response.
First, the system is designed so that all major delayed signals (represented as delay paths) fall within the guard interval, so all delay paths outside the guard interval are considered as interference noise (interference or noise). Cut.
Next, T is higher than the path with the largest power (in the example shown, the direct path corresponding to the direct wave). _h A path lower than [dB] (second path in the illustrated example) is also considered as interference noise and cut. In FIG. 7, the cut path is marked with an X mark.
By the above operation, the impulse response shown in FIG. 7A is corrected to the impulse response shown in FIG.
[0045]
Returning to FIG. 5 again, the description will be continued.
Thus, in the following description, the propagation vector a for each user k _k Will be described.
In S33, impulse responses a, which are the propagation path estimation results, for the propagation paths of all users including the desired user (k = 1), respectively. _k To Fourier transform _k Get.
In S34, Fa according to the expression of the second row of the correlation vector v of Expression (9). ₁ And spreading code vector C ₁ Correlation vector v is calculated from (spread code assigned to user k = 1).
On the other hand, in S35, for all users (k = 1 to K), the noise power P is calculated from the power of the path cut by the impulse response correction processing in S32 described above. _N Is estimated.
[0046]
Noise power P _N The estimation of is described in detail.
The received signal from the mobile station of the user other than the desired user is not necessarily symbol-synchronized with the received signal from the mobile station of the desired user. For this reason, within one symbol of the received signal from the mobile station of the desired user, the received signal from the mobile station of the other user can partially include two adjacent symbols.
However, to simplify the calculation here, it is assumed that other users also transmit symbols at the same timing as the desired user.
The system is designed so that the mobile station transmits a transmission signal in synchronization with the downlink pilot signal supplied by the base station on the downlink, and the mobile station is within a predetermined distance from the base station. Since the received signal from the user's mobile station can be accommodated within the guard interval, it is equivalent to other users transmitting symbols at the same timing as the desired user within the range of discrete Fourier transform.
[0047]
In the above-mentioned second formula of R in formula (9), Fnn ^† F ^† The term is difficult to evaluate because the interference component with other users and the noise component are mixed.
Therefore, with reference to equation (9), the correlation matrix taking into account the propagation vectors of other users is as follows:
[Formula 13]

Where a _k Is the propagation vector of user k, C _k Is the spreading code vector of user k, P _N Represents noise power, and I represents a unit matrix.
P _N Is the propagation vector a for each user k in S32 of FIG. _k , The total power of the cut path is obtained, and the average value of this total power for all users k is obtained as the noise power P _N And
Although the multi-user interference and the noise are not completely evaluated by the above-described equations, the correlation matrix R is obtained after evaluating them substantially separately.
In S36, the correlation matrix R is calculated.
In S37, a weight vector is calculated. Estimation results (propagation vector a) of all users k subjected to Fourier transform _k ) And the spreading code vector C assigned to all users k _k Information (replica) and noise power P _N Based on the estimated value, the weight vector w of the desired user (k = 1) shown in Equation (9) is calculated.
[0048]
Finally, the results of computer simulation of the multicarrier CDMA receiver of the present invention will be described with reference to FIGS.
FIG. 8 is a diagram showing BER (bit error rate) characteristics when the impulse response estimation value is not corrected when only a single station user is present and no other station user is present (no multi-user interference).
E _b / N _o The BER against (signal energy per bit / one-sided noise power spectral density) is the number of pilot symbols (q ₀ ) As a parameter.
The spreading code vector C uses an orthogonal Gold code having a code length (number of chips) of 32. There are no interfering mobile stations and an additional white Gaussian noise (AWGN) transmission line.
At this time, the bit error rate characteristic is the number of pilot symbols (q ₀ The smaller the), the more the characteristics deteriorate, which is considered to be caused by an estimation error of the impulse response estimation value.
[0049]
FIG. 9 is a diagram showing BER (bit error rate) characteristics when the impulse response estimation value is corrected when only a single station user is present and no other station user exists (no multi-user interference).
In the figure, for comparison, a part of the BER characteristic when the impulse response estimation value is not corrected is partially copied from FIG. 8 and shown in black.
The conditions such as the propagation path are the same as in FIG. It can be seen that the bit error rate characteristic is improved by correcting the impulse response estimation value shown in FIG. Here, the threshold T _h However, it is necessary to select an optimum value depending on the propagation path.
[0050]
FIG. 10 is a diagram showing an average SINR (signal-to-interference and noise ratio) characteristic with respect to the number of users when another station user exists for multi-access (there is multiuser interference).
E _b / N _o = 30 [dB], number of pilot symbols q ₀ = 8, propagation path is a two-wave model, fading fluctuation is sufficiently slow, propagation vector a in one frame (a set of pilot and data symbols) _k Is constant. For simplification, it was evaluated with a synchronous system.
As comparison targets, a case in which a fading compensation method used in single user reception is used and a case in which an MMSE combining method based on an RLS algorithm known for rapid convergence are also illustrated.
[0051]
In the fading compensation method for single user reception, since the propagation path characteristics of other stations are not taken into consideration, SINR rapidly decreases in two-user multiplexing. In addition, since the number of pilot symbols is insufficient in RLS, the weight vector does not converge, and the SINR is greatly reduced when the number of multiplexing is large. On the other hand, in the method of the present invention, although the SINR decreases as the number of multiplexing increases, no rapid deterioration is observed, and it can be seen that better characteristics are obtained than the two compared methods.
[0052]
In the above description, the weight vector is calculated using the propagation vector based on the impulse response. However, the weight vector calculated in this way may be used as an initial value, and the weight vector may be sequentially updated using a conventional MMSE type adaptive algorithm for calculating a weight vector. As the number is sequentially updated, the estimation accuracy of the weight vector is improved, and the weight vector can follow the propagation path fluctuation.
[0053]
At this time, from the first pilot symbol of the same frame to the last data symbol of this frame, the weight vector previously calculated is used as an initial value, and the weight vector is sequentially updated using the MMSE type adaptive algorithm. Alternatively, the weight vector is sequentially updated from the beginning to the end of the same one frame data symbol using the previously calculated weight vector as an initial value using an MMSE type adaptive algorithm.
In any case, in the data symbol interval, the MMSE type adaptive algorithm is applied by using the data symbols synthesized one after another using this weight vector as a temporary decision data symbol of the transmitted data symbol.
Alternatively, the weight vector calculated earlier may be sequentially updated using the MMSE-type adaptive algorithm in the pilot symbol section of the same one frame, using the previously calculated weight vector as an initial value. The transmitted data symbols may be combined in the same one-frame data symbol section using the updated predetermined weight vector.
[0054]
In addition, a data symbol synthesized using a weight vector generated in a pilot symbol section in one frame can be regarded as a pilot symbol together with the pilot symbol as a provisionally determined data symbol. Therefore, the transmitted data symbols can be synthesized by updating the weight vector again based on the impulse response estimation.
Further, in a new pilot symbol period, the weight vector is sequentially updated by the conventional MMSE type adaptive algorithm, and the transmitted data symbol can be synthesized using the updated weight vector.
An embodiment for improving the accuracy of the weight vector using the temporarily determined data symbol will be described below.
[0055]
FIG. 11 is a block diagram showing another embodiment of the multi-carrier CDMA receiver according to the present invention. In the figure, parts similar to those in FIG.
41 is a channel estimation unit, and 42 is a data symbol replica generation unit.
The data symbol replica generation unit 42 tentatively determines a transmitted data symbol, assuming that the received data subjected to QPSK demodulation is transmitted data. Based on the information of the temporary determination data symbol, a transmission signal (waveform) of the temporary determination data symbol, that is, a data symbol replica is generated.
[0056]
The channel estimation unit 41 inputs the data symbol replica output from the data symbol replica generation unit 42 as the data symbol replica of the desired user. At the same time, similarly to the above-described receiving device of the desired user, the receiving data is provisionally determined to be transmitted data from all other receiving devices of the user, and based on the information of the temporary determination data symbol of each user. Enter the generated data symbol replica.
The data symbol replicas of all users including the desired user and other users are known as the pilot symbol replicas of all users used in FIG.
[0057]
Therefore, a symbol replica in one frame including a pilot symbol replica and a data symbol replica of each user can be used in the same manner as the pilot symbol replica of each user used in FIG.
As a result, the channel estimator 41 also receives the received signal in the data symbol period, and can perform channel estimation in the data symbol period in addition to the pilot symbol period.
Here, it is not essential to make a temporary determination for all data symbols in the data symbol section. If the data symbol from the beginning of the data symbol interval to at least one symbol is provisionally determined, the number of known symbols can be increased and the weight vector can be calculated over a longer period, thereby improving the accuracy of the weight vector.
However, not only the desired user but also other users' temporary determination data symbols may be required. However, in the “multi-user reception” like the uplink reception in the base station, the provisional determination data of other users can be easily used, but there is no particular problem although the processing amount increases.
[0058]
The data symbol replica generation unit 42 will be described supplementarily.
Based on the received data as provisional determination data, a data symbol replica is generated on the side of the receiving device. A block that realizes this function is used as a data symbol replica generation unit 42.
More specifically, similar to the transmission apparatus configuration of FIG. 1, tentative determination data symbols are generated by QPSK modulation of temporary determination data, branching, multiplication of spreading codes, despread Fourier transform, parallel-serial conversion, guard interval. Insert.
Therefore, it is as if the receiving apparatus shown in FIG. 11 has a transmitting apparatus, which complicates the apparatus scale. However, since there are only four patterns of transmission signals (data symbol replicas) in the QPSK symbol, four patterns of transmission signals are stored in the memory space, and the memory space is determined according to the pattern corresponding to the temporarily determined data symbol information. If the transmission signal is read out from the above, complicated processing such as the inverse diffusion Fourier transform described above becomes unnecessary.
[0059]
In the following description, a certain frame is provisionally determined from the beginning of the data symbol period to a predetermined data symbol (at least one symbol). Using this together with the pilot symbol, a calculation is performed to update the weight vector for the pilot symbol period and the data symbol period of the same frame from the beginning to a predetermined data symbol, and using the updated weight vector, The transmitted data symbols are combined again in all data symbol sections of the same frame.
However, since processing delay occurs, it is suitable for packet data of one frame length, file transfer, and the like.
Here, the “predetermined data symbol” described above may be one symbol or all data symbols in a data symbol section.
In FIG. 11, for comparison with FIG. 4, the pilot symbol replica and the data symbol replica are separately generated. However, a replica may be generated collectively for a combination of pilot symbols in one frame and a predetermined data symbol from the beginning of the data symbol interval.
[0060]
FIG. 12 is a schematic diagram for explaining a first updating method of the weight vector performed using the provisional determination data symbol. This corresponds to the receiver configuration shown in FIG.
In step (S1), the same processing as in FIG. 6 is performed in the pilot symbol section in one frame. However, not only the desired user but also the received signals of all users k (k = 1 to K) are performed. That is, using a replica of pilot symbols of all users k, the weight vector of all users k (hereinafter referred to as w _k Respectively).
[0061]
In step (S2), received signals of all users k (k = 1 to K) and weight vectors w of all users k _k Are used to temporarily determine the data symbols of all users k. That is, the weight vector w of all users k obtained using pilot symbols _k To temporarily determine data symbols of all users.
The processing interval is from the beginning of the data symbol interval to a predetermined data symbol. As already described, all data symbols in the data symbol section may be used.
From the beginning of the data symbol interval in one frame to a predetermined data symbol, the data symbol can be regarded as the same as the pilot symbol. For all users k, a replica (time axis) of transmission signals in one frame (from the beginning of the pilot symbol period and the data symbol period to a predetermined data symbol) based on the data symbols provisionally determined to be known pilot symbols. Signal).
[0062]
In step (S3), the received signal (from the beginning of the pilot symbol period and the data symbol period to a predetermined data symbol) and the already obtained replicas of [pilot symbols + provisional determination data symbols] of all users k are newly set. The impulse response value is estimated again using what is regarded as the pilot symbol, and the weight vector w of the desired user is recalculated.
In step (S4), the data symbol of the desired user is synthesized (determined) using the received signal and the weight vector w of the desired user in the data symbol period.
[0063]
FIG. 13 is a schematic diagram illustrating a second updating method of the weight vector performed using the provisional determination data symbol.
Basically, the receiver configuration shown in FIG. 11 is adopted, but the configuration for updating the weight vector is different. The estimation of the impulse response is the same as in the upper half of the explanatory diagram shown in FIG.
In FIG. 12 described above, in step (S3), recalculation for updating the weight vector is performed by estimating the impulse response, whereas in this second updating method, the conventional MMSE (minimum 2) is used. It is updated by a mean square error) type adaptive algorithm. Due to the difference in the update calculation, the processing of steps (S1) and (S2) until the generation of the temporary determination data symbol is changed, and only the data symbol of the desired user needs to be temporarily determined.
Further, it is only necessary to obtain provisional determination data symbol information, and a replica thereof is not required. The same applies to step (S4).
When the number of symbols in the frame is sufficiently large, the MMSE type adaptive algorithm based on the sequential updating of the weight vector is effective.
[0064]
In step (S11), the same processing as in FIG. 6 is performed in the pilot symbol section in one frame. A weight vector w of a desired user is calculated using a replica of pilot symbols of all users.
In step (S12), the received signal and the weight vector w of the desired user are used from the beginning of the data symbol period in one frame to a predetermined symbol (at least one symbol or all data symbols in the data symbol period). The data symbol of the desired user is provisionally determined.
For the desired user, transmission symbol (pilot symbol and provisionally determined data symbol) information in one frame is reconstructed based on the known pilot symbol and the temporarily determined data symbol.
[0065]
In step (S13), the weight vector w of the desired user is recalculated by the MMSE type adaptive algorithm for each pilot signal and data signal of the received signal in one frame.
Specifically, the weight vector w obtained in the pilot symbol period of step (S11) is used as an initial value, the received signal (from the beginning of the pilot symbol period and the data symbol period to a predetermined data symbol), and step (S12). The accuracy of the weight vector is improved by recalculating the weight vector w of the desired user by using the information regarding the [pilot symbol + provisional determination data symbol] information of the desired user obtained in step 1 as pilot symbol information. .
[0066]
In the MMSE type adaptive algorithm, as described above, the ensemble average of the first correlation equation for the received signal Y (q) subjected to the discrete Fourier transform is calculated according to the first relational expression of R in the equation (9). Take.
On the other hand, according to the first relational expression of v, an ensemble average of the second correlation expression regarding the received signal Y (q) subjected to discrete Fourier transform is taken. Where d ^* The value of the transmission symbol (pilot symbol + provisional determination data symbol) of the desired user described above is substituted into (q). As a result, finally, at the end of one frame, w = R in equation (9) ^-1 The weight vector w is updated by v.
In step (S4), as in FIG. 12, the data symbol of the desired user is determined in the data symbol period.
In this updating method, if (S12) is omitted and the number of provisional determination data symbols is set to 0, the pilot symbol section of the same one frame with the previously calculated weight vector as an initial value as described above is used. This is the case when the weight vector is sequentially updated using the MMSE-type adaptive algorithm and the transmitted data symbols are synthesized in the same one-frame data symbol section using the updated weight vector of the predetermined value.
[0067]
In the weight vector updating method described above with reference to FIG. 12, the propagation path estimation is performed as follows: “estimation of propagation path using pilot signal → provisional determination → estimation of propagation path using pilot signal and data signal → determination”. And data determination is repeated.
Therefore, if delay of processing time in reception is allowed, that is, if it is not real-time communication such as a telephone but packet communication or file transfer such as a still image, the weight vector is obtained by repeatedly estimating the propagation path. That is, the estimation accuracy of the transmitted data symbols can be improved.
[0068]
FIG. 14 is a schematic diagram illustrating a third updating method of the weight vector performed using the provisional determination data symbol.
In the figure, steps that perform the same processing as in FIG. 12 are given the same step numbers. By inserting steps (S21) and (S22) for estimating a propagation path using a pilot signal and a data signal between step (S2) and step (S3) in FIG. 12, all users k (k = k = 1 to K), provisional determination of data symbols is performed twice.
[0069]
Similarly, “estimation of propagation path by pilot signal → temporary determination → estimation of propagation path by pilot signal and data signal → temporary determination → estimation of propagation path by pilot and data signal → temporary determination → ...” The number of repetitions can be one or more and can be performed a plurality of times.
In any process of repetition, the predetermined data symbol from the beginning of the data symbol period, that is, the provisional determination data symbol may be only the first one symbol or all data symbols in the data symbol period.
You may change the number of temporary determination data symbols for every update. For example, each time the weight vector is updated, the number of provisional determination data symbols is sequentially increased. More specifically, the provisional decision data is initially set to the first i symbols (i is a positive integer) in the data symbol section, and next time, this is set to 2i symbols. Increase the number of data by + i.
[0070]
Theoretically, the accuracy of channel estimation improves as the number of provisional determination data symbols increases. However, as the number of provisional determination data symbols increases, the influence of the provisional determination error of the transmission data symbol increases, and the time required for processing increases. Therefore, by gradually increasing the number of temporary determination data symbols in this way, the accuracy of the weight vector can be improved according to the number of updates while suppressing the processing delay required for the update operation.
Note that the weight vector calculation method at the time of updating the weight vector may alternately use or mix the propagation path estimation based on the impulse response as in the present invention and the conventional MMSE type adaptive algorithm. Is possible. In addition, if there is another method that can calculate the weight vector, another calculation method may be used. However, information and data necessary for use in the calculation differ depending on each.
[0071]
FIG. 15 is a schematic diagram illustrating a fourth updating method of the weight vector performed using the provisional determination data symbol. This is an example using both the estimation of the propagation path by the impulse response and the conventional MMSE type adaptive algorithm.
In the figure, steps that perform the same processing as in FIGS. 12 and 14 are given the same step numbers. Between step (S2) and step (S3) in FIG. 12, a conventional MMSE type adaptive algorithm is used to estimate the propagation path using pilot signals and data signals for all users k (k = 1 to K). By inserting the steps (S31) and (S22) to be performed, provisional determination of the data symbol is performed twice. Also in this example, the provisional determination data symbol may be only the first one symbol, all the data symbols in the data symbol section, or the number of provisional determination data symbols may be changed each time the data is updated.
[0072]
In the above description, the calculation of the weight vector and the synthesis process of the transmitted data symbol has been described for each frame. However, as long as changes in the surrounding environment, such as fluctuations in propagation path characteristics, do not affect the ensemble average using the propagation path estimation data in the previous one or several frames in the subsequent frames, Alternatively, the weight vector may be updated using a weight vector in several frames as an initial value.
[0073]
【The invention's effect】
As is apparent from the above description, according to the present invention, the pilot signal is inserted into the data transmission / reception apparatus, and the received pilot signal is processed on the time axis to obtain the impulse response value. Since no estimation DS-CDMA transmission / reception equipment is required, the apparatus scale is reduced.
Even when there is interference from other station users as in the case of using multicarrier CDMA in the uplink, this interference can be removed, and the weight vector does not converge in conventional MMSE combining by the RLS method, and SINR and the like are improved. Therefore, MMSE combining is possible even with a small number of pilot symbols.
Furthermore, if the temporary determination data symbol is used, the weight vector is updated and the accuracy of the weight vector is improved.
[Brief description of the drawings]
FIG. 1 is a block diagram of a multicarrier CDMA transmitter used with the multicarrier CDMA receiver of the present invention.
FIG. 2 is a frame configuration diagram of a multicarrier CDMA system in which data symbols and pilot symbols are time-multiplexed.
FIG. 3 is a waveform diagram showing a transmission signal waveform of a multi-carrier CDMA system.
FIG. 4 is a block diagram of a multicarrier CDMA receiver according to the present invention.
FIG. 5 is a flowchart showing a process for calculating a weight vector from a pilot signal.
FIG. 6 is a schematic diagram for explaining an impulse response estimation method;
FIG. 7 is a schematic diagram illustrating a method for correcting an impulse response.
FIG. 8 is a diagram showing a BER (bit error rate) characteristic when impulse response correction is not performed when only a single station user and no other station user exist.
FIG. 9 is a diagram showing BER (bit error rate) characteristics when impulse response values are corrected when only a single-station user and no other-station user exist.
FIG. 10 is a diagram showing an average SINR (signal-to-interference and noise ratio) characteristic with respect to the number of users when another station user exists to perform multiple access.
FIG. 11 is a block diagram showing another embodiment of the multi-carrier CDMA receiver according to the present invention.
12 is a schematic diagram for explaining a first updating method of the weight vector performed using the provisional determination data symbol, corresponding to the receiver configuration shown in FIG. 11. FIG.
FIG. 13 is a schematic diagram illustrating a second updating method of weight vectors performed using provisional determination data symbols.
FIG. 14 is a schematic diagram illustrating a third updating method of the weight vector performed using the provisional determination data symbol.
FIG. 15 is a schematic diagram illustrating a fourth updating method of the weight vector performed using the provisional determination data symbol.
[Explanation of symbols]
DESCRIPTION OF SYMBOLS 1 ... Multiplexer, 2 ... QPSK modulator, 3 ... Branch circuit, 4 ₁ ~ 4 _L DESCRIPTION OF SYMBOLS ... Multiplier, 5 ... Inverse discrete Fourier transform part, 6 ... Parallel-serial converter, 7 ... Guard interval insertion part, 11 ... Demultiplexing part, 12 ... Guard interval removal part, 13 ... Channel estimation part, 14 ... Serial-parallel conversion 15, discrete Fourier transform, 16 ₁ ~ 16 _L ... multiplier, 17 ... adder, 18 ... QPSK demodulator, 19 ... discrete Fourier transform unit, 20 ... weight vector calculation unit

Claims (12)

A transmission symbol sequence symbol in which a pilot symbol is time-multiplexed with a data symbol is multiplied by the value of each chip of the spread code, a plurality of orthogonal subcarriers are modulated by each of the multiplication values, and a guard interval is inserted. A multi-carrier CDMA receiver for receiving a multi-carrier CDMA signal,
The multi-carrier CDMA signal is received, and a pilot symbol period is extracted. Based on the multi-carrier CDMA signal in the pilot symbol period and pilot symbol information of all users, the transmissions from the transmission apparatuses of all the users are received. Impulse response estimation means for outputting an impulse response estimation value of the propagation path;
A vector obtained by performing a discrete Fourier transform on the impulse response estimation value of the propagation path from the transmitting device of the desired user, and a correlation vector generating means for generating a correlation vector of the desired user according to the spreading code vector information of the desired user; ,
A vector obtained by performing a discrete Fourier transform on the estimated impulse response values of the propagation paths from the transmission devices of all users, and the impulse response estimation means outputs the estimated impulse response values of the propagation paths from the transmission devices of all users. Correlation matrix generating means for generating a correlation matrix of the received multicarrier CDMA signal according to interference noise components removed from the impulse response estimation value and spread code vector information of all users,
Weight vector generation means for generating a weight vector for the desired user based on the correlation vector and the correlation matrix;
A composite signal for multiplying a vector obtained by discrete Fourier transform of a data symbol section of the received multicarrier CDMA signal and a conjugate transpose of the weight vector to synthesize a data symbol transmitted from the transmission device of the desired user Generating means,
A multi-carrier CDMA receiver characterized by comprising: The correlation vector generation means multiplies a vector obtained by performing discrete Fourier transform on the impulse response estimated value of the propagation path from the transmission device of the desired user and the spread code vector information of the desired user for each vector element. ,
The correlation matrix generation means, for all the users, a vector obtained by multiplying a vector obtained by discrete Fourier transform of an impulse response estimation value of a propagation path from each transmission device and each spread code vector information for each vector element, and the vector A matrix obtained by multiplying a conjugate transpose of each of the users, and adding a matrix obtained by multiplying a value obtained by averaging the interference noise components for all users by a unit matrix.
The multi-carrier CDMA receiver according to claim 1. The impulse response estimation means averages and outputs the impulse response estimation values calculated for a plurality of pilot symbols.
The multi-carrier CDMA receiver according to claim 1 or 2. The impulse response estimation means outputs the impulse response estimation value based only on a path within a guard interval and exceeding a threshold set in accordance with a maximum power path.
The multicarrier CDMA receiver according to any one of claims 1 to 3, wherein Weight vector updating means for sequentially updating the weight vector using the least mean square error adaptive algorithm with the weight vector generated by the weight vector generating means as an initial value;
5. The multicarrier CDMA receiver according to claim 1, comprising: A multicarrier CDMA receiver according to any one of claims 1 to 4 and weight vector update means,
The weight vector updating means causes the multi-carrier CDMA receiver according to any one of claims 1 to 4 to output a weight vector for the desired user, and for the desired user as an initial value. The weight vector is sequentially updated by a least mean square error type adaptive algorithm for the multicarrier CDMA signal in the pilot symbol interval using a weight vector, and the weight vector is sequentially updated. The updated weight vector is given to the combined signal generating means in the multi-carrier CDMA receiver described in the above, and the data symbol of the desired user is combined.
And a multi-carrier CDMA receiver. A transmission symbol sequence symbol in which a pilot symbol is time-multiplexed with a data symbol is multiplied by the value of each chip of the spread code, a plurality of orthogonal subcarriers are modulated by each of the multiplication values, and a guard interval is inserted. A multi-carrier CDMA receiver for receiving a multi-carrier CDMA signal,
The multi-carrier CDMA signal is received, and a pilot symbol period is extracted. Based on the multi-carrier CDMA signal in the pilot symbol period and pilot symbol information of all users, the transmissions from the transmission apparatuses of all the users are received. Impulse response estimation means for outputting an impulse response estimation value of the propagation path;
A vector obtained by performing a discrete Fourier transform on the impulse response estimation value of the propagation path from the transmitting device of the desired user, and a correlation vector generating means for generating a correlation vector of the desired user according to the spreading code vector information of the desired user; ,
A vector obtained by performing discrete Fourier transform on the estimated impulse response values of the propagation paths from the transmission devices of all the users, and the impulse response estimation means outputs the estimated impulse response values of the propagation paths from the transmission devices of all the users. A correlation matrix generating means for generating a correlation matrix of the received multicarrier CDMA signal according to the interference noise component removed from the estimated impulse response value and the spread code vector information of all the users,
Weight vector generation means for generating a weight vector for the desired user based on the correlation vector and the correlation matrix;
A vector obtained by performing discrete Fourier transform on at least one symbol from the beginning of the data symbol interval of the received multicarrier CDMA signal is multiplied by a conjugate transpose of the weight vector, and transmitted from the transmitting device of the desired user. A synthesized signal generating means for synthesizing at least one data symbol;
A multi-carrier CDMA receiver characterized by comprising: A multicarrier CDMA receiver according to claim 7 and weight vector update means,
8. The weight vector updating means, for the multi-carrier CDMA receiver according to claim 7, wherein at least one desired data symbol of at least the desired user among all the users from the beginning of the data symbol period. Are combined as temporary decision data symbols, based on the pilot symbol interval and the multicarrier CDMA signal from the beginning of the data symbol interval to at least one symbol, and at least the pilot symbol information and the temporary decision data symbol of the desired user. The weight vector of at least the desired user is updated, and the updated weight vector is provided to the combined signal generating means in the multi-carrier CDMA receiver according to claim 7, and at least the desired user is provided. The day of Is intended to synthesize a symbol,
And a multi-carrier CDMA receiver. The weight vector update means repeats the update operation a plurality of times.
The multi-carrier CDMA receiver according to claim 8. The multi-carrier CDMA receiver according to claim 7, wherein each time the weight vector updating unit performs the updating operation, the number of the data symbols to be combined as the provisional determination data symbol is sequentially increased.
The multi-carrier CDMA receiver according to claim 9. A multicarrier CDMA receiver according to claim 7 and weight vector update means,
8. The weight vector updating means, with respect to the multi-carrier CDMA receiver according to claim 7, wherein each of all the users is regarded as the desired user and at least 1 from the beginning of the data symbol period of all the users. Data symbols up to symbols are combined as provisional decision data symbols, the pilot symbol information and the provisional decision data symbol information of all the users are regarded as new pilot symbol information, and the pilot symbol period and 8. The multicarrier CDMA receiver according to claim 7, wherein the weight vector is updated by the multicarrier CDMA receiver according to claim 7, considering at least one symbol from the beginning of the data symbol period as a new pilot symbol period, and the data of the desired user. Synthesize symbols,
And a multi-carrier CDMA receiver. A multicarrier CDMA receiver according to claim 7 and weight vector update means,
The weight vector updating means synthesizes at least one symbol from the beginning of the data symbol period of the desired user as a provisional decision data symbol to the multi-carrier CDMA receiver according to claim 7, A weight vector for the desired user is output, the pilot symbol information of the desired user and the information of the provisional determination data symbol are regarded as new pilot symbol information, and the weight vector for the desired user is used as an initial value, The weight vector is sequentially updated by a least mean square error type adaptive algorithm for the multi-carrier CDMA signal from the beginning of the pilot symbol period and the data symbol period to at least one symbol, and the weight vector is sequentially updated. Multi Gives the weight vector the updated synthesis signal generating means in Yariya CDMA receiver, is intended to combine the data symbols of the desired user,
And a multi-carrier CDMA receiver.

Images