JP5111699B2 - Interference wave suppressor - Google Patents

Description

【００１３】
平均二乗誤差最小の意味で、主アンテナ１００に連なる遅延器５の出力信号の周波数特性と、補助アンテナ１−ｎに連なる補償フィルタＥ_ｎ（ｚ）の出力信号の周波数特性とを一致させるように、補償フィルタＥ_ｎ（ｚ）の特性近似問題を以下のように定式化する。つまり、式（２）の誤差関数Ｅｒｒを最小にする補償フィルタのインパルス応答ｅ_ｎ（ｍ）を求める。ここで、Ｅ_ｎ（Ｆ）はＥ_ｎ（ｚ）の周波数応答、Ｗ（Ｆ）は適当な荷重関数である。ΣはＦＦＴでの周波数サンプル点における和を意味する。
【数２】

【００１４】
式（２）の最小化問題は、式（４）の最小二乗問題に帰着できる。Ｗは式（９）で定義される対角行列である。ｄｉａｇはかっこ内を要素とする対角行列を表し、式（４）を最小化するベクトルｅ_ｎは一般に多元１次連立方程式（ＷＡ）^Ｈ（ＷＡ）ｅ_ｎ＝（ＷＡ）^Ｈ（Ｗｂ_ｎ）を解けば得られる（肩文字Ｈは共役転置）。これをｎ＝１，２，…，Ｎに対して行う。Ｆ_１，Ｆ_２，…，Ｆ_ＩはＦＦＴにおける周波数サンプル点である。
【数３】

【００１５】
一方、従来技術文献２（Ｒ．Ｔ．Ｃｏｍｐｔｏｎ，Ｊｒ．“Ａｄａｐｔｉｖｅ Ａｎｔｅｎｎａｓ”Ｐｒｅｎｔｉｃｅ−Ｈａｌｌ，１９８８）に示されているように、補償フィルタを使わずに、アダプティブアレーアンテナがタップ付き遅延線回路１５−ｎを持つ構成にして（図８、図９を参照）、その係数をまとめて干渉波抑圧を行うように適応的に制御すれば、結果的にチャネル間の周波数特性の不整合を考慮せずに干渉波の抑圧ができる。図８において、１５−ｎは図９の内部構造を持つタップ付き遅延線回路である。図９において、１６は信号を１サンプル遅延させる遅延器、１７は干渉波抑圧を行うように制御する適応荷重、１８は加算器である。
【００１６】
【発明が解決しようとする課題】
従来の干渉波抑圧装置は以上のように構成されているので、信号ケーブル３１，３１−ｎの長さがほぼ等しくなければならないが、やむを得ず等しくできない場合には、せっかく補償フィルタを用いても、チャネル間の周波数特性の不整合を補償することができない課題があった。主アンテナ１００から信号入力点３２までの信号ケーブルの長さと、補助アンテナ１−ｎから信号入力点３２−ｎまでの信号ケーブルの長さが異なる場合も同様である。これは広帯域干渉波抑圧性能の劣化を引き起こす可能性がある。
【００１７】
図８，図９のような構成でタップ付き遅延線回路１５−ｎの係数を適応的に制御すれば、チャネル間の周波数特性の不整合を考慮せずに干渉波を抑圧することができるが、適応的に制御すべき荷重の数が非常に多くなる。チャネル間の周波数特性不整合は、干渉波の数、入射方向、電力が変わるなどの電波環境の変化に比べたら、それほど頻繁に時間的に変化するものではない。それにもかかわらず、電波環境の変化のたびに多くの荷重を制御するのは信号処理演算量の点で負担が大きい。また、適応アルゴリズムにＬＭＳアルゴリズム等のフィードバック形のアルゴリズムを用いた場合、一般にタップ付き遅延線回路１５−ｎの収束が遅い傾向にある。
【００１８】
この発明は上記のような課題を解決するためになされたもので、各信号ケーブルの長さに相違がある場合でも、チャネル間の周波数特性の不整合を補償することができる干渉波抑圧装置を得ることを目的とする。さらに、±０．５ｆｓ近傍で不連続な周波数特性不整合があっても良好な干渉波抑圧を得ることができる干渉波抑圧装置を得ることを目的とする。
【００１９】
【課題を解決するための手段】
この発明に係る干渉波抑圧装置は、係数設定手段が、主アンテナから第１の信号入力点に至る信号の遅延時間及び複数の補助アンテナから各々の第２の信号入力点に至る信号の遅延時間と、信号発生器から第１の信号入力点に至る信号の遅延時間及び信号発生器から各々の第２の信号入力点に至る信号の遅延時間とを考慮して、補償手段を構成するディジタルフィルタの係数を設定するものであって、信号発生器から第１の信号入力点までの第１の信号ケーブルの長さと、信号発生器から第２の信号入力点までの第２の信号ケーブルの長さとの差が実測されていない場合、第１及び第２の変換手段から出力された同相・直交信号のフーリエ変換結果を取得するとともに、第１の信号ケーブルの接続先が第１の信号入力点から第２の信号入力点に切り替えられて、第２の信号ケーブルの接続先が第２の信号入力点から第１の信号入力点に切り替えられる信号ケーブルの繋ぎ替えが実施されたのち、第１及び第２の変換手段から出力された同相・直交信号のフーリエ変換結果を取得し、信号ケーブルの繋ぎ替え前後の同相・直交信号のフーリエ変換結果を用いて、補償手段を構成するディジタルフィルタの係数を設定するようにしたものである。
【００２０】
この発明に係る干渉波抑圧装置は、係数設定手段が、複数のアンテナ素子から各々の信号入力点に至る信号の遅延時間と、信号発生器から各々の信号入力点に至る信号の遅延時間とを考慮して、補償手段を構成するディジタルフィルタの係数を設定するものであって、信号発生器から第１の信号入力点までの第１の信号ケーブルの長さと、信号発生器から第２の信号入力点までの第２の信号ケーブルの長さとの差が実測されていない場合、第１及び第２の変換手段から出力された同相・直交信号のフーリエ変換結果を取得するとともに、第１の信号ケーブルの接続先が第１の信号入力点から第２の信号入力点に切り替えられて、第２の信号ケーブルの接続先が第２の信号入力点から第１の信号入力点に切り替えられる信号ケーブルの繋ぎ替えが実施されたのち、第１及び第２の変換手段から出力された同相・直交信号のフーリエ変換結果を取得し、信号ケーブルの繋ぎ替え前後の同相・直交信号のフーリエ変換結果を用いて、補償手段を構成するディジタルフィルタの係数を設定するようにしたものである。
【００２１】
この発明に係る干渉波抑圧装置は、遅延手段の出力信号のスペクトルと補償手段の出力信号のスペクトルとの差の重み付き絶対値２乗平均値が最小になるように、固有値分解もしくは特異値分解を利用して補償手段を構成するディジタルフィルタの係数を設定するようにしたものである。
【００２２】
この発明に係る干渉波抑圧装置は、補償手段の補償帯域幅が荷重加算手段でのサンプリング周波数より狭くなるように設定し、この補償帯域幅にて補償手段を構成するディジタルフィルタの係数を設定するようにしたものである。
【００２３】
【発明の実施の形態】
以下、この発明の実施の一形態を説明する。
実施の形態１．
図１はこの発明の実施の形態１による干渉波抑圧装置を示す構成図であり、図において、１００は所定の方向（所望波方向、あるいは、信号を受信したいと思う方向）に主ビームを向けた主アンテナ、１−１，１−２，…，１−Ｎは指向性が等方性に近いＮ個の補助アンテナ、２は主アンテナ１００により受信されたＲＦ信号を増幅し、そのＲＦ信号をＩＦ信号にダウンコンバートする受信機、２−１，２−２，…，２−Ｎは補助アンテナ１−ｎ（ｎ＝１，２，…，Ｎ，断らない限り以下同様）により受信されたＲＦ信号を増幅し、そのＲＦ信号をＩＦ信号にダウンコンバートする受信機である。
【００２４】
３，３−ｎはアナログ信号をディジタル信号に変換するＡ／Ｄ変換器、４，４−ｎはディジタルＩＦ信号をディジタル同相・直交信号に変換する位相検波器である。なお、受信機２，Ａ／Ｄ変換器３及び位相検波器４から第１の変換手段が構成され、受信機２−ｎ，Ａ／Ｄ変換器３−ｎ及び位相検波器４−ｎから第２の変換手段が構成されている。
【００２５】
５は位相検波器４の出力信号を遅延させる遅延器（遅延手段）、６−ｎは自己の出力における周波数特性と、遅延器５の出力における周波数特性との不整合を補償するディジタルフィルタである補償フィルタＥ_ｎ（ｚ）（補償手段）、７−ｎは適応荷重、８は加算器である。適応荷重７−ｎと加算器８から荷重加算手段９を構成しており、適応荷重７−ｎは適応荷重制御手段２５により適当な適応アルゴリズムで制御される。１０は遅延器５の出力信号から荷重加算手段９の出力信号を差し引く減算器（減算手段）である。
【００２６】
２１は遅延器５の出力信号と各補償フィルタＥ_ｎ（ｚ）の出力信号の和との差の電力が最小になるように、各補償フィルタＥ_ｎ（ｚ）の係数を設定する補償フィルタ係数設定手段である。なお、補償フィルタ係数設定手段２１及び適応荷重制御手段２５から係数設定手段が構成されている。
なお、補償フィルタＥ_ｎ（ｚ）をＦＩＲ形とするとき、遅延器５の遅延量ｄの目安は、補償フィルタＥ_ｎ（ｚ）のインパルス応答長の半分程度である。
【００２７】
次に動作について説明する。
この実施の形態１では、補償フィルタＥ_ｎ（ｚ）の係数を計算する際に、信号発生器３０から信号ケーブルを介してＲＦ信号を入力するのではなく、干渉波信号を利用する。よって、図７の信号発生器３０，信号ケーブル３１，３１−ｎ，信号入力点３２，３２−ｎは図１にはない。
【００２８】
サイドローブキャンセラ構成の干渉波抑圧装置では、補償フィルタＥ_ｎ（ｚ）の係数計算に受信される広帯域干渉波を利用する。その手順について説明する。
補償フィルタＥ_ｎ（ｚ）の係数が決まっていない場合、まず、適応荷重７−ｎの値をすべて同じ値、例えば１に固定する。その状態でサイドローブキャンセラの出力信号の電力をなるべく小さくするように、後述する方法で補償フィルタＥ_ｎ（ｚ）の係数を計算する。この結果、この時点では干渉波が抑圧される。
【００２９】
チャネル間（主アンテナ１００と各補助アンテナ１−ｎ間）の周波数特性の不整合は、干渉波の入射方向等には関係しない。そこで、一旦、補償フィルタＥ_ｎ（ｚ）の係数を計算した後は、その係数を固定しておく。チャネル間の周波数特性不整合がなければ干渉波抑圧ができるから、補償フィルタＥ_ｎ（ｚ）の係数を計算した後は、干渉波の入射方向等が変わっても、今度は適応荷重７−ｎだけを制御することによって干渉波抑圧を行える。そうすれば、チャネル間の周波数特性の不整合があっても、図８のような適応荷重を持つタップ付き遅延線回路構成にせずに適応荷重７−ｎだけの制御で、比較的少ない信号処理演算量で干渉波抑圧性能を保持することが可能となる。補償フィルタＥ_ｎ（ｚ）の係数計算は、適応荷重７−ｎの計算ほど頻繁に行う必要はない。ただし、補償フィルタＥ_ｎ（ｚ）の係数を計算する際の干渉波は広帯域である必要がある。
【００３０】
次に、干渉波を利用した補償フィルタＥ_ｎ（ｚ）の係数計算、即ち、補償フィルタ係数設定手段２１の動作について説明する。
補償フィルタＥ_ｎ（ｚ）の係数計算時には、適応荷重７−ｎの値をすべて同じ値に固定する。例えば、その値を１とする。このとき、サイドローブキャンセラの出力信号の電力の最小化は、遅延器５の出力信号と、各補償フィルタＥ_ｎ（ｚ）の出力信号の和との差信号の電力最小化と等価である。
【００３１】
図１において、位相検波器４の出力信号をｘ_０（ｋ）、位相検波器４−ｎの出力信号をｘ_ｎ（ｋ）、遅延器５の出力信号をｙ_０（ｋ），補償フィルタＥ_ｎ（ｚ）の出力信号をｙ_ｎ（ｋ）とする。これらは、同相成分を実部、直交成分を虚部とする複素信号として扱う。ｋはサンプリング間隔を単位とする時刻である。このとき、式（１０）と式（１１）の関係が成立する。式（１２）のように遅延器５の出力信号ｙ_０（ｋ）と各補償フィルタＥ_ｎ（ｚ）の出力信号ｙ_ｎ（ｋ）の和との差信号ｚ（ｋ）を定義し、ｚ（ｋ）の電力式（１３）（Ｅ[・]は統計平均）を最小化する補償フィルタＥ_ｎ（ｚ）の係数ｅ_ｎ（ｍ）（ｍ＝０，１，…，Ｍ−１）を決める。式（１３）はさらに式（１４）のように変形できる。肩文字＊は複素共役を表す。
【数４】

【００３２】
式（１４）の評価関数Ｊを最小とするｅ_ｎ（ｍ）を求めるために、Ｊをｅ_ｎ（ｍ）に関して偏微分する。式（１４）のｅ_ｎ（ｍ）に関する偏微分として式（１５）を得る。ｎ＝１，２，…，Ｎ、ｍ＝０，１，…，Ｍ−１について、式（１５）を０にするようなｅ_ｎ（ｍ）が、求める補償フィルタの係数である。よって、式（１６）の正規方程式、つまりｅ_ｎ（ｍ）（ｎ＝１，２，…，Ｎ；ｍ＝０，１，…，Ｍ−１）を未知数とする連立方程式を解けば、ｅ_ｎ（ｍ）が求められる。式（１６）の行列Ｒ、ベクトルｅ、ｐは式（１７）〜（２０）のようになる。実際には、式（１９），（２０）で統計平均Ｅ[・]は単なる時間平均で代用してよい。
【数５】

【００３３】
上記のようにして補償フィルタＥ_ｎ（ｚ）の係数を計算すると、今度はその係数を固定し、適応荷重制御手段２５が減算器１０の出力信号の電力が最小になるように、適応荷重７−ｎの荷重を決定する。
図１では、適応アルゴリズムとしてフィードフォワード形のＳＭＩ法を想定しているため、遅延器５の出力信号を適応荷重制御に用いる。フィードバック形のＬＭＳアルゴリズムなどでは、遅延器５の出力信号の代わりに減算器１０の出力信号を用いる。
図１ではＩＦ信号をＡ／Ｄ変換・サンプリング後、ディジタル信号処理によりディジタル同相・直交信号を得る構成としたが、ＩＦ信号をアナログ信号処理によりアナログ同相・直交信号に変換し、その後Ａ／Ｄ変換・サンプリングしてディジタル同相・直交信号を得る構成としてもよい。他の実施の形態でも同様である。
【００３４】
以上で明らかなように、この実施の形態１によれば、従来技術のようにタップ付き遅延線回路１５−ｎの係数を電波環境の変化のたびに適応的に制御する必要がなく、干渉波抑圧処理とは独立して、干渉波を利用してチャネル間の周波数特性の不整合を補償することが可能となる。また、チャネル間の周波数特性の不整合を補償するために信号を特定箇所にケーブルで入力する構成ではないため、信号入力点までの信号遅延の相違を考えずに済むという利点がある。
【００３５】
実施の形態２．
図２はこの発明の実施の形態２による干渉波抑圧装置を示す構成図であり、図において、図１と同一符号は同一または相当部分を示すので説明を省略する。
３５はＲＦ信号を主アンテナ１００及び補助アンテナ１−ｎに電波によって送信するための外部アンテナ（信号送信手段）、２２は遅延器５の出力信号と補償フィルタＥ_ｎ（ｚ）の出力信号との差信号の電力が最小になるように、補償フィルタＥ_ｎ（ｚ）の係数を設定する補償フィルタ係数設定手段（係数設定手段）であり、補償フィルタ係数設定手段２２は、上記実施の形態１における補償フィルタ係数設定手段２１と異なり、各補償フィルタについて、独立に係数計算を行う。
【００３６】
次に動作について説明する。
補償フィルタＥ_ｎ（ｚ）の係数計算は干渉波抑圧処理前に行っておく。一度計算すれば、干渉波抑圧のための適応荷重７−ｎほど頻繁に計算する必要はない。外部アンテナ３５から送信するＲＦ信号の帯域幅は、従来技術と同様である。
【００３７】
次に、補償フィルタＥ_ｎ（ｚ）の係数計算、即ち、補償フィルタ係数設定手段２２の動作について説明する。
補償フィルタＥ_ｎ（ｚ）の係数ｅ_ｎ（ｍ）（ｍ＝０，１，…，Ｍ−１）は、遅延器５の出力信号ｙ_０（ｋ）と、補償フィルタＥ_ｎ（ｚ）の出力信号ｙ_ｎ（ｋ）の差信号ｚ_ｎ（ｋ）の電力Ｅ[｜ｚ_ｎ（ｋ）｜^２]がなるべく小さくなるように、各ｎ＝１，２，…，Ｎごとに決める。式（２２）のように、Ｅ[｜ｚ_ｎ（ｋ）｜２]を評価関数Ｊ_ｎとおく。式（１０），（１１）から、式（２２）はさらに式（２３）のように変形できる。
【数６】

【００３８】
評価関数Ｊ_ｎを最小とするｅ_ｎ（ｍ）を求めるために、Ｊ_ｎのｅ_ｎ（ｍ）に関する偏微分を求める。それは式（２４）のようになる。ｍ＝０，１，…，Ｍ−１について式（２４）を０にするようなｅ_ｎ（ｍ）が、求める補償フィルタの係数である。よって、式（２５）の正規方程式、つまりｅ_ｎ（ｍ）（ｍ＝０，１，…，Ｍ−１）を未知数とする連立方程式を解けば、ｅ_ｎ（ｍ）が求められる。実際には、式（２５）で統計平均Ｅ[・]は単なる時間平均で代用してよい。また、式（２５）左辺の正方行列対角項はｘ_ｎ（ｋ）の電力で同じ値をとる。以上の連立方程式を解く操作をｎ＝１，２，…，Ｎに対して行う。
【数７】

【００３９】
以上説明した方法は、差信号電力を最小化するという時間域での計算といえる。従来技術（補償フィルタ係数設定手段２３）と同様に、周波数領域での誤差、つまり複素スペクトルの誤差をなるべく小さくするように、周波数領域で補償フィルタＥ_ｎ（ｚ）の係数を計算することもできる。補償帯域幅は荷重加算手段９でのサンプリング周波数より狭くせざるを得ないので、以下に説明するように特異値分解を利用して計算することが望ましい。これは、式（７）の行列Ａは事実上フルランクでなくなることが多いためである。そこで、連立方程式を解くときの数値的不安定性を防ぐために特異値分解を利用する。
【００４０】
補償フィルタＥ_ｎ（ｚ）による補償帯域幅を荷重加算手段９でのサンプリング周波数（ｆ_ｓとする）より狭くせざるを得ない理由について説明する。
チャネル間の周波数特性不整合は主にアナログフィルタの周波数特性のばらつきに起因する。ＩＦ帯では、ＩＦ周波数をｆ_ＩＦとすれば、−０．５ｆ_ｓ＋ｆ_ＩＦから０．５ｆ_ｓ＋ｆ_ＩＦの周波数帯域の特性不整合が問題になる。ＩＦ帯のアナログフィルタの周波数特性は、一般に周波数−０．５ｆ_ｓ＋ｆ_ＩＦと周波数０．５ｆ_ｓ＋ｆ_ＩＦにおける周波数特性は不連続である。これは、ベースバンドでは、補償フィルタで近似すべき特性が、２つの周波数−０．５ｆ_ｓと０．５ｆ_ｓで不連続であることを意味する。ところが、サンプリング周波数ｆ_ｓのディジタルフィルタはこの２つの周波数での周波数特性が連続である。従って、ディジタルフィルタである補償フィルタで荷重加算手段９における全帯域（−０．５ｆ_ｓから０．５ｆ_ｓ）にわたってチャネル間の周波数特性不整合を補償することが困難となる。近似度を良好に保つには、補償帯域幅を荷重加算手段９でのサンプリング周波数より少し狭くして、２つの周波数−０．５ｆ_ｓと０．５ｆ_ｓ近傍では近似を避ける必要がある。
【００４１】
式（２）の評価関数Ｅｒｒは式（２６）のように設定する。式（２６）などで、Ｆはサンプリング周波数で正規化した周波数である。式（２６）で、Ｈ_ｒｅｆ（Ｆ）は式（３）と同じであり、Ｆ_ｍｉｎとＦ_ｍａｘはそれぞれ補償帯域の下端と上端の周波数（絶対値は０．５より小）、ΣはＦＦＴによる周波数サンプル点での和を意味する。行列Ａとベクトルｂ_ｎはそれぞれ式（７），（６）と同じ形であるが、Ｆ_１，Ｆ_２，…，Ｆ_ＩをＦ_ｍｉｎからＦ_ｍａｘの間のＦＦＴにおける周波数サンプル点とする。
【数８】

【００４２】
式（２６）の最小化問題は、式（２７）の最小二乗問題に帰着できる。式（２７）を最小化するベクトルｅ_ｎは、一般に多元１次連立方程式Ａ^ＨＡｅ_ｎ＝Ａ^Ｈｂ_ｎを解けば得られるが、実際には、行列Ａ^ＨＡのランクは事実上Ｍより小さくなることが多い。つまり、行列Ａの特異値には大きいものと無視できる小さいものが現れることが多い。そうなると数値的不安定性により連立方程式をうまく解くことができなくなる。このとき、仮に補償フィルタＥ_ｎ（ｚ）の係数が得られても、その絶対値は非常に大きな値となってしまうことがあり、事実上実装不可能になってしまう。そのようなときは特異値分解を利用するとうまく最小２乗問題を解くことができ、補償フィルタＥ_ｎ（ｚ）の係数絶対値が非常に大きな値となることを防げる。
【００４３】
行列Ａを式（２９）のように特異値分解する。
Ａ＝ＵＳＶ^Ｈ （２９）
Ｓ＝ｄｉａｇ[σ_１，σ_２，…，σ_Ｍ] （３０）
ここで、Ｕは列が正規直交しているＩ×Ｍ行列、ＳはＭ×Ｍ対角行列、ＶはＭ×Ｍ直交行列である。Ｓの対角要素σ_ｍ（ｍ＝１，２，…，Ｍ）は特異値であり、０でない特異値の数がＡのランクである。実際には雑音があるので０の特異値は存在しない。しかし、特異値の大きさにかなりの差を生ずることがある。そこで、無視し得る小さい値を除いた特異値が寄与する分のみを考慮する。
【００４４】
次に、ベクトルβ_ｎを式（３１）のように定義する。そして、式（３２）のベクトルγ_ｎを計算する。ここで、[・]_ｍはそのベクトルの第ｍ要素を表し、補償フィルタＥ_ｎ（ｚ）のインパルス応答は式（３３）のようになる。以上の操作をｎ＝１，２，…，Ｎについて行う。
【数９】

【００４５】
ここでは、補償フィルタＥ_ｎ（ｚ）の係数を計算するための最小２乗問題を特異値分解を用いて解いたが、従来技術文献３（鷹尾、内田、“ビームスペース・パーシャリーアダプティブアレイに関する検討”電子情報通信学会技術研究報告Ａ・Ｐ８８−５２，１９８８）に示されているように、行列ＡＨＡの固有値分解を利用して解いてもよい。
【００４６】
以上で明らかなように、この実施の形態２では、チャネル間の周波数特性の不整合補償のために外部から電波を送信する構成として、信号をケーブルで特定箇所に入力する構成としないため、信号入力点までの信号遅延の相違を考えずに済むという利点がある。
また、補償フィルタＥ_ｎ（ｚ）の係数を周波数領域で行う場合、計算に特異値分解法や固有値分解法を利用すれば、数値的不安定性を生じることなく係数計算が可能となる。そして、補償フィルタＥ_ｎ（ｚ）の係数絶対値が非常に大きな値となることを防ぐことが可能となる。
【００４７】
実施の形態３．
図７で示す従来技術のように、補償フィルタＥ_ｎ（ｚ）の係数を計算するために外部からＲＦ信号をケーブルを介して入力する場合、主アンテナ１００から信号入力点３２（第１の信号入力点）までの信号遅延時間と、補助アンテナ１−ｎから信号入力点３２−ｎ（第２の信号入力点）までの信号遅延時間がほぼ等しいことが前提となる。これは主アンテナ１００と補助アンテナ１−ｎから信号入力点までのケーブルの長さがほぼ等しいということである。また、信号ケーブル３１，３１−ｎの長さがほぼ等しいことも前提となる。実際には、実装上そのような条件が満足されない場合がある。その場合、補償フィルタでチャネル間の周波数特性不整合を補償したとしても、上記のような信号遅延差のために干渉波抑圧性能が向上しない可能性がある。
【００４８】
もし主アンテナ１００から信号入力点３２までのケーブルの長さと、補助アンテナ１−ｎから信号入力点３２−ｎまでのケーブルの長さの差、および信号発生器３０から信号入力点３２までのケーブルの長さと、信号発生器３０から信号入力点３２−ｎまでのケーブルの長さの差が計測でき、それらに対応する信号遅延時間差がわかる場合、それらを考慮して補償フィルタＥ_ｎ（ｚ）の係数を計算することができる。
【００４９】
主アンテナ１００から信号入力点３２までの信号遅延時間に対する補助アンテナ１−ｎから信号入力点３２−ｎまでの信号遅延時間の差をΔ_ｎ（時間の次元）、信号発生器３０から信号入力点３２までの信号遅延時間に対する信号発生器３０から信号入力点３２−ｎまでの信号遅延時間の差をδ_ｎ（時間の次元）とする。主アンテナ側に対して補助アンテナ側の方が信号の遅れがある場合、Δ_ｎあるいはδ_ｎを正の量とする。説明はおもに周波数領域で補償フィルタＥ_ｎ（ｚ）の係数計算を行う場合について行う。上の２つのケースは異なるので、それぞれ説明する。
【００５０】
［１］主アンテナ・補助アンテナから信号入力点までに遅延時間差がある場合
１つの補助アンテナ１−ｎ（ｎ＝１，２，…，Ｎ）について、周波数応答ブロック構成図を示すと図３（ａ）のようになる。ここでは簡単のため、周波数応答は等価低域系で表現する。図３において、Ｂ_０（Ｆ），Ｂ_ｎ（Ｆ）は信号入力点３２，３２−ｎから位相検波器４，４−ｎの出力までの周波数応答、Ｈ_ｒｅｆ（Ｆ）は式（３）と同じである。３６は主アンテナ側に対して補助アンテナ側の方が信号の遅れがあることを示す遅延素子である。この場合、観測される位相検波器４，４−ｎの出力信号のフーリエ変換Ｃ０（Ｆ），Ｃ_ｎ（Ｆ）とＢ_０（Ｆ），Ｂ_ｎ（Ｆ）はそれぞれ等しい。
【００５１】
補償フィルタＥ_ｎ（ｚ）の係数を計算した場合、得られる補償フィルタの周波数応答Ｅ_ｎ（Ｆ）は、理想的には式（３４）のようになる。しかし、補助アンテナ側には補助アンテナ１−ｎから信号入力点３２−ｎまでにΔ_ｎの遅延があるから、補償フィルタが本来持つべき周波数応答Ｅ_ｎ’（Ｆ）は式（３５）のようになる。式（３４）と式（３５）を比べると、式（３４）には主アンテナ系の信号の進みが足りないことになる。これを補償フィルタ係数計算に反映させるには、式（３）のＨ_ｒｅｆ（Ｆ）の代わりに、信号の進みを加えた式（３６）のＨ_ｒｅｆ’（Ｆ）を用いる。
【数１０】

【００５２】
［２］信号発生器から信号入力点までに遅延時間差がある場合
１つの補助アンテナ番号１−ｎについて、周波数応答ブロック構成図を示すと図３（ｂ）のようになる。３７は信号発生器３０から信号入力点３２，３２−ｎまでに主アンテナ側に対して補助アンテナ側の方が信号の遅れがあることを示す遅延素子である。この場合、補助アンテナ側において、観測される位相検波器４−ｎの出力信号のフーリエ変換Ｃ_ｎ（Ｆ）と信号入力点３２−ｎから位相検波器４−ｎの出力までの周波数応答Ｂ_ｎ（Ｆ）は異なる。つまり、式（３７）のようになる。Ｕ（Ｆ）は信号発生器３０で発生する信号のフーリエ変換である。
【数１１】

【００５３】
補償フィルタＥ_ｎ（ｚ）の係数を計算した場合、得られる補償フィルタの周波数応答Ｅ_ｎ（Ｆ）は、理想的には式（３８）のようになる。一方、実際に信号を受信したときの信号経路には遅延がないから、補償フィルタが本来持つべき周波数応答Ｅ_ｎ”（Ｆ）は式（３９）のようになる。式（３８）と式（３９）を比べると、式（３８）は主アンテナ側の信号の進みが余分である。これを補償フィルタ係数計算に反映させるには、式（３）のＨ_ｒｅｆ（Ｆ）の代わりに、信号の遅れを加えた式（４０）のＨ_ｒｅｆ”（Ｆ）を用いる。
【数１２】

【００５４】
［３］主アンテナ・補助アンテナから信号入力点までと、信号発生器から信号入力点までの両方に遅延時間差がある場合
主アンテナ１００と補助アンテナ１−ｎから信号入力点３２，３２−ｎまでの主アンテナ系と補助アンテナ系との間の信号遅延時間差、および信号発生器３０から主アンテナ側と補助アンテナ側の信号入力点３２，３２−ｎまでの信号遅延時間差の両方がある場合、以上の議論から、式（３）のＨ_ｒｅｆ（Ｆ）の代わりに式（４１）のＨ_ｒｅｆ’’’（Ｆ）を用いる。
【数１３】

【００５５】
［４］時間域で補償フィルタの係数を計算する場合
式（３７），（４０），（４１）は、遅延器５の遅延時間ｄサンプルをそれぞれｄ−ｆ_ｓΔ_ｎ，ｄ＋ｆ_ｓδ_ｎ，ｄ＋ｆ_ｓ（δ_ｎ−Δ_ｎ）サンプルの遅延に置き換えることに等しい。これらは一般に非整数であるので、遅延器５の代わりに、上記遅延時間相当の遅延を近似的に持つディジタルフィルタを用いればよい。あるいは、これらの遅延をＦＦＴと逆ＦＦＴを介して周波数領域で与えてもよい。実際に干渉波抑圧を行うときには、元の遅延器５に戻しておく。
【００５６】
以上で明らかなように、この実施の形態３によれば、補償フィルタＥ_ｎ（ｚ）の係数計算時にＲＦ信号をケーブルを介して入力する場合、主アンテナ１００と補助アンテナ１−ｎから信号入力点３２，３２−ｎ、および信号発生器３０から信号入力点３２，３２−ｎまでの信号の遅延時間差を考慮してチャネル間の周波数特性不整合を補償できるため、良好な広帯域干渉波抑圧性能を得ることができる。
【００５７】
実施の形態４．
上記実施の形態３では、図７において信号発生器３０から信号入力点３２，３２−ｎまでのケーブルの長さの差が計測でき、それらに対応する信号遅延時間差がわかるものとしたが、信号発生器３０から信号入力点３２，３２−ｎまでの信号ケーブル３１（第１の信号ケーブル），３１−ｎ（第２の信号ケーブル）の長さの差を実測できなくても、それらの長さの差の影響を排除してチャネル間の周波数特性不整合を補償することができる。
【００５８】
ここでは、主アンテナ１００から信号入力点３２までのケーブルの長さと、補助アンテナ１−ｎから信号入力点３２−ｎまでの信号ケーブルの長さがほぼ等しいか、あるいは、それが計測でき、対応する信号遅延時間差がわかるものとする。以下、主アンテナ１００から信号入力点３２までの信号ケーブルの長さと補助アンテナ１−ｎから信号入力点３２−ｎまでのケーブルの長さの差がほぼ等しいとして説明する。等しくないが長さの差が計測できて対応する信号遅延時間差がわかる場合は、これから説明する方法に加えて、上記実施の形態３と同様な操作を行えばよい。また、補償フィルタＥ_ｎ（ｚ）の係数計算は周波数領域で行うものとする。信号等は等価低域系で表現する。また、ｎ＝１，２，…，Ｎである。
【００５９】
主アンテナ側の信号入力点３２から位相検波器４の出力までの周波数特性をＭ_０（Ｆ）、補助アンテナ側の信号入力点３２−ｎから位相検波器４−ｎの出力までの周波数特性をＭ_ｎ（Ｆ）とする。また、信号発生器３０から信号入力点３２までの信号ケーブル３１の周波数特性をＡ_０（Ｆ）、信号発生器３０から信号入力点３２−ｎまでの信号ケーブル３１−ｎの周波数特性をＡ_ｎ（Ｆ）とする。これをブロック構成図で示すと図４（ａ）のようになる。
【００６０】
信号発生器３０から信号ｕ（ｋ）を入力する。主アンテナ側の位相検波器４の出力信号のフーリエ変換をＤ_０（Ｆ）、補助アンテナ側の位相検波器４−ｎの出力信号のフーリエ変換をＤ_ｎ（Ｆ）とすると、式（４２），（４３）のようになる。Ｕ（Ｆ）はｕ（ｋ）のフーリエ変換である。
Ｄ_０（Ｆ）＝Ａ_０（Ｆ）Ｍ_０（Ｆ）Ｕ（Ｆ） （４２）
Ｄ_ｎ（Ｆ）＝Ａ_ｎ（Ｆ）Ｍ_ｎ（Ｆ）Ｕ（Ｆ） （４３）
【００６１】
次に、図５のように、信号ケーブル３１を信号入力点３２−ｎに接続し、信号ケーブル３１−ｎを信号入力点３２に接続する。つまり。信号ケーブルのつなぎ替えを行って再び信号を入力する。信号発生器３０から信号ｕ（ｋ）を入力したときの主アンテナ側の位相検波器４の出力信号のフーリエ変換をＤ_０’（Ｆ）、補助アンテナ側の位相検波器４−ｎの出力信号のフーリエ変換をＤ_ｎ’（Ｆ）とすると、式（４４），（４５）のようになる。ブロック構成図で示すと図４（ｂ）のようになる。なお、信号発生器３０から入力する信号は図４（ａ）と同じでなくてよい。式（４２），（４３）から式（４６）を得て、式（４４），（４５）から式（４７）を得る。なお、図４と図５では、補助アンテナは番号ｎのみを抜き出している。
【数１４】

【００６２】
チャネル間の周波数特性不整合を補償する場合、補償フィルタ係数計算には位相検波器の出力信号のフーリエ変換Ｄ_０（Ｆ）とＤ_ｎ（Ｆ）をそれぞれ式（２６）のＣ_０（Ｆ）とＣ_ｎ（Ｆ）として用いる。しかし、それはＡ_０（Ｆ）＝Ａ_ｎ（Ｆ）が前提である。ここでは、Ａ_０（Ｆ）≠Ａ_ｎ（Ｆ）であり、このままではＤ_０（Ｆ）とＤ_ｎ（Ｆ）を使えない。これは次のように解決できる。式（４６），（４７）から式（４８）を得る。式（４８）左辺はすべて得られる量である。式（４８）左辺をＱ（Ｆ）とおく（式（４９））と、式（４６）から式（５０）を得る。
【数１５】

【００６３】
式（２６）のＣ_０（Ｆ）／Ｃ_ｎ（Ｆ）は、Ａ_０（Ｆ）＝Ａ_ｎ（Ｆ）のときのＭ０（Ｆ）／Ｍ_ｎ（Ｆ）に等しい。そこで、式（２６）のＣ_０（Ｆ）／Ｃ_ｎ（Ｆ）の代わりに、すべて計測できる量である式（５０）左辺を用いれば、信号発生器３０から信号入力点３２，３２−ｎまでの信号ケーブル３１，３１−ｎの長さの差が実測できなくても、それらの長さの差の影響を排除してチャネル間の周波数特性不整合を補償する補償フィルタの係数を計算することができる。
【００６４】
以上で明らかなように、この実施の形態４によれば、補償フィルタＥ_ｎ（ｚ）の係数計算時にＲＦ信号をケーブルを介して入力する場合、信号発生器３０から信号入力点３２，３２−ｎまでの信号ケーブル３１，３１−ｎの長さの差が実測できなくても、これらの信号ケーブルの繋ぎ替えを行う前と後の主アンテナ側の位相検波器４の出力信号のフーリエ変換と補助アンテナ側の位相検波器４−ｎの出力信号のフーリエ変換から、信号ケーブルの長さの差の影響を排除してチャネル間の周波数特性不整合を補償できる補償フィルタ係数を計算することができる効果を奏する。
【００６５】
実施の形態５．
上記実施の形態２〜４を各アンテナ素子が対等なアダプティブアレーアンテナに適用することが可能である。図６はこの発明の実施の形態５による干渉波抑圧装置を示す構成図であり、図において、図１と同一符号は同一または相当部分を示すので説明を省略する。
５０−１，５０−２，…，５０−ＮはＮ個のアンテナ素子であり、アレーアンテナを構成する。各アンテナ素子５０−ｎには受信機２−ｎ，Ａ／Ｄ変換器３−ｎ，位相検波器４−ｎが順に接続されている。ここでは、チャネル間の周波数特性の不整合は、アンテナ素子５０−１に連なるチャネル番号１を基準とし、他チャネルの周波数特性をチャネル番号１に合わせることにする。従って、遅延器５は位相検波器４−１に接続され、補償フィルタＥ_ｎ（ｚ）は位相検波器４−２，４−３，…，４−ｎに接続されている。
【００６６】
遅延器５および補償フィルタＥ_ｎ（ｚ）（ｎ≠１）の出力には、適応荷重制御手段２５によって適当な適応アルゴリズムで制御される適応荷重７−１，７−２，…，７−Ｎが接続されている。各適応荷重７−ｎの出力信号は加算器８で加算され、荷重加算手段９の出力信号は干渉波が抑圧された信号となる。適応荷重制御手段２５は、例えば、荷重加算手段９の出力信号と参照信号との差信号電力がなるべく小さくなるように適応荷重７−ｎを制御する。
【００６７】
アンテナ素子５０−１に連なる位相検波器４−１の出力における周波数特性と、他のアンテナ素子５０−２，…，５０−Ｎに連なる位相検波器４−２，…，４−Ｎの出力の周波数特性がなるべく一致するように、あるいは、遅延器５の出力信号と各補償フィルタＥ_ｎ（ｚ）（ｎ＝２，…，Ｎ）の出力信号との差の電力がなるべく小さくなるように、補償フィルタＥ_ｎ（ｚ）で補償する。その方法は今まで説明してきた実施の形態２〜４と同様である。図６では外部に設けたアンテナ３５からＲＦ信号を電波により送信するようにしている。
【００６８】
上記実施の形態２で説明したのと同様に、チャネル間の周波数特性の不整合補償のために外部から電波を送信すれば、信号をケーブルで特定箇所に入力しないため、信号入力点までの信号遅延の相違を考えずに済むという利点がある。また、補償フィルタＥ_ｎ（ｚ）の係数計算に特異値分解法や固有値分解法を利用すれば、数値的不安定性を生じることなく係数計算が可能となる。そして、補償フィルタＥ_ｎ（ｚ）の係数絶対値が非常に大きな値となることを防ぐことが可能となる。
【００６９】
また、上記実施の形態３で説明したのと同様に、補償フィルタＥ_ｎ（ｚ）の係数計算時にＲＦ信号をケーブルを介して入力する場合、アンテナ素子５０−ｎから信号入力点、および信号発生器３０から信号入力点までの信号の遅延時間差を計測できれば、それらを考慮してチャネル間の周波数特性不整合を補償できるため、良好な広帯域干渉波抑圧性能を得ることができる。
【００７０】
さらに、上記実施の形態４で説明したのと同様に、補償フィルタＥ_ｎ（ｚ）の係数計算時にＲＦ信号をケーブルを介して入力する場合、信号発生器３０から複数の信号入力点までの信号ケーブルの長さの差が実測できなくても、これらの信号ケーブルの繋ぎ替えを行う前と後の位相検波器の出力信号のフーリエ変換から、信号ケーブルの長さの差の影響を排除してチャネル間の周波数特性不整合を補償できる補償フィルタ係数を計算することができる。
【００７１】
【発明の効果】
以上のように、この発明によれば、係数設定手段が、主アンテナから第１の信号入力点に至る信号の遅延時間及び複数の補助アンテナから各々の第２の信号入力点に至る信号の遅延時間と、信号発生器から第１の信号入力点に至る信号の遅延時間及び信号発生器から各々の第２の信号入力点に至る信号の遅延時間とを考慮して、補償手段を構成するディジタルフィルタの係数を設定するものであって、信号発生器から第１の信号入力点までの第１の信号ケーブルの長さと、信号発生器から第２の信号入力点までの第２の信号ケーブルの長さとの差が実測されていない場合、第１及び第２の変換手段から出力された同相・直交信号のフーリエ変換結果を取得するとともに、第１の信号ケーブルの接続先が第１の信号入力点から第２の信号入力点に切り替えられて、第２の信号ケーブルの接続先が第２の信号入力点から第１の信号入力点に切り替えられる信号ケーブルの繋ぎ替えが実施されたのち、第１及び第２の変換手段から出力された同相・直交信号のフーリエ変換結果を取得し、信号ケーブルの繋ぎ替え前後の同相・直交信号のフーリエ変換結果を用いて、補償手段を構成するディジタルフィルタの係数を設定するように構成したので、主アンテナ及び複数の補助アンテナから各信号入力点までの信号ケーブルの長さや、信号発生器から各信号入力点までの信号ケーブルの長さに相違があり、信号発生器から各信号入力点までの信号ケーブルの長さを実測できなくても、チャネル間の周波数特性の不整合を補償することができる効果がある。
【００７２】
この発明によれば、係数設定手段が、複数のアンテナ素子から各々の信号入力点に至る信号の遅延時間と、信号発生器から各々の信号入力点に至る信号の遅延時間とを考慮して、補償手段を構成するディジタルフィルタの係数を設定するものであって、信号発生器から第１の信号入力点までの第１の信号ケーブルの長さと、信号発生器から第２の信号入力点までの第２の信号ケーブルの長さとの差が実測されていない場合、第１及び第２の変換手段から出力された同相・直交信号のフーリエ変換結果を取得するとともに、第１の信号ケーブルの接続先が第１の信号入力点から第２の信号入力点に切り替えられて、第２の信号ケーブルの接続先が第２の信号入力点から第１の信号入力点に切り替えられる信号ケーブルの繋ぎ替えが実施されたのち、第１及び第２の変換手段から出力された同相・直交信号のフーリエ変換結果を取得し、信号ケーブルの繋ぎ替え前後の同相・直交信号のフーリエ変換結果を用いて、補償手段を構成するディジタルフィルタの係数を設定するように構成したので、複数のアンテナ素子から各信号入力点までの信号ケーブルの長さや、信号発生器から各信号入力点までの信号ケーブルの長さに相違があり、信号発生器から各信号入力点までの信号ケーブルの長さを実測できなくても、チャネル間の周波数特性の不整合を補償することができる効果がある。
【００７３】
この発明によれば、遅延手段の出力信号のスペクトルと補償手段の出力信号のスペクトルとの差の重み付き絶対値２乗平均値が最小になるように、固有値分解もしくは特異値分解を利用して補償手段を構成するディジタルフィルタの係数を設定するように構成したので、数値的不安定性を招くことなくディジタルフィルタの係数を設定することができるとともに、その係数の絶対値が非常に大きな値になることを防ぐことができる効果がある。
【００７４】
この発明によれば、補償手段の補償帯域幅が荷重加算手段でのサンプリング周波数より狭くなるように設定し、この補償帯域幅にて補償手段を構成するディジタルフィルタの係数を設定するように構成したので、±０．５ｆ_ｓ近傍で不連続な周波数特性不整合があっても良好な干渉波抑圧を得ることができる効果がある。
【図面の簡単な説明】
【図１】 この発明の実施の形態１による干渉波抑圧装置を示す構成図である。
【図２】 この発明の実施の形態２による干渉波抑圧装置を示す構成図である。
【図３】 遅延時間差がある場合の周波数応答を示すブロック構成図である。
【図４】 補償フィルタ係数計算に関して、信号ケーブルの繋ぎ替えを行う前と後の周波数応答に関するブロック構成図である。
【図５】 補償フィルタ係数計算に関して、信号ケーブルの繋ぎ替えを行うことを模式的に表したブロック構成図である。
【図６】 この発明の実施の形態５による干渉波抑圧装置を示す構成図である。
【図７】 従来の干渉波抑圧装置を示す構成図である。
【図８】 係数を適応的に制御するタップ付き遅延線回路を用いた従来の干渉波抑圧装置を示す構成図である。
【図９】 タップ付き遅延線回路の内部を示す構成図である。
【符号の説明】
１−１，…，１−Ｎ 補助アンテナ、２ 受信機（第１の変換手段）、２−１，…，２−Ｎ 受信機（第２の変換手段）、３ Ａ／Ｄ変換機（第１の変換手段）、３−１，…，３−Ｎ Ａ／Ｄ変換機（第２の変換手段）、４ 位相検波器（第１の変換手段）、４−１，…，４−Ｎ 位相検波器（第２の変換手段）、５ 遅延器（遅延手段）、６−１，…，６−Ｎ 補償フィルタ（補償手段）、７−１，…，７−Ｎ 適応荷重、８ 加算器、９ 荷重加算手段、１０ 減算器（減算手段）、２１，２２ 補償フィルタ係数設定手段（係数設定手段）、２５ 適応荷重制御手段（係数設定手段）、３０ 信号発生器、３１，３１−１，３１−２，…，３１−Ｎ 信号ケーブル、３２，３２−１，３２−２，…，３２−Ｎ 信号入力点、３５ 外部アンテナ（信号送信手段）、３６，３７ 遅延素子、５０−１，…，５０−Ｎ アンテナ素子、１００ 主アンテナ。[0001]
BACKGROUND OF THE INVENTION
  The present invention relates to an interference wave suppressing device that suppresses interference waves.
[0002]
[Prior art]
  As is well known in the art, it is possible to suppress unintentionally received interference waves by signal processing using an adaptive array antenna. In an adaptive array antenna, an adaptive load coefficient connected to each antenna element is appropriately controlled, and an interference signal is suppressed by adding a signal weighted by the adaptive load. However, if there is a mismatch (hereinafter referred to as mismatch) of frequency characteristics (amplitude characteristics, phase characteristics, group delay characteristics) between each channel (input of adaptive load coefficient corresponding to each antenna element), it is particularly wideband interference. Wave suppression performance may be degraded.
[0003]
  The mismatch of frequency characteristics between channels is mainly caused by characteristic variations of analog filters or the like in the receiver. For example, the ripple of the amplitude characteristic and the phase characteristic of the analog filter varies depending on the filter within a generally allowable range. Here, “broadband” means a bandwidth close to the sampling frequency in the interference wave suppression signal processing by weighting of the adaptive load.
[0004]
  Weighting as shown in Prior Art Document 1 (K. Treitelbaum, “A flexible processor for a digital adaptive array radar”, IEEE AES Systems Magazine, pp. 18-22, May 1991). A digital filter can be connected to compensate for frequency characteristic mismatch between channels.
[0005]
  FIG. 7 is a block diagram showing a conventional interference wave suppression device disclosed in the above-mentioned prior art document 1. In the figure, 100 is mainly in a predetermined direction (a desired wave direction or a direction in which a signal is desired to be received). .., 1-N are N auxiliary antennas whose directivities are close to isotropic, and 2 is an RF (radio frequency) signal received by the main antenna 100. , And 2-N are auxiliary antennas 1-n (n = 1, 2,..., N), which downconvert the RF signal into IF (intermediate frequency) signals. , The same unless otherwise noted) is a receiver that amplifies the received RF signal and down-converts the RF signal into an IF signal.
[0006]
  3, 3-n is an A / D converter that converts an analog signal into a digital signal, 4, 4-n is a phase detector that converts a digital IF signal into a digital in-phase / quadrature signal, and 5 is an output of the phase detector 4 A delay device for delaying the signal, 6-n is a digital filter (hereinafter referred to as compensation filter E) that compensates for a mismatch between the frequency characteristic at its output and the frequency characteristic at the output of the delay device 5._n(Referred to as (z)), 7-n is an adaptive load, and 8 is an adder. The load adding means 9 is composed of the adaptive load 7-n and the adder 8. The adaptive load 7-n is controlled by the adaptive load control means 25 with an appropriate adaptive algorithm. A subtracter 10 subtracts the output signal of the load adding means 9 from the output signal of the delay unit 5.
[0007]
  30 is a compensation filter E_nA signal generator for generating an RF signal to calculate the coefficient of (z), 31 and 31-n are signal cables for transmitting the RF signal generated by the signal generator 30, and 32 and 32-n are signals of the RF signal. The input point 23 is a compensation filter E from the output signal of the phase detector 4, 4-n._nCompensation filter coefficient setting means for calculating and setting the coefficient of (z).
  Here, z^-1Is a sampling interval of signals in the load adding means 9 and represents a signal delay of one sample, and the delay amount of the delay unit 5 is d samples.
[0008]
  Next, the operation will be described.
  The RF signals received by the main antenna 100 and the auxiliary antenna 1-n are amplified by the receivers 2, 2-n and down-converted to become IF signals. The A / D converters 3, 3-n convert the analog IF signals into digital IF signals, and the phase detectors 4, 4-n convert the digital IF signals into digital in-phase / quadrature signals. The signal after the phase detectors 4, 4-n is treated as a complex signal in which the real part is an in-phase component and the imaginary part is a quadrature component. The delay unit 5 delays the output signal of the phase detector 4 by d samples. This is the compensation filter E_nThis is because the insertion of (z) causes some delay, so that the delay is matched.
[0009]
  Compensation filter E_nThe output signal of (z) is weighted by the adaptive load 7-n, and the adder 8 adds the output signal of the adaptive load 7-n. Then, the subtracter 10 subtracts the output signal of the adder 8 from the output signal of the delay unit 5 to obtain a reception signal in which the interference wave is suppressed.
  In FIG. 7, a feedforward type SMI (sample matrix inversion) method is assumed as an adaptive algorithm, and therefore the output signal of the delay unit 5 is used for adaptive load control. In the feedback type LMS (least mean squares) algorithm or the like, the output signal of the subtractor 10 is used instead of the output signal of the delay unit 5.
[0010]
  Next, the compensation filter E_nThe coefficient calculation of (z) will be described. It is desirable that the signal input points 32 and 32-n be as close to the antenna as possible. Here, it is assumed that the signal delay time from the main antenna 100 to the signal input point 32 is substantially equal to the signal delay time from the auxiliary antenna 1-n to the signal input point 32-n. Further, it is assumed that the lengths of the signal cables 31 and 31-n are substantially equal.
[0011]
  The RF signal generated by the signal generator 30 is simultaneously input to the signal input points 32 and 32-n via the signal cables 31 and 31-n. The bandwidth of the RF signal is about the sampling frequency of the signal in the load adding means 9 or more. For simultaneously input signals, the spectrum of the output signal of the delay unit 5 and the compensation filter E_nThe compensation filter coefficient setting means 23 makes the compensation filter E so that the spectra of the output signals of (z) are as equal as possible._nCalculate the coefficient of (z). In Prior Art Document 1, the compensation filter E_n(Z) is the FIR type, and the coefficient is calculated so that the mean square error of the spectral difference between the two is minimized. This will be described below. Here, the spectrum is a complex number.
[0012]
  Compensation filter E_nThe transfer function of (z) is set as in equation (1). M is the compensation filter E_nIt is the impulse response length of (z).
  FFT (Fast Fourier Transform) is performed on the output signal of the phase detector 4 and the output signal of the phase detector 4-n. The FFT result for the output signal of the phase detector 4 is C₀(F), the FFT result for the output signal of the phase detector 4-n is C_n(F). F is a normalized frequency that takes discrete values at equal intervals of −0.5 or more and less than 0.5.
[Expression 1]

[0013]
  The frequency characteristic of the output signal of the delay unit 5 connected to the main antenna 100 and the compensation filter E connected to the auxiliary antenna 1-n in the meaning of the mean square error being minimum._nThe compensation filter E so as to match the frequency characteristic of the output signal of (z)._nThe characteristic approximation problem of (z) is formulated as follows. In other words, the impulse response e of the compensation filter that minimizes the error function Err in equation (2)._nFind (m). Where E_n(F) is E_nThe frequency response of (z), W (F), is a suitable load function. Σ means the sum at the frequency sampling points in the FFT.
[Expression 2]

[0014]
  The minimization problem of equation (2) can be reduced to the least square problem of equation (4). W is a diagonal matrix defined by equation (9). diag represents a diagonal matrix with elements in parentheses, and a vector e that minimizes the expression (4)_nIs generally a multidimensional system of simultaneous equations (WA)^H(WA) e_n= (WA)^H(Wb_n) Is obtained (the superscript H is a conjugate transposition). This is performed for n = 1, 2,. F₁, F₂, ..., F_IIs a frequency sampling point in FFT.
[Equation 3]

[0015]
  On the other hand, as shown in Prior Art Document 2 (RT Compton, Jr. “Adaptive Antennas”, Prentice-Hall, 1988), an adaptive array antenna has a tapped delay line circuit 15 without using a compensation filter. -N (see FIGS. 8 and 9), and adaptively controlling the coefficients together to suppress the interference wave, the result is to consider the mismatch in frequency characteristics between channels. Interference signal can be suppressed. In FIG. 8, 15-n is a tapped delay line circuit having the internal structure of FIG. In FIG. 9, 16 is a delay unit that delays a signal by one sample, 17 is an adaptive weight that is controlled so as to suppress interference waves, and 18 is an adder.
[0016]
[Problems to be solved by the invention]
  Since the conventional interference wave suppression device is configured as described above, the lengths of the signal cables 31 and 31-n must be approximately equal. If this is unavoidable, even if a compensation filter is used, There was a problem that frequency characteristic mismatch between channels could not be compensated. The same applies to the case where the length of the signal cable from the main antenna 100 to the signal input point 32 is different from the length of the signal cable from the auxiliary antenna 1-n to the signal input point 32-n. This may cause degradation of broadband interference wave suppression performance.
[0017]
  If the coefficients of the tapped delay line circuit 15-n are adaptively controlled with the configurations shown in FIGS. 8 and 9, interference waves can be suppressed without considering the mismatch of frequency characteristics between channels. The number of loads to be adaptively controlled becomes very large. The frequency characteristic mismatch between channels does not change so frequently in time compared to changes in the radio wave environment such as the number of interference waves, the incident direction, and the power. Nevertheless, controlling a large amount of load each time the radio wave environment changes is a heavy burden in terms of the amount of signal processing computation. When a feedback type algorithm such as an LMS algorithm is used as the adaptive algorithm, the tapped delay line circuit 15-n generally tends to converge slowly.
[0018]
  The present invention has been made in order to solve the above-described problems. An interference wave suppressing device capable of compensating for mismatch in frequency characteristics between channels even when there is a difference in the length of each signal cable. The purpose is to obtain. It is another object of the present invention to provide an interference wave suppressing device that can obtain good interference wave suppression even if there is a discontinuous frequency characteristic mismatch near ± 0.5 fs.
[0019]
[Means for Solving the Problems]
  In the interference wave suppressing device according to the present invention, the coefficient setting means includes a signal delay time from the main antenna to the first signal input point and a signal delay time from the plurality of auxiliary antennas to each second signal input point. Filter that constitutes compensation means in consideration of the delay time of the signal from the signal generator to the first signal input point and the delay time of the signal from the signal generator to each second signal input point Set the coefficient ofThe difference between the length of the first signal cable from the signal generator to the first signal input point and the length of the second signal cable from the signal generator to the second signal input point is actually measured. If not, the Fourier transform result of the in-phase / quadrature signal output from the first and second conversion means is acquired, and the connection destination of the first signal cable is connected to the second signal from the first signal input point. After switching to the input point and switching the signal cable so that the connection destination of the second signal cable is switched from the second signal input point to the first signal input point, the first and second conversions are performed. The Fourier transform result of the in-phase / quadrature signal output from the means is acquired, and the coefficient of the digital filter constituting the compensation means is set using the Fourier transform result of the in-phase / quadrature signal before and after the signal cable switching. It is intended.
[0020]
  In the interference wave suppressing device according to the present invention, the coefficient setting means includes a delay time of a signal from a plurality of antenna elements to each signal input point and a delay time of a signal from the signal generator to each signal input point. Taking into account, set the coefficient of the digital filter that constitutes the compensation meansThe difference between the length of the first signal cable from the signal generator to the first signal input point and the length of the second signal cable from the signal generator to the second signal input point is actually measured. If not, the Fourier transform result of the in-phase / quadrature signal output from the first and second conversion means is acquired, and the connection destination of the first signal cable is connected to the second signal from the first signal input point. After switching to the input point and switching the signal cable so that the connection destination of the second signal cable is switched from the second signal input point to the first signal input point, the first and second conversions are performed. The Fourier transform result of the in-phase / quadrature signal output from the means is acquired, and the coefficient of the digital filter constituting the compensation means is set using the Fourier transform result of the in-phase / quadrature signal before and after the signal cable switching. It is intended.
[0021]
  The interference wave suppressing apparatus according to the present invention is characterized in that eigenvalue decomposition or singular value decomposition is performed so that a weighted absolute mean square value of a difference between the spectrum of the output signal of the delay means and the spectrum of the output signal of the compensation means is minimized. Is used to set the coefficient of the digital filter constituting the compensation means.
[0022]
  In the interference wave suppressing apparatus according to the present invention, the compensation bandwidth of the compensation means is set to be narrower than the sampling frequency of the load adding means, and the coefficient of the digital filter constituting the compensation means is set by this compensation bandwidth. It is what I did.
[0023]
DETAILED DESCRIPTION OF THE INVENTION
  An embodiment of the present invention will be described below.
Embodiment 1 FIG.
  FIG. 1 is a block diagram showing an interference wave suppressing apparatus according to Embodiment 1 of the present invention. In FIG. 1, reference numeral 100 designates a main beam in a predetermined direction (a desired wave direction or a direction in which a signal is desired to be received). Main antennas 1-1, 1-2,..., 1-N are N auxiliary antennas whose directivities are close to isotropic, and 2 is an RF signal received by the main antenna 100. , 2-N are received by auxiliary antennas 1-n (n = 1, 2,..., N, and so on unless otherwise noted). A receiver that amplifies an RF signal and down-converts the RF signal into an IF signal.
[0024]
  3, 3-n is an A / D converter that converts an analog signal into a digital signal, and 4, 4-n is a phase detector that converts a digital IF signal into a digital in-phase / quadrature signal. The receiver 2, the A / D converter 3 and the phase detector 4 constitute a first conversion means, and the receiver 2-n, the A / D converter 3-n and the phase detector 4-n Two conversion means are configured.
[0025]
  Reference numeral 5 denotes a delay device (delay means) that delays the output signal of the phase detector 4, and 6 -n denotes a digital filter that compensates for mismatch between the frequency characteristic of its own output and the frequency characteristic of the output of the delay device 5. Compensation filter E_n(Z) (compensation means), 7-n is an adaptive load, and 8 is an adder. The load adding means 9 is composed of the adaptive load 7-n and the adder 8. The adaptive load 7-n is controlled by the adaptive load control means 25 with an appropriate adaptive algorithm. Reference numeral 10 denotes a subtracter (subtracting means) for subtracting the output signal of the load adding means 9 from the output signal of the delayer 5.
[0026]
  Reference numeral 21 denotes an output signal of the delay unit 5 and each compensation filter E._nEach compensation filter E is set so that the power of the difference from the sum of the output signals of (z) is minimized._nCompensation filter coefficient setting means for setting the coefficient of (z). The compensation filter coefficient setting means 21 and the adaptive load control means 25 constitute coefficient setting means.
  The compensation filter E_nWhen (z) is the FIR type, the standard of the delay amount d of the delay device 5 is the compensation filter E_nIt is about half of the impulse response length of (z).
[0027]
  Next, the operation will be described.
  In the first embodiment, the compensation filter E_nWhen calculating the coefficient of (z), an RF signal is not input from the signal generator 30 via the signal cable, but an interference wave signal is used. Therefore, the signal generator 30, the signal cables 31, 31-n, and the signal input points 32, 32-n in FIG. 7 are not shown in FIG.
[0028]
  In the interference wave suppression device having the sidelobe canceller configuration, the compensation filter E_nThe received broadband interference wave is used for calculating the coefficient of (z). The procedure will be described.
  Compensation filter E_nIf the coefficient of (z) has not been determined, first, all the values of the adaptive load 7-n are fixed to the same value, for example, 1. In this state, the compensation filter E is applied by a method described later so as to reduce the power of the output signal of the sidelobe canceller as much as possible._nCalculate the coefficient of (z). As a result, the interference wave is suppressed at this point.
[0029]
  The mismatch in frequency characteristics between channels (between the main antenna 100 and each auxiliary antenna 1-n) is not related to the incident direction of interference waves. Therefore, once compensation filter E_nAfter calculating the coefficient of (z), the coefficient is fixed. If there is no frequency characteristic mismatch between the channels, the interference wave can be suppressed._nAfter calculating the coefficient of (z), even if the incident direction or the like of the interference wave is changed, the interference wave can be suppressed by controlling only the adaptive load 7-n. By doing so, even if there is a mismatch in frequency characteristics between channels, relatively little signal processing can be performed by controlling only the adaptive load 7-n without using the tapped delay line circuit configuration having the adaptive load as shown in FIG. Interference wave suppression performance can be maintained with the amount of computation. Compensation filter E_nThe coefficient calculation of (z) does not need to be performed as frequently as the calculation of the adaptive load 7-n. However, compensation filter E_nThe interference wave at the time of calculating the coefficient of (z) needs to be a wide band.
[0030]
  Next, a compensation filter E using an interference wave_nThe coefficient calculation of (z), that is, the operation of the compensation filter coefficient setting means 21 will be described.
  Compensation filter E_nAt the time of calculating the coefficient of (z), all values of the adaptive load 7-n are fixed to the same value. For example, the value is 1. At this time, the power of the output signal of the sidelobe canceller is minimized by the output signal of the delay unit 5 and each compensation filter E._nThis is equivalent to the power minimization of the difference signal from the sum of the output signals of (z).
[0031]
  In FIG. 1, the output signal of the phase detector 4 is x₀(K) The output signal of the phase detector 4-n is expressed as x_n(K) The output signal of the delay unit 5 is y₀(K), compensation filter E_nThe output signal of (z) is y_n(K). These are treated as complex signals having the in-phase component as the real part and the quadrature component as the imaginary part. k is a time with a sampling interval as a unit. At this time, the relationship of Formula (10) and Formula (11) is materialized. The output signal y of the delay device 5 as shown in Expression (12)₀(K) and each compensation filter E_nOutput signal y of (z)_nCompensation filter E which defines a difference signal z (k) from the sum of (k) and minimizes power equation (13) of z (k) (E [•] is a statistical average)_nCoefficient e of (z)_n(M) (m = 0, 1,..., M−1) is determined. Equation (13) can be further transformed into Equation (14). The superscript * represents a complex conjugate.
[Expression 4]

[0032]
  E which minimizes the evaluation function J of Equation (14)_nTo find (m), let J be e_nPartial differentiation is performed with respect to (m). E in equation (14)_nEquation (15) is obtained as a partial differential with respect to (m). n = 1, 2,..., N, m = 0, 1,._n(M) is a coefficient of the compensation filter to be obtained. Therefore, the normal equation of equation (16), that is, e_n(M) Solving simultaneous equations with n = 1, 2,..., N; m = 0, 1,._n(M) is required. The matrix R and the vectors e and p in Expression (16) are as shown in Expressions (17) to (20). In practice, the statistical average E [•] may be replaced with a simple time average in the equations (19) and (20).
[Equation 5]

[0033]
  Compensation filter E as described above_nWhen the coefficient of (z) is calculated, this coefficient is fixed, and the adaptive load control means 25 determines the load of the adaptive load 7-n so that the power of the output signal of the subtracter 10 is minimized.
  In FIG. 1, since the feedforward type SMI method is assumed as an adaptive algorithm, the output signal of the delay unit 5 is used for adaptive load control. In the feedback type LMS algorithm, the output signal of the subtracter 10 is used instead of the output signal of the delay unit 5.
  In FIG. 1, after the A / D conversion / sampling of the IF signal, a digital in-phase / quadrature signal is obtained by digital signal processing. However, the IF signal is converted into an analog in-phase / quadrature signal by analog signal processing and then A / D A configuration may be adopted in which digital in-phase / quadrature signals are obtained by conversion / sampling. The same applies to other embodiments.
[0034]
  As apparent from the above, according to the first embodiment, it is not necessary to adaptively control the coefficient of the tapped delay line circuit 15-n every time the radio wave environment changes as in the prior art. Independent of the suppression process, it is possible to compensate for mismatch in frequency characteristics between channels using interference waves. In addition, there is an advantage that it is not necessary to consider the difference in signal delay up to the signal input point because the signal is not input to a specific location with a cable in order to compensate for mismatch in frequency characteristics between channels.
[0035]
Embodiment 2. FIG.
  2 is a block diagram showing an interference wave suppressing apparatus according to Embodiment 2 of the present invention. In the figure, the same reference numerals as those in FIG.
  Reference numeral 35 denotes an external antenna (signal transmission means) for transmitting RF signals to the main antenna 100 and the auxiliary antenna 1-n by radio waves, and reference numeral 22 denotes an output signal of the delay device 5 and a compensation filter E._nThe compensation filter E so that the power of the difference signal from the output signal of (z) is minimized._nCompensation filter coefficient setting means (coefficient setting means) for setting the coefficient of (z). Unlike the compensation filter coefficient setting means 21 in the first embodiment, the compensation filter coefficient setting means 22 is independent for each compensation filter. Calculate the coefficient.
[0036]
  Next, the operation will be described.
  Compensation filter E_nThe coefficient calculation of (z) is performed before the interference wave suppression process. Once calculated, it is not necessary to calculate as frequently as the adaptive load 7-n for interference wave suppression. The bandwidth of the RF signal transmitted from the external antenna 35 is the same as in the prior art.
[0037]
  Next, the compensation filter E_nThe coefficient calculation of (z), that is, the operation of the compensation filter coefficient setting means 22 will be described.
  Compensation filter E_nCoefficient e of (z)_n(M) (m = 0, 1,..., M−1) is the output signal y of the delay unit 5₀(K) and compensation filter E_nOutput signal y of (z)_nDifference signal z of (k)_nElectric power E [| z of (k)_n(K) |²] Is determined for each n = 1, 2,... As shown in Equation (22), E [| z_n(K) | 2] is an evaluation function J_nfar. From equations (10) and (11), equation (22) can be further transformed into equation (23).
[Formula 6]

[0038]
  Evaluation function J_nE to minimize_nTo find (m), J_nE_nA partial derivative with respect to (m) is obtained. It becomes like Formula (24). e such that equation (24) is set to 0 for m = 0, 1,..., M−1._n(M) is a coefficient of the compensation filter to be obtained. Therefore, the normal equation of equation (25), that is, e_n(M) Solving simultaneous equations with m = 0, 1,..., M−1 as unknowns, e_n(M) is required. In practice, the statistical average E [•] in equation (25) may be replaced with a simple time average. Also, the square matrix diagonal term on the left side of equation (25) is x_nIt takes the same value with the power of (k). The operation for solving the simultaneous equations is performed for n = 1, 2,.
[Expression 7]

[0039]
  The method described above can be said to be a calculation in the time domain in which the difference signal power is minimized. Similar to the prior art (compensation filter coefficient setting means 23), the compensation filter E in the frequency domain is designed to minimize the error in the frequency domain, that is, the error in the complex spectrum._nThe coefficient of (z) can also be calculated. Since the compensation bandwidth must be narrower than the sampling frequency in the load adding means 9, it is desirable to calculate using the singular value decomposition as described below. This is because the matrix A in equation (7) is often not in full rank. Therefore, singular value decomposition is used to prevent numerical instability when solving simultaneous equations.
[0040]
  Compensation filter E_nThe compensation bandwidth according to (z) is used as the sampling frequency (f_sExplain why you have to make it narrower.
  The frequency characteristic mismatch between the channels is mainly caused by variation in the frequency characteristic of the analog filter. In the IF band, the IF frequency is f_IF-0.5f_s+ F_IFTo 0.5f_s+ F_IFThe frequency band characteristic mismatch is a problem. The frequency characteristic of an IF band analog filter generally has a frequency of −0.5 f._s+ F_IFAnd frequency 0.5f_s+ F_IFThe frequency characteristic at is discontinuous. This is because, in the baseband, the characteristics to be approximated by the compensation filter are two frequencies −0.5f._sAnd 0.5f_sMeans discontinuous. However, the sampling frequency f_sThis digital filter has continuous frequency characteristics at these two frequencies. Accordingly, the entire band (−0.5 f) in the load adding means 9 is compensated by a compensation filter which is a digital filter._sTo 0.5f_sIt is difficult to compensate for frequency characteristic mismatch between channels. In order to maintain a good degree of approximation, the compensation bandwidth is slightly narrower than the sampling frequency in the load adding means 9, and the two frequencies -0.5f are set._sAnd 0.5f_sIt is necessary to avoid approximation in the vicinity.
[0041]
  The evaluation function Err in Expression (2) is set as in Expression (26). In Expression (26) and the like, F is a frequency normalized with the sampling frequency. In equation (26), H_ref(F) is the same as equation (3), and F_minAnd F_maxMeans the frequency at the lower end and the upper end of the compensation band (absolute value is smaller than 0.5), and Σ means the sum at the frequency sampling points by FFT. Matrix A and vector b_nAre in the same form as equations (7) and (6) respectively, but F₁, F₂, ..., F_IF_minTo F_maxThe frequency sampling points in the FFT between.
[Equation 8]

[0042]
  The minimization problem of equation (26) can be reduced to the least square problem of equation (27). Vector e that minimizes equation (27)_nIs generally a multidimensional system of simultaneous equations A^HAe_n= A^Hb_nIn practice, the matrix A^HIn many cases, the rank of A is effectively smaller than M. That is, the singular values of the matrix A often appear large and negligible small ones. Then, the simultaneous equations cannot be solved well due to numerical instability. At this time, it is assumed that the compensation filter E_nEven if the coefficient of (z) is obtained, its absolute value may become a very large value, which makes it practically impossible to implement. In such a case, using the singular value decomposition, the least square problem can be solved well, and the compensation filter E_nIt is possible to prevent the coefficient absolute value of (z) from becoming a very large value.
[0043]
  The matrix A is subjected to singular value decomposition as shown in Expression (29).
      A = USV^H                                  (29)
      S = diag [σ₁, Σ₂, ..., σ_M] (30)
  Here, U is an I × M matrix whose columns are orthonormal, S is an M × M diagonal matrix, and V is an M × M orthogonal matrix. Diagonal element σ of S_m(M = 1, 2,..., M) is a singular value, and the number of non-zero singular values is the rank of A. There is actually no noise, so there is no singular value of 0. However, there can be considerable differences in the magnitude of the singular values. Therefore, only the contribution of singular values excluding small negligible values is considered.
[0044]
  Next, the vector β_nIs defined as in equation (31). And the vector γ in equation (32)_nCalculate here,[·]_mRepresents the m-th element of the vector, and the compensation filter E_nThe impulse response of (z) is as shown in Equation (33). The above operation is performed for n = 1, 2,.
[Equation 9]

[0045]
  Here, the compensation filter E_nThe least-squares problem for calculating the coefficient of (z) was solved using singular value decomposition, but the prior art document 3 (Takao, Uchida, “Study on Beamspace and Partial Adaptive Array”, IEICE Technology As shown in Research Report A · P88-52, 1988), eigenvalue decomposition of the matrix AHA may be used.
[0046]
  As is apparent from the above, in the second embodiment, the signal is not input to a specific location with a cable as a configuration for transmitting radio waves from the outside in order to compensate for mismatch in frequency characteristics between channels. There is an advantage that it is not necessary to consider the difference in signal delay to the input point.
  The compensation filter E_nWhen the coefficient (z) is performed in the frequency domain, the coefficient calculation can be performed without causing numerical instability if a singular value decomposition method or an eigenvalue decomposition method is used for the calculation. And the compensation filter E_nIt becomes possible to prevent the coefficient absolute value of (z) from becoming a very large value.
[0047]
Embodiment 3 FIG.
  As in the prior art shown in FIG._nWhen inputting an RF signal from the outside via a cable in order to calculate the coefficient of (z), a signal input point 32 from the main antenna 100 is used.(First signal input point)Signal delay time from the auxiliary antenna 1-n to the signal input point 32-n(Second signal input point)It is assumed that the signal delay time until is almost equal. This means that the cable lengths from the main antenna 100 and the auxiliary antenna 1-n to the signal input point are substantially equal. It is also assumed that the lengths of the signal cables 31 and 31-n are substantially equal. In practice, such a condition may not be satisfied in implementation. In this case, even if the frequency characteristic mismatch between the channels is compensated by the compensation filter, the interference wave suppression performance may not be improved due to the signal delay difference as described above.
[0048]
  If the length of the cable from the main antenna 100 to the signal input point 32 is different from the length of the cable from the auxiliary antenna 1-n to the signal input point 32-n, and the cable from the signal generator 30 to the signal input point 32 And the length of the cable from the signal generator 30 to the signal input point 32-n can be measured, and when the corresponding signal delay time difference is known, the compensation filter E_nThe coefficient of (z) can be calculated.
[0049]
  The difference in signal delay time from the auxiliary antenna 1-n to the signal input point 32-n with respect to the signal delay time from the main antenna 100 to the signal input point 32 is Δ_n(Time dimension), the difference of the signal delay time from the signal generator 30 to the signal input point 32-n with respect to the signal delay time from the signal generator 30 to the signal input point 32 is expressed by δ_n(Time dimension). If there is a signal delay on the auxiliary antenna side relative to the main antenna side, Δ_nOr δ_nIs a positive amount. The explanation is mainly for the compensation filter E in the frequency domain._nThis is performed for the case of calculating the coefficient (z). The above two cases are different and will be described separately.
[0050]
[1] When there is a delay time difference from the main / auxiliary antenna to the signal input point
  FIG. 3A shows a frequency response block configuration diagram for one auxiliary antenna 1-n (n = 1, 2,..., N). Here, for simplicity, the frequency response is expressed by an equivalent low-frequency system. In FIG. 3, B₀(F), B_n(F) is the frequency response from the signal input points 32, 32-n to the output of the phase detectors 4, 4-n, H_ref(F) is the same as equation (3). Reference numeral 36 denotes a delay element indicating that there is a signal delay on the auxiliary antenna side with respect to the main antenna side. In this case, the Fourier transforms C0 (F), C of the output signals of the observed phase detectors 4, 4-n_n(F) and B₀(F), B_n(F) is equal to each other.
[0051]
  Compensation filter E_nWhen the coefficient of (z) is calculated, the frequency response E of the obtained compensation filter_n(F) is ideally given by equation (34). However, on the auxiliary antenna side, Δ from the auxiliary antenna 1-n to the signal input point 32-n_nFrequency response E that the compensation filter should originally have_n′ (F) is represented by Expression (35). Comparing equation (34) with equation (35), equation (34) has insufficient signal advance of the main antenna system. In order to reflect this in the compensation filter coefficient calculation, H in Equation (3) is used._refInstead of (F), H in equation (36) with the advance of the signal added_ref'(F) is used.
[Expression 10]

[0052]
[2] When there is a delay time difference from the signal generator to the signal input point
  The frequency response block diagram for one auxiliary antenna number 1-n is as shown in FIG. Reference numeral 37 denotes a delay element indicating that there is a signal delay on the auxiliary antenna side from the main antenna side from the signal generator 30 to the signal input points 32 and 32-n. In this case, the Fourier transform C of the output signal of the phase detector 4-n observed on the auxiliary antenna side_n(F) and the frequency response B from the signal input point 32-n to the output of the phase detector 4-n_n(F) is different. That is, it becomes like a formula (37). U (F) is a Fourier transform of the signal generated by the signal generator 30.
## EQU11 ##

[0053]
  Compensation filter E_nWhen the coefficient of (z) is calculated, the frequency response E of the obtained compensation filter_n(F) is ideally given by equation (38). On the other hand, since there is no delay in the signal path when the signal is actually received, the frequency response E that the compensation filter should originally have._n"(F) becomes as shown in Equation (39). Comparing Equation (38) with Equation (39), Equation (38) has an extra signal advance on the main antenna side. In order to reflect in H_refInstead of (F), H in equation (40) with a signal delay added_ref"(F) is used.
[Expression 12]

[0054]
[3] When there is a difference in delay time from the main antenna / auxiliary antenna to the signal input point and from the signal generator to the signal input point
  Signal delay time difference between the main antenna system and the auxiliary antenna system from the main antenna 100 and the auxiliary antenna 1-n to the signal input points 32 and 32-n, and signals from the signal generator 30 on the main antenna side and the auxiliary antenna side When there is both signal delay time differences up to the input points 32 and 32-n, from the above discussion, H in Equation (3)_refH in formula (41) instead of (F)_ref"" (F) is used.
[Formula 13]

[0055]
[4] When calculating compensation filter coefficients in the time domain
  Expressions (37), (40), and (41) are expressed as df for the delay time d samples of the delay unit 5, respectively._sΔ_n, D + f_sδ_n, D + f_s(Δ_n-Δ_n) Equivalent to replacing with sample delay. Since these are generally non-integer numbers, a digital filter having an approximate delay corresponding to the delay time may be used instead of the delay unit 5. Alternatively, these delays may be given in the frequency domain via FFT and inverse FFT. When actually performing interference wave suppression, it is returned to the original delay unit 5.
[0056]
  As apparent from the above, according to the third embodiment, the compensation filter E_nWhen an RF signal is input via a cable at the time of calculating the coefficient of (z), signal input points 32 and 32-n from the main antenna 100 and the auxiliary antenna 1-n, and signal input points 32 and 32- Since the frequency characteristic mismatch between channels can be compensated in consideration of the delay time difference of signals up to n, a good broadband interference wave suppression performance can be obtained.
[0057]
Embodiment 4 FIG.
  In the third embodiment, the difference in cable length from the signal generator 30 to the signal input points 32 and 32-n in FIG. 7 can be measured and the corresponding signal delay time difference can be understood. Signal cable 31 from the generator 30 to the signal input points 32 and 32-n(First signal cable), 31-n(Second signal cable)Even if the length difference cannot be measured, it is possible to eliminate the influence of the length difference and compensate for the frequency characteristic mismatch between the channels.
[0058]
  Here, the length of the cable from the main antenna 100 to the signal input point 32 is substantially equal to the length of the signal cable from the auxiliary antenna 1-n to the signal input point 32-n, or it can be measured and handled. It is assumed that the difference in signal delay time to be understood. In the following description, it is assumed that the difference between the length of the signal cable from the main antenna 100 to the signal input point 32 and the length of the cable from the auxiliary antenna 1-n to the signal input point 32-n is substantially equal. If they are not equal but the length difference can be measured and the corresponding signal delay time difference is known, in addition to the method described below, the same operation as in the third embodiment may be performed. The compensation filter E_nThe coefficient calculation of (z) is performed in the frequency domain. Signals etc. are expressed in an equivalent low-frequency system. Further, n = 1, 2,..., N.
[0059]
  The frequency characteristic from the signal input point 32 on the main antenna side to the output of the phase detector 4 is represented by M₀(F) M is the frequency characteristic from the signal input point 32-n on the auxiliary antenna side to the output of the phase detector 4-n._n(F). The frequency characteristic of the signal cable 31 from the signal generator 30 to the signal input point 32 is represented by A₀(F) A frequency characteristic of the signal cable 31-n from the signal generator 30 to the signal input point 32-n is represented by A_n(F). This is shown in a block diagram in FIG.
[0060]
  A signal u (k) is input from the signal generator 30. D Fourier transform of the output signal of the phase detector 4 on the main antenna side₀(F) D is the Fourier transform of the output signal of the phase detector 4-n on the auxiliary antenna side._nAssuming (F), equations (42) and (43) are obtained. U (F) is the Fourier transform of u (k).
      D₀(F) = A₀(F) M₀(F) U (F) (42)
      D_n(F) = A_n(F) M_n(F) U (F) (43)
[0061]
  Next, as shown in FIG. 5, the signal cable 31 is connected to the signal input point 32-n and the signal cable 31-n is connected to the signal input point 32. In other words. Reconnect the signal cable and input the signal again. D Fourier transform of the output signal of the phase detector 4 on the main antenna side when the signal u (k) is input from the signal generator 30₀′ (F), D Fourier transform of the output signal of the phase detector 4-n on the auxiliary antenna side_nIf ′ (F), then equations (44) and (45) are obtained. A block configuration diagram is as shown in FIG. The signal input from the signal generator 30 may not be the same as that shown in FIG. Equation (46) is obtained from equations (42) and (43), and equation (47) is obtained from equations (44) and (45). 4 and 5, only the number n is extracted from the auxiliary antenna.
[Expression 14]

[0062]
  When compensating for the frequency characteristic mismatch between channels, the compensation filter coefficient calculation uses the Fourier transform D of the output signal of the phase detector.₀(F) and D_n(F) is represented by C in formula (26).₀(F) and C_nUsed as (F). But it is A₀(F) = A_n(F) is a premise. Here, A₀(F) ≠ A_n(F) and D as it is₀(F) and D_nCannot use (F). This can be solved as follows. Expression (48) is obtained from Expressions (46) and (47). The left side of equation (48) is the amount obtained. When the left side of Expression (48) is set to Q (F) (Expression (49)), Expression (50) is obtained from Expression (46).
[Expression 15]

[0063]
  C in formula (26)₀(F) / C_n(F) is A₀(F) = A_nM0 (F) / M for (F)_nEqual to (F). Therefore, C in equation (26)₀(F) / C_nIf the left side of Equation (50), which is a measurable amount, is used instead of (F), the difference in the length of the signal cables 31, 31-n from the signal generator 30 to the signal input points 32, 32-n is Even if it cannot be actually measured, it is possible to calculate the coefficient of the compensation filter that eliminates the influence of the difference in length and compensates for the frequency characteristic mismatch between the channels.
[0064]
  As apparent from the above, according to the fourth embodiment, the compensation filter E_nWhen an RF signal is input via a cable when calculating the coefficient of (z), the difference in length of the signal cables 31, 31-n from the signal generator 30 to the signal input points 32, 32-n cannot be measured. From the Fourier transform of the output signal of the phase detector 4 on the main antenna side and the Fourier transform of the output signal of the phase detector 4-n on the auxiliary antenna side before and after reconnection of these signal cables, It is possible to calculate the compensation filter coefficient that can compensate for the frequency characteristic mismatch between the channels by eliminating the influence of the difference in length.
[0065]
Embodiment 5 FIG.
  Embodiments 2 to 4 can be applied to an adaptive array antenna in which each antenna element is equal. 6 is a block diagram showing an interference wave suppressing apparatus according to Embodiment 5 of the present invention. In the figure, the same reference numerals as those in FIG.
  Reference numerals 50-1, 50-2,..., 50-N denote N antenna elements, which constitute an array antenna. A receiver 2-n, an A / D converter 3-n, and a phase detector 4-n are sequentially connected to each antenna element 50-n. Here, the mismatch in frequency characteristics between channels is based on channel number 1 connected to the antenna element 50-1, and the frequency characteristics of other channels are matched with channel number 1. Therefore, the delay unit 5 is connected to the phase detector 4-1, and the compensation filter E_n(Z) is connected to the phase detectors 4-2, 4-3,..., 4-n.
[0066]
  Delay device 5 and compensation filter E_n(Z) The adaptive loads 7-1, 7-2,..., 7-N controlled by the adaptive load control means 25 with an appropriate adaptive algorithm are connected to the output of (z ≠ 1). The output signals of the respective adaptive loads 7-n are added by the adder 8, and the output signal of the load adding means 9 is a signal in which the interference wave is suppressed. The adaptive load control means 25 controls the adaptive load 7-n so that the difference signal power between the output signal of the load addition means 9 and the reference signal becomes as small as possible, for example.
[0067]
  The frequency characteristics at the output of the phase detector 4-1 connected to the antenna element 50-1, and the outputs of the phase detectors 4-2,..., 4-N connected to the other antenna elements 50-2,. The frequency characteristics are matched as much as possible, or the output signal of the delay unit 5 and each compensation filter E_n(Z) The compensation filter E so that the power of the difference from the output signal (n = 2,..., N) is as small as possible._nCompensate with (z). The method is the same as in Embodiments 2 to 4 described so far. In FIG. 6, an RF signal is transmitted by radio waves from an antenna 35 provided outside.
[0068]
  In the same manner as described in the second embodiment, if a radio wave is transmitted from the outside to compensate for mismatch in frequency characteristics between channels, a signal is not input to a specific location with a cable. There is an advantage that it is not necessary to consider the difference in delay. The compensation filter E_nIf the singular value decomposition method or the eigenvalue decomposition method is used for the coefficient calculation of (z), the coefficient calculation can be performed without causing numerical instability. And the compensation filter E_nIt becomes possible to prevent the coefficient absolute value of (z) from becoming a very large value.
[0069]
  In addition, as described in the third embodiment, the compensation filter E_nWhen the RF signal is input through the cable when calculating the coefficient of (z), if the delay time difference between the signal from the antenna element 50-n to the signal input point and the signal generator 30 to the signal input point can be measured, these are taken into consideration. Thus, the frequency characteristic mismatch between the channels can be compensated, so that a good broadband interference wave suppression performance can be obtained.
[0070]
  Further, as described in the fourth embodiment, the compensation filter E_nWhen the RF signal is input through the cable when calculating the coefficient of (z), even if the difference in the length of the signal cable from the signal generator 30 to the plurality of signal input points cannot be measured, the connection of these signal cables is performed. Compensation filter coefficients that can compensate for frequency characteristic mismatch between channels by eliminating the effect of signal cable length difference can be calculated from the Fourier transform of the output signal of the phase detector before and after switching. .
[0071]
【The invention's effect】
  As described above, according to the present invention, the coefficient setting means allows the signal delay time from the main antenna to the first signal input point and the signal delay from the plurality of auxiliary antennas to each second signal input point. The digital constituting the compensation means in consideration of the time, the delay time of the signal from the signal generator to the first signal input point, and the delay time of the signal from the signal generator to each second signal input point Set filter coefficientsThe difference between the length of the first signal cable from the signal generator to the first signal input point and the length of the second signal cable from the signal generator to the second signal input point is actually measured. If not, the Fourier transform result of the in-phase / quadrature signal output from the first and second conversion means is acquired, and the connection destination of the first signal cable is connected to the second signal from the first signal input point. After switching to the input point and switching the signal cable so that the connection destination of the second signal cable is switched from the second signal input point to the first signal input point, the first and second conversions are performed. The Fourier transform result of the in-phase / quadrature signal output from the means is acquired, and the coefficient of the digital filter that constitutes the compensation means is set using the Fourier transform result of the in-phase / quadrature signal before and after the signal cable switching.Therefore, there is a difference in the length of the signal cable from the main antenna and the plurality of auxiliary antennas to each signal input point and the length of the signal cable from the signal generator to each signal input point.Even if the length of the signal cable from the signal generator to each signal input point cannot be measured,There is an effect that it is possible to compensate for mismatch in frequency characteristics between channels.
[0072]
  According to this invention, the coefficient setting means takes into account the delay time of the signal from the plurality of antenna elements to each signal input point and the delay time of the signal from the signal generator to each signal input point, Set the coefficients of the digital filter that composes the compensation meansThe difference between the length of the first signal cable from the signal generator to the first signal input point and the length of the second signal cable from the signal generator to the second signal input point is actually measured. If not, the Fourier transform result of the in-phase / quadrature signal output from the first and second conversion means is acquired, and the connection destination of the first signal cable is connected to the second signal from the first signal input point. After switching to the input point and switching the signal cable so that the connection destination of the second signal cable is switched from the second signal input point to the first signal input point, the first and second conversions are performed. The Fourier transform result of the in-phase / quadrature signal output from the means is acquired, and the coefficient of the digital filter that constitutes the compensation means is set using the Fourier transform result of the in-phase / quadrature signal before and after the signal cable switching.Because it was configured asMultiple antenna elementsThere is a difference in signal cable length from the signal generator to each signal input point and signal cable length from the signal generator to each signal input point.Even if the length of the signal cable from the signal generator to each signal input point cannot be measured,There is an effect that it is possible to compensate for mismatch in frequency characteristics between channels.
[0073]
  According to the present invention, eigenvalue decomposition or singular value decomposition is used so that the weighted absolute mean square value of the difference between the spectrum of the output signal of the delay means and the spectrum of the output signal of the compensation means is minimized. Since the coefficient of the digital filter constituting the compensation means is set, the coefficient of the digital filter can be set without causing numerical instability, and the absolute value of the coefficient becomes a very large value. There is an effect that can be prevented.
[0074]
  According to the present invention, the compensation bandwidth of the compensation means is set to be narrower than the sampling frequency of the load adding means, and the coefficient of the digital filter constituting the compensation means is set with this compensation bandwidth. So ± 0.5f_sEven if there is a discontinuous frequency characteristic mismatch in the vicinity, there is an effect that good interference wave suppression can be obtained.
[Brief description of the drawings]
FIG. 1 is a configuration diagram showing an interference wave suppressing device according to Embodiment 1 of the present invention;
FIG. 2 is a block diagram showing an interference wave suppressing apparatus according to Embodiment 2 of the present invention.
FIG. 3 is a block configuration diagram showing a frequency response when there is a delay time difference.
FIG. 4 is a block configuration diagram relating to frequency response before and after signal cable reconnection with respect to compensation filter coefficient calculation;
FIG. 5 is a block diagram schematically showing that signal cable reconnection is performed with respect to compensation filter coefficient calculation.
FIG. 6 is a block diagram showing an interference wave suppressing apparatus according to Embodiment 5 of the present invention.
FIG. 7 is a block diagram showing a conventional interference wave suppressing device.
FIG. 8 is a block diagram showing a conventional interference wave suppressing apparatus using a tapped delay line circuit that adaptively controls coefficients.
FIG. 9 is a configuration diagram showing the inside of a tapped delay line circuit.
[Explanation of symbols]
  1-1,..., 1-N auxiliary antenna, 2 receiver (first conversion means), 2-1,..., 2-N receiver (second conversion means), 3 A / D converter (first 1,..., 3-N A / D converter (second conversion means), 4 phase detector (first conversion means), 4-1,..., 4-N phase Detector (second conversion means), 5 delay elements (delay means), 6-1, ..., 6-N compensation filter (compensation means), 7-1, ..., 7-N adaptive load, 8 adder, 9 load adding means, 10 subtractor (subtracting means), 21, 22 compensation filter coefficient setting means (coefficient setting means), 25 adaptive load control means (coefficient setting means), 30 signal generator, 31, 31-1, 31 -2, ..., 31-N Signal cable, 32, 32-1, 32-2, ..., 32-N Signal input point, 35 External antenna (signal) Transmitting means), 36 and 37 delay elements, 50-1, ..., 50-N antenna elements, 100 main antenna.

Claims (6)

First conversion means for converting the reception signal of the main antenna into in-phase / quadrature signals, delay means for delaying the in-phase / quadrature signals output from the first conversion means, and reception signals of the plurality of auxiliary antennas, respectively A plurality of second conversion means for converting into in-phase / quadrature signals, a plurality of compensation means for compensating frequency characteristics of the in-phase / quadrature signals output from the plurality of second conversion means, and the plurality of compensation means Load adding means for weighted addition of the output signals, subtracting means for subtracting the output signal of the load adding means from the output signal of the delay means, and a signal cable connecting the main antenna and the first converting means A predetermined signal is injected into the first signal input point above, and a second signal on each signal cable connecting the plurality of auxiliary antennas and the plurality of second conversion means is connected. A signal generator for injecting a predetermined signal to the input point, and when the predetermined signal is injected from the first and second signal input points by the signal generator, the output signal of the delay means and the compensation means In the interference wave suppressing device including the coefficient setting means for setting the coefficient of the digital filter constituting the compensation means so that the power of the difference from the output signal is minimized,
Said coefficient setting means includes a delay time of each of said second signal reaches the signal input point from the delay time of the signal to reach to the first signal input point from the main antenna and said plurality of auxiliary antennas, the signal generator vessels from consideration of the delay time of the first signal reaches the signal input point delay time and each from the signal generator of the second signal reaches the signal input point, constituting the compensating means digital filter It is used to set the coefficient,
The difference between the length of the first signal cable from the signal generator to the first signal input point and the length of the second signal cable from the signal generator to the second signal input point is measured. If not, the Fourier transform result of the in-phase / quadrature signal output from the first and second conversion means is acquired, and the connection destination of the first signal cable is connected to the first signal input point from the first signal input point. After switching to the second signal input point and switching the signal cable so that the connection destination of the second signal cable is switched from the second signal input point to the first signal input point. The Fourier transform result of the in-phase / quadrature signal output from the first and second transform means is acquired, and the Fourier transform result of the in-phase / quadrature signal before and after the signal cable reconnection is used to configure the compensation means. Interference suppression apparatus characterized by setting the coefficients of the barrel filter. First conversion means for converting the reception signal of the main antenna into in-phase / quadrature signals, delay means for delaying the in-phase / quadrature signals output from the first conversion means, and reception signals of the plurality of auxiliary antennas, respectively A plurality of second conversion means for converting into in-phase / quadrature signals, a plurality of compensation means for compensating frequency characteristics of the in-phase / quadrature signals output from the plurality of second conversion means, and the plurality of compensation means Load adding means for weighted addition of the output signals, subtracting means for subtracting the output signal of the load adding means from the output signal of the delay means, and a signal cable connecting the main antenna and the first converting means A predetermined signal is injected into the first signal input point above, and a second signal on each signal cable connecting the plurality of auxiliary antennas and the plurality of second conversion means is connected. A signal generator for injecting a predetermined signal to the input point, and a spectrum of the output signal of the delay means and the compensation when the predetermined signal is injected from the first and second signal input points by the signal generator In the interference wave suppressing device comprising coefficient setting means for setting the coefficient of the digital filter constituting the compensation means so that the spectrum of the output signal of the means matches.
Said coefficient setting means includes a delay time of each of said second signal reaches the signal input point from the delay time of the signal to reach to the first signal input point from the main antenna and said plurality of auxiliary antennas, the signal generator vessels from consideration of the delay time of the first signal reaches the signal input point delay time and each from the signal generator of the second signal reaches the signal input point, constituting the compensating means digital filter It is used to set the coefficient,
The difference between the length of the first signal cable from the signal generator to the first signal input point and the length of the second signal cable from the signal generator to the second signal input point is measured. If not, the Fourier transform result of the in-phase / quadrature signal output from the first and second conversion means is acquired, and the connection destination of the first signal cable is connected to the first signal input point from the first signal input point. After switching to the second signal input point and switching the signal cable so that the connection destination of the second signal cable is switched from the second signal input point to the first signal input point. The Fourier transform result of the in-phase / quadrature signal output from the first and second transform means is acquired, and the Fourier transform result of the in-phase / quadrature signal before and after the signal cable reconnection is used to configure the compensation means. Interference suppression apparatus characterized by setting the coefficients of the barrel filter. From a plurality of conversion means for converting received signals of a plurality of antenna elements constituting an array antenna into in-phase / quadrature signals, and among the plurality of conversion means , a first conversion means that is a conversion means related to a reference channel Delay means for delaying the output in-phase / quadrature signal, and among the plurality of conversion means, the in- phase / quadrature signal output from the plurality of second conversion means which are conversion means related to channels other than the reference channel Among the plurality of antenna elements, a plurality of compensation means for compensating frequency characteristics, weight addition means for weighted addition of the output signals of the delay means and the output signals of the plurality of compensation means, Injecting a predetermined signal to the first signal input point on the signal cable connecting the antenna element and the first conversion means, and the plurality of antenna elements A signal generator for injecting a predetermined signal to a second signal input point on each signal cable connecting the antenna element related to a channel other than the reference channel and the plurality of second conversion means; , when a predetermined signal is injected from each of the signal input point by the signal generator, so that the difference in power between the output signal of the output signal and the compensation means of the delay means is minimized, constituting the compensating means In the interference wave suppressing device comprising coefficient setting means for setting the coefficient of the digital filter to be
The coefficient setting means takes into account the delay time of the signal from the plurality of antenna elements to each signal input point and the delay time of the signal from the signal generator to each signal input point. It is one that sets the coefficient of the digital filter constituting the,
The difference between the length of the first signal cable from the signal generator to the first signal input point and the length of the second signal cable from the signal generator to the second signal input point is measured. If not, the Fourier transform result of the in-phase / quadrature signal output from the first and second conversion means is acquired, and the connection destination of the first signal cable is connected to the first signal input point from the first signal input point. After switching to the second signal input point and switching the signal cable so that the connection destination of the second signal cable is switched from the second signal input point to the first signal input point. The Fourier transform result of the in-phase / quadrature signal output from the first and second transform means is acquired, and the Fourier transform result of the in-phase / quadrature signal before and after the signal cable reconnection is used to configure the compensation means. Interference suppression apparatus characterized by setting the coefficients of the barrel filter. From a plurality of conversion means for converting received signals of a plurality of antenna elements constituting an array antenna into in-phase / quadrature signals, and among the plurality of conversion means , a first conversion means that is a conversion means related to a reference channel Delay means for delaying the output in-phase / quadrature signal, and among the plurality of conversion means, the in- phase / quadrature signal output from the plurality of second conversion means which are conversion means related to channels other than the reference channel Among the plurality of antenna elements, a plurality of compensation means for compensating frequency characteristics, weight addition means for weighted addition of the output signals of the delay means and the output signals of the plurality of compensation means, Injecting a predetermined signal to the first signal input point on the signal cable connecting the antenna element and the first conversion means, and the plurality of antenna elements A signal generator for injecting a predetermined signal to a second signal input point on each signal cable connecting the antenna element related to a channel other than the reference channel and the plurality of second conversion means; , when a predetermined signal is injected from each of the signal input point by the signal generator, so that the spectrum of the output signal of the spectrum and the compensating means of the output signal of the delay means are matched, constituting the compensating means In the interference wave suppressing device comprising coefficient setting means for setting the coefficient of the digital filter,
The coefficient setting means takes into account the delay time of the signal from the plurality of antenna elements to each signal input point and the delay time of the signal from the signal generator to each signal input point. It is one that sets the coefficient of the digital filter constituting the,
The difference between the length of the first signal cable from the signal generator to the first signal input point and the length of the second signal cable from the signal generator to the second signal input point is measured. If not, the Fourier transform result of the in-phase / quadrature signal output from the first and second conversion means is acquired, and the connection destination of the first signal cable is connected to the first signal input point from the first signal input point. After switching to the second signal input point and switching the signal cable so that the connection destination of the second signal cable is switched from the second signal input point to the first signal input point. The Fourier transform result of the in-phase / quadrature signal output from the first and second transform means is acquired, and the Fourier transform result of the in-phase / quadrature signal before and after the signal cable reconnection is used to configure the compensation means. Interference suppression apparatus characterized by setting the coefficients of the barrel filter.   The coefficient setting means uses the eigenvalue decomposition or the singular value decomposition so that the weighted absolute mean square value of the difference between the spectrum of the output signal of the delay means and the spectrum of the output signal of the compensation means is minimized. 5. The interference wave suppressing device according to claim 2, wherein a coefficient of a digital filter constituting the compensating means is set.   The coefficient setting means sets the compensation bandwidth of the compensation means to be narrower than the sampling frequency of the load adding means, and sets the coefficient of the digital filter constituting the compensation means with the compensation bandwidth. 5. The interference wave suppressing device according to claim 2, wherein the interference wave suppressing device is characterized in that:

Images