JPWO2002035774A1 - Data multiplexing network mechanism - Google Patents

Abstract

The main purpose is to replace a star-type host computer, which is a conventional data multiplexing device, with a subnet by token passing in asynchronous communication. However, each node constituting the subnet of the present invention does not need to hold the token during transmission. In addition, a method is also provided in which nodes connected in a ring topology are logically disconnected to make a logical bus type of linear connection.
The main functions and mechanisms of subnets and nodes are as follows.
1) Token (41) Transmits data (42-45) in the direction opposite to the traveling direction.
2) Each node (29 to 39) has a FIFO memory (43) having a capacity equal to the data length.
3) Flow control data for avoiding the congestion state is transmitted following the token signal.
4) Provide data and methods that can grasp the network connection status.

Description

【表２】ＭＡＲＳ−ＩＦがＣＰＵ部に受信端子を表示するルーティングビット

ノード３６のＣＰＵ部は、時刻Ｔ２で、データ３５＿３６Ｓを送信する準備をする。３５＿３６Ｓの先頭２ビットは、“１１”である。［表１］
備考：ノード３６は、ノード３５の接続方向を知らないため、“１１”として全接続端子にデータを送信する。
暫くしてノード３６のＭＡＲＳ−ＩＦは、端子Ｘからトークンを受信した。
ＣＰＵ部からの指示に従い、３５＿３６Ｓを、ＸおよびＹ端子から送信する。［１−４，表１，２−２］
備考：ＣＰＵ部３６は、“１１＋３５＿３６Ｓ”のデータをＭＡＲＳ−ＩＦ３６に送り、ＭＡＲＳ−ＩＦ３６は、データ“１１＋３５＿３６Ｓ”を“ｎｎ＋３５＿３６Ｓ”に変換し、伝送路２５、２６に送出する。従って、伝送路２５、２６に流れているデータは、ＣＰＵ部３６が作成した“１１”が削除されている。［１−４］は、先行技術のノード機能である。ＭＡＲＳ−ＩＦの機能で、先頭２ビットの“１１”は［２−２］に記載した“指定”に相当する。
ノード３５とノード３７は、データ３５＿３６Ｓの受信を開始した。
ＭＡＲＳ−ＩＦ３５は、端子Ｙから受信したデータ３５＿３６Ｓを自己宛と判断したので、ＣＰＵ部３５に“０１＋３５＿３６Ｓ”を伝える。［表２］
ノード３７は、時刻Ｔ２で、最初のデータ３５＿３７Ｓの先頭２ビットを“１１”として、送信する準備をしていた。［表１］
しかし、トークンではなくデータ３５＿３６Ｓを受信したので、端子Ｘから伝送路２７へ、転送データとして３５＿３６Ｓを送出する。［１−３，２−３］
ノード３６は、最初のデータ３５＿３６Ｓの送信が完了したので、トークンを端子Ｙから送出する。［１−２，２−１］
ノード３７は、端子Ｙからトークンを受信した。ＣＰＵ部からの指示に従い、３５＿３７Ｓを、ＸおよびＹ端子から送信する。［１−４，表１，２−２］
ノード３６は、データ３５＿３７Ｓの受信を開始した。自己宛データでないので、端子Ｘから伝送路２５へ、転送データとして送出する。［１−３，２−３］
ＭＡＲＳ−ＩＦ３５は、端子Ｙから受信したデータ３５＿３７Ｓを自己宛と判断したので、ＣＰＵ部３５に“０１＋３５＿３７Ｓ”を伝える。［表２］
ノード３７は、最初のデータ３５＿３７Ｓの送信が完了したので、トークンを端子Ｘから送出する。［１−２，２−１］
ＣＰＵ部３５は、“０１”付きデータ３５＿３６Ｓ”の受信を完了した。返事をするデータとして、“０１＋３６＿３５Ｒ”を作成した。
引き続き“０１＋３５＿３７Ｓ”の受信を完了した。返事をするデータとして、“０１＋３７＿３５Ｒ”を作成した。
備考：データ受信端子は、送信元ノードの接続方向を示していることになるので、受信ルーティングビット状態を、返信ルーティングビットとする。
ＭＡＲＳ−ＩＦ３５は、ＣＰＵ部３５から“０１＋３６＿３５Ｒ”および“０１＋３７＿３５Ｒ”を受け取り、端子Ｙに送出するデータであることを知った。端子Ｙからトークンが来るまで待機する。［表１，２−２］
備考：サブネットによるデータ多重化機構における最大回線容量として、図８の例では、ノード３５は“４”であり、他のノードは“１”であると説明した。ここでは、データ３６＿３５Ｒおよび３７＿３５Ｒの異なる２パケットを連続して送信する予定であり、トークン制御データとして加算する数字を［＋２］として待機したとする。［３−１］
ノード３５は、トークンが端子Ｙからきたので、データ３６＿３５Ｒおよび３７＿３５Ｒを連続して送出する。［表１，２−２］
ＭＡＲＳ−ＩＦ３６は、自己宛データ３６＿３５Ｒの受信を開始した。端子Ｘから受信したので、ＣＰＵ部３６に“１０＋３６＿３５Ｒ”を伝える。［表２］
ＭＡＲＳ−ＩＦ３７は、自己宛データ３７＿３５Ｒの受信を開始した。端子Ｙから受信したので、ＣＰＵ部３７に“０１＋３７＿３５Ｒ”を伝える。［表２］
以上の様に、第１表と第２表の処理をすることで、ルーティングが施される。従って、ルーティングテーブルを作成し参照しなくとも、受信データを参照するだけで、返送ルーティングが確保できる。
ノード３６と３７のＣＰＵ部が、自己ＭＡＲＳ−ＩＦから受け取るルーティングビットは“１０”“０１”と異なっている。このことは、図９でノード３７の伝送路接続において、右隣の伝送路と左隣の伝送路が逆に接続していたことに起因する。しかし、いづれにしても、データを送信したノードは、受信した端子側にあるので、伝送路を逆に接続しても、正しくルーティングがなされる。
◆　データ性質ビット処理の実施形態
〔発明の開示〕で、４種類のデータ性質、送出時のトークン利用状況、更に、ネットワークのトポロジーを調査することができると述べた。これらのデータ性質を、第３表にデータ性質ビットとして、例示し、まとめて掲げる。第３表のデータ性質ビットとして、４ビットを使用したが、“ｘ”記載したビットは、“０”あるいは“１”と、自由に規定できる。なお、この“ｘ”と同様に、図５や図６の“ｙ”については、“０”あるいは“１”と、自由に規定できる。
〔発明の開示〕で、ＭＡＲＳ−ＩＦの処理は、図３のノード３５、３６、３７、あるいは、図４のノード３３の様な２端子バスノードであり、図３の１端子を有する端末ノード、終端器、分岐器のデータ処理について、明確に示されていないので、第３表の分類に沿って、データ受信についての処理を、まとめて記載する。
【表３】データ性質ビットとデータ性質の対応

分岐器は、データ性質ビット“１１００”の受信データを除き、データ到来の伝送路を除くすべての伝送路へ送出する。
“００ｘｘ”：ブロードキャストチャネル用データの受信
２端子バスノード：すべてのデータを受信し、他の伝送路へ送出する。
端末ノード：すべてのデータを受信し、送出しない。
終端器：送出しない。
“０１ｘｘ”：宛先付ブロードキャストチャネル用データの受信
２端子バスノード：自己宛データを受信し、他の伝送路へすべて送出する。
端末ノード：自己宛データを受信し、他の伝送路へすべて送出しない。
終端器：すべて送出しない。
“１０ｘｘ”：宛先を特定するデータの受信
２端子バスノード：自己宛データを受信し、
自己宛データでないならば、他の伝送路へ送出する。
端末ノード：自己宛データを受信し、他の伝送路へすべて送出しない。
終端器：すべて送出しない。
“１１００”：トークンを利用したトポロジー調査用データの受信
２端子バスノード：すべて受信し、ＣＰＵ部で送出判断をする。
端末ノード：すべて受信し、ＣＰＵ部で送出判断をする。。
分岐器：あらかじめ決められている１つの伝送路へ送出する。
終端器：“１１００”のデータ性質ビットを示すコードのみ、もしくは、
すべて受信データを、トークンの到来で送出する。
なお、トークンは、コードまたはデータ送出後に、送出される。
“１１０１”：トークンを利用する隣接ノードを宛先とするデータ、及び、
“１１１ｘ”：トークンを無視する隣接ノードを宛先とするデータの受信
２端子バスノード：すべて受信し、他の伝送路へ送出しない。
端末ノード：すべて受信し、他の伝送路へ送出しない。
終端器：すべて送出しない。
以上が、データ性質ビットにより、受信したサブネットを構築する部材の処理である。データ性質ビット“１１０１”と“１１１ｘ”のデータについて、最良の実施形態を述べる。
本発明のデータ多重化ネットワーク機構は、トークンによる調停機構を有するネットワークを先行技術として用いている。もし、トークンが消滅したならば、システムダウンになり、ネットワークとして機能しなくなる。通常、トークンの消滅は、ネットワーク構築部材や伝送路の一部の故障、あるいは、トークン信号伝達時に、ノイズによりトークン信号が変化した場合である。従って、トークンが消滅しても、大部分の構築部材は正常であり、時には、すべて正常である。よって、ネットワークの故障を調べるのに、トークンを必要としない通信形態をネットワークに有しているならば、故障原因を探るのために、故障しているネットワークを利用できる。なお、トークンが巡回している正常時に、データ性質ビット“１１１ｘ”のデータを送出することもできるが、データ競合が発生することになり、ネットワークのシステムダウンになる。
以下、正常にトークンが巡回しているものとする。従って、データ性質ビット“１１１ｘ”のデータを送信する故障個所調査用通信形態を除いて述べる。
データ性質ビットで分類した“１１００”や“１１０１“は、隣接ノードを宛先とする。隣接ノードへの通信形態は、ポイントツーポイントチャネル用の基本形式であり、一般に良く知られているＲＳ−２３２−Ｃ（ＣＣＩＴＴ勧告Ｖ２．４）などが挙げられるであろう。本発明のデータ多重化ネットワーク機構は、ポイントツーポイントチャネル用をブロードキャストチャネルに変える先行技術を使っているが、チャネル形態の基本は、ポイントツーポイントチャネルである。従って、ポイントツーポイントチャネルを利用する場合、必ずしも、ノードの宛先を必要としない。データ性質ビット処理の発明たる部分は、ポイントツーポイントチャネルを利用して、サブネットのトポロジーを調査する方法である。よって、データ性質ビット“１１００”の処理に関する実施形態を詳述する。
前掲の第３表の分類に沿ってのまとめた“１１００”の受信処理で、ノードに関して、データの送出はＣＰＵ部で送出判断をすると記載した。ＣＰＵ部の送出判断基準を第４表に示す。なお、分岐器と終端器は、元々ＣＰＵ部が必要ないため、第４表は適用されない。
サブネットが図１の場合において、データ性質ビット“１１００”を使い、ノード３２がトポロジー調査を希望した例を挙げ、時間経過に沿って説明する。
ノード３２は、フレーム６１の調査ノード番号を“３２”とし、“ｙｙ”に適当な２進数を記入して、“０１１１００ｙｙ／３２”とし、ＭＡＲＳ−ＩＦ３２に送出する。［表４−１］
備考：ノード３２の調査ノード番号を“３２”とした。もし、ノード番号が未定であれば、“３２”を“００”などの未使用ノード番号にしても構わない。接続調査を希望したノードを区別できれば充分である。
ＭＡＲＳ−ＩＦ３２は、ルーティング用ビットが“０１”なので、ノード３２の端子Ｙから“ｎｎ１１００ｙｙ／３２”を送出する。［表１］
備考：図１でノード３２の端子Ｙに接続している伝送路は、ノード３１に接続されても、分岐器７０に接続されていても構わないが、ノード３２の端子Ｙとノード３１が接続されていたとする。
【表４】データ性質ビットが“１１００”のとき、ＣＰＵ部の処理

ＭＡＲＳ−ＩＦ３１は、“ｎｎ１１００ｙｙ／３２”を受信し、ＣＰＵ部３１に“０１１１００ｙｙ／３２”を送出する。［表２］
備考：ＣＰＵ部３１がＭＡＲＳ−ＩＦ３１から受け取るルーティングビットは、端子Ｘから受信したなら“１０”であり、端子Ｙから受信したなら“０１”である。端子Ｙから受信したとする。［表２］
ＣＰＵ部３１は、“０１１１００ｙｙ／３２”のデータを受信した。ルーティング用ビットをローテートさせ、“１０１１００ｙｙ／３２”とする。［表４−３］
ＣＰＵ部３１は、トポロジー調査の初期送信をしていないので、フレーム６２の値として“１０１１００ｙｙ／３２／３１”をＭＡＲＳ−ＩＦ３１に送信データとして渡す。［表４−４］
備考：ビット列“１００”を右ローテートさせると“０１０”になり、左ローテートさせると“００１”になる。しかし、“０１”または“１０”である２ビットならば、右ローテートも左ローテートも同じ値になり、反転した２ビットと同じ結果となる。［表４−３］
ＭＡＲＳ−ＩＦ３１は、“ｎｎ１１００ｙｙ／３２／３１”を、端子Ｘから終端器５０に送信する。［表１］
終端器５０は、“ｎｎ１１００ｙｙ／３２／３１”を受信した。“ｎｎ１１００ｙｙ”を送出する。
備考：前掲の第３表の分類に沿ってのまとめた“１１００”受信で、端末器の送出は“１１００”のデータ性質ビットを示すコード、もしくは、全受信データであるから、簡単なコードだけを送出した。
ＭＡＲＳ−ＩＦ３１は、“ｎｎ１１００ｙｙ”を端子Ｘから受信し、“１０１１００ｙｙ”をＣＰＵ部３１に送出する。［表２］
備考：ＣＰＵ部３１の接続端子は、端子Ｙがノード３２側であるとした。従って、終端器５０からの受信は、端子Ｘからの受信となる。
ＣＰＵ部３１は、“１０１１００ｙｙ”のデータを受け取った。コードだけの受信であるから、前回送信したデータ“１０１１００ｙｙ／３２／３１”を、受信したとみなす。［表４−２］
ＣＰＵ部３１は、トポロジー調査の初期送信をしていないので、フレーム６３の値として“０１１１００ｙｙ／３２／３１／３１”をＭＡＲＳ−ＩＦ３１に送信データとして渡す。［表４−３，表４−４］
ＭＡＲＳ−ＩＦ３１は、“ｎｎ１１００ｙｙ／３２／３１／３１”を、端子Ｙからノード３２に送信する。［表１］
ＭＡＲＳ−ＩＦ３２は、“ｎｎ１１００ｙｙ／３２／３１／３１”を端子Ｙから受信し、ＣＰＵ部３２に“０１１１００ｙｙ／３２／３１／３１”を送出する。［表２］
ＣＰＵ部３２は、“０１１１００ｙｙ／３２／３１／３１”のデータを受け取り、ルーティング用ビットをローテートさせ“１０１１００ｙｙ／３２／３１／３１”とする。［表４−２］
ＣＰＵ部３２は、トポロジー調査の初期送信を“０１１１００ｙｙ／３２”としていた。ＡＣＭ部ついて合致判断をすると、先頭のルーティング用ビットが異なるので、データ“１０１１００ｙｙ／３２／３１／３１／３２”を、フレーム６３の値として送信する。［表４−４］
ＭＡＲＳ−ＩＦ３１は、“ｎｎ１１００ｙｙ／３２／３１／３１／３２”を、端子Ｘから分岐器７０に送信する。［表１］
分岐器７０は、“ｎｎ１１００ｙｙ／３２／３１／３１／３２”を受信した。
一般データであれば、分岐器が受信したデータを他のすべての端子から送信するが、データ性質ビットが“１１００”なので、トークンの様に巡回転送され、ノード３３に送出する。
備考：前掲の第３表の分類に沿ってのまとめた“１１００”データの受信で、分岐器の処置として、あらかじめ決められている１つの伝送路へ送出する、と記載した。元々、分岐器はトークン信号の送出先をあらかじめ決めている。もちろん、“１１００”データの受信端子に対応する送出端子を、トークン信号の送出先と異なっていても構わないが、同じとした。
ノード３３は、“？？１１００ｙｙ／３２／３１／３１／３２”を分岐器７０から受信し、送出データを“ｎｎ１１００ｙｙ／３２／３１／３１／３２／３３”として、端末ノード２９に送出する。［表１，表４−４］
備考：図６の６３や上記でルーティングビットを“？？”として記載したが、実際には、“０１”あるいは“１０”である。
端末ノード２９は、“？？１１００ｙｙ／３２／３１／３１／３２／３３”をノード３３から受信し、送出データを“ｎｎ１１００ｙｙ／３２／３１／３１／３２／３３／２９”として、ノード３３に送出する。［表１，表４−４］
以下、ノード３３、分岐器７０、ノード３５、ノード３４、端末器６０、ノード３４、ノード３５、分岐器７０で、同様な処理がなされる。ＣＰＵ部３２がＭＡＲＳ−ＩＦ３２から受け取ったデータのルーティング用ビットを［表４−２］でローテートさせたフレーム６３のデータは、“０１１１００ｙｙ／３２／３１／３１／３２／３３／２９／３３／３５／３４／３４／３５”となる。ノード３２は、トポロジー調査の初期送信を“０１１１００ｙｙ／３２”としていた。ＡＣＭ部ついて合致判断をすると、同一なので、受信データの分析によりサブネット接続状況を知る。なお、分析するデータはコードを除く。［表４−４］
よって、“３２／３１／３１／３２／３３／２９／３３／３５／３４／３４／３５”のデータ列が、ノード３２を中心としたサブネットの接続状況を表現している。このデータ列の分析方法を、簡単に例示する。
１）同一の番号が続いているならば、隣接は“終端器”である。
例：“３１／３１”、“３４／３４”が続いている。
従って、“３１”と“３４”は、“終端器”に接している。
２）接続データ列に、番号が一つしかなければ、“端末ノード”である。
例：“２９”は一つしかないので、“端末ノード２９”である。
３）終端器および端末ノードを中心にして、左右対象番号があれば直線接続。
例：“３２／３１（端末器）３１／３２”から、　─ノード３２─ノード３１─終端器。
“３３／端末ノード２９／３３”から、　─ノード３３─端末ノード２９。
“３５／３４（端末）３４／３５”から、　─ノード３５─ノード３４─終端器。
４）３）で得られた直線接続と、調査開始ノード、分岐器接続を、考慮すれば、調査開始ノードは、サブネットのトポロジーを得ることができる。
以上で、個別の請求項に基づく最良の形態説明を終え、請求項を総合した最良の形態を示して説明する。
図１１は、図８や図１０と同様に、時間経過で表現した機構説明図である。時間は図１１の上側から下側に向かって流れており、時間軸としての図１１の左側に示した矢印で示される。図８や図１１と異なる点を挙げる。
終端器６０を端末ノード２９に変えた。ノード３１、３２、３３および３９を付け加えた。一枚の紙面に収める図１１にするため、ノード３５についてのみ、“宛先＿送信元”を記載した。引用符号４４のデータは、ノード３１と３３間で送受信される。データ３１＿３３または３３＿３１を表し、明確に示すため“Ｘ”を書き入れた。また、引用符号４５のデータは、ノード３７と３２間で送受信されるデータ３２＿３７を表し、データ３２＿３７及び３７＿３２を明確に示すため“／”を書き入れた。
図８のノード３５において、左隣に送出したデータ３５＿３６Ａ、３５＿３８Ｙ、右隣に送出したデータ３５＿３４、左隣に送出したデータ３５＿３６Ｂ、３５＿３７、３５＿３８Ｚについて、“省略”と記載した。また、図１０に記載したデータも“省略”と記載した。しかし、宛先が示されているので、データ性質ビットを“１０ｘｘ（宛先を特定するデータ）”とすれば、第３表の分類に沿ってのまとめた受信処理から、“省略”ではなく“消滅”になり、図１１では削除した。
図１１で“Ｘ”が書き込まれているデータ４４は、ノード３３と３１間のノード３２を単純に通過する。また、“／”が書き込まれているデータ４５は、ノード３７と３２間のノード３６、３５、３４、３３を単純に通過する。その他のデータは、ノード３５と送受信をする。従って、宛先と送信元のノード以外は、“Ｘ”や“／”のデータと同様に、中間にあるノードを単純に通過する。
ノード３５の送信データに注目すると、左側の３４から到来したトークン信号をトリガーとして、データ３４＿３５、３３＿３５、６９＿３５を送信し、右側の３６から到来したトークン信号をトリガーとして、データ３９＿３５、３６＿３５、３８＿３５を送信し、続いて、左側の３４から到来したトークン信号をトリガーとして、データ３４＿３５、３３＿３５、６９＿３５を送信した。データ３３＿３５と６９＿３５の送信時間間隔は、“／”のデータ４５の有無により異なる。もちろん、データ３３＿３５と６９＿３５の送信時間間隔を最小にするならば、“／”のデータ４５は、データ６９＿３５送信後にノード３５で転送される。しかし、データ３３＿３５と６９＿３５の送信時間間隔を最小にする条件として、ＦＩＦＯメモリ４３の容量は大きくなる。
ＦＩＦＯメモリの容量を最小にするには、次の順で送出端子の優先権を持たせることが、最良の実施形態である。
１．転送データ：ＭＡＲＳ−ＩＦが伝送路端子から受取るデータ
２．送信データ：ＭＡＲＳ−ＩＦに従属しているＣＰＵ部から受取るデータ
３．トークン信号
トークン信号の送出は、最下位の優先権であるため、［２−１］に記載された様に、送出予定の伝送路端子が使用されていると、トークン送出が留保される。なお、ノード３５は、トークンが１巡する間に、３４＿３５、３３＿３５、６９＿３５、および、３９＿３５、３６＿３５、３８＿３５の６個のパケットを送信する。この意味は、本発明の目的が図２での広域伝送路における多重化を実現することにあり、上位の広域伝送路１４から受取ったデータを下位の広域伝送路１１へすみやかに転送するためである。
もし、図１１の端末ノード２９と終端器５０が、図４に記載したリング接続のノード２９と同値であれば、ノード３５を単純に通過させるデータ“／”が多くなる処置として、端末ノード２９を２端子バスノードとし、ノード３６または３７で論理切断することで、ノード３５を単純に通過させるデータ“／”を、非常に少なくし、サブネットの伝送路利用効率を高めることができる。
ノードをＣＰＵ部とＭＡＲＳ−ＩＦに分割した。処理変化が大きい輻輳制御部分をＣＰＵ部のプログラムで実現させ、ＭＡＲＳ−ＩＦを小さなＬＳＩとするならば、先行技術に記載されている回路の一部変更で実現できる。
よって、ＭＡＲＳ−ＩＦの機能を満足させるインターフェースＬＳＩを製作するための指針となるブロック図を〔産業上の利用可能性〕で述べる。
産業上の利用可能性
図１２は、ＣＰＵ部に、主に輻輳制御を担当させ、ＭＡＲＳ−ＩＦをインターフェースＬＳＩ７６にすることで、本発明を産業に利用できるようにしたものである。以下、図１２について説明する。
ＣＰＵ部は、上位広域伝送路１４と下位広域伝送路１１を通じて、ネットワークを構築し、また、インターフェースＬＳＩ７６を通じてサブネットと接続する。ＣＰＵ部と７６の信号伝達線として、ＲＴ、ＳＴ、ＣＴＲ、ＣＴＷ、ＳＤ、ＲＤの名称を与える。
ＲＴ，ＳＴ：ＲＴは、制御データ付きトークンを７６からＣＰＵ部に、ＳＴは制御データ付きトークンをＣＰＵ部から７６に、伝達する信号線である。
ＣＴＲ，ＣＴＷ：ＣＴＲは７６からＣＰＵ部に、ＣＴＷはＣＰＵ部から７６に、制御信号を伝達する信号線である。
ＳＤ，ＲＤ：ＳＤはサブネットに送出するデータをＣＰＵ部から７６に、ＲＤはサブネットから得るデータを７６からＣＰＵ部に、伝達する信号線である。
インターフェースＬＳＩ７６は、伝送路接続端子Ｘ７８と伝送路接続端子Ｙ７９を持っている。７６の内部ブロックは、Ｘ側の伝送路からの受信信号をトークン信号とデータ信号に分離する８０、Ｙ側の伝送路からの受信信号をトークンとデータに分離する８１、Ｘ側の伝送路へ送信信号を与えるパラレル／シリアル変換器８２、Ｙ側の伝送路へ送信信号を与えるパラレル／シリアル変換器８３、受信トークンを一時保管するバッファ８４、受信データを記憶するＦＩＦＯメモリ８５、８４と８５に記憶した信号を制御するコントローラ８６、ＣＰＵ部が送出した制御データ付きトークンを一時保管するバッファ８７、ＣＰＵ部が送出したデータを一時保管するバッファ８８、各ブロック間の信号伝達を制御するコントローラ８９、および、バッファ８４内の制御データ付きトークンをＣＰＵ部への送出するゲート９０、ＦＩＦＯ８５内のデータをＣＰＵ部へ送出するゲート９１、ＦＩＦＯ８５内の転送データをＸ側へ送出するゲート９２、Ｙ側へ送出するゲート９３、バッファ８４内の制御データが付いていないトークンを直接Ｘ側へ送出するゲート９４、直接Ｙ側へ送出するゲート９５、バッファ８８内の送信データをＸ側へ送出するゲート９６、Ｙ側へ送出するゲート９７、バッファ８７内の制御データ付きトークンをＸ側へ送出するゲート９８、Ｙ側へ送出するゲート９９、で構成されている。なお、ゲート９０〜９９の制御線は、８９に接続されているが、煩雑な図になるのを避けるため省略した。
図１２の構成において、終端器をＸ側とするには、８０の“Ｄａｔａ”出力を図６のコード以外のデータならば停止させ、８０がトークン信号の到来を８９に知らせることで、９４のゲートを開く制御を行う。もし、図６のコードを８０が受信したことを８９に知らせたならば、８５にデータが記憶されるが、先頭のコードのみを９２で通過させることで対応できる。もちろん、８５のデータをすべて９２から送出しても良い。なお、サブネットに、只一つのトークン信号しか存在しないので、終端器側でない７９からトークンが到来したとき、受送信制御をしても、データやトークンが論理切断されていることになる。なぜならば、図１１で、伝送路端子ＸとＹの両方から同時にデータを受取ることはないことは、明白である。
図１２の構成において、終端器をＹ側とするには、８１の“Ｄａｔａ”出力を図６のコード以外のデータならば停止させ、８１がトークン信号を８９に知らせることで、９５のゲートを開く制御を行う。もし、図６のコードを８１が受信したことを８９に知らせたならば、８５にデータが記憶され、記憶データすべて、あるいは、データ先頭のコードのみを、９３で通過させることで対応できる。
図１２の構成において、ゲート９０、９８、９９を開く制御をせず、コントローラ８９が〔先行技術の概略〕で示した［１−１］〜［１−５］の制御をすれば、“先行技術で示されたノード”が実現できる。なお、ＦＩＦＯ８５は、先入れ先出しメモリで有るが故に、先行技術で示されたノードにおいて、単純なバッファの役目をする。
図１２の構成において、ゲート９０、９８、９９を開く制御をせず、コントローラ８９が〔発明の開示〕で示した［２−１］〜［２−３］の制御をすれば、２端子バスノードにおけるＭＡＲＳ−ＩＦ機能が実現できる。なお、端末ノードに使われるＭＡＲＳ−ＩＦ機能を実現させるには、８０または８１において、終端器としての機能を優先させれば良い。
図１２の構成において、ゲート９４を開く制御をせず、制御データ付トークンを、ゲート９０を通過させてＣＰＵ部に送り、ゲート９８、９９を開く制御を８９で行うことで、ＣＰＵ部が〔発明の開示〕で示した［３−１］〜［３−２］の制御をすれば、輻輳状態を回避する制御データの処理をするノード機能が実現できる。
ルーティングビットの処理の内、受信端子を受信データに付加するには、データ受信時に８０または８１がデータの先頭に付加すれば良い。なお、受信データは、ＦＩＦＯ８５に送られる。
ルーティングビットの処理の内、ＣＰＵ部から送られたデータの先頭２ビットで、送出端子を選択しデータを変更する処置については、コントローラ８９が送信データバッファ８８を調べ、ゲート９６または９７を開くことで送出端子を選択できる。データの変更処理は、パラレル／シリアル変換器８２および８３でシリアルデータにする際に、ルーティングビットを消去すれば良い。
データ性質ビットによる受信処理は、第３表の説明を、ゲート９１、９２および９３に適用させれば良い。
以上の様に、図１２の各ゲートを制御することで、ＭＡＲＳ−ＩＦがインターフェースＬＳＩで実現できる。
本発明の主目的は、多重化装置であるスター型のホストコンピュータを、サブネットで置き換えることにあった。しかし、産業上の利用として、サブネットをＬＡＮ（Ｌｏｃａｌ　Ａｒｉａ　Ｎｅｔｗｏｒｋ）として利用しても、何ら差し障りも生じない。なぜならば、階層構造の図２で、最上層の広域伝送路１４を接続するホストコンピュータに相当する機材は、現実社会において存在しない。また、逆に、最下層の広域伝送路には、パーソナルコンピュータや産業機械のセンサーなどが接続されている。しかし、信号伝達速度は光速を越えることができないので、サブネットの大きさに制限がある。
最後に、サブネットを構成するノード間の距離と伝送路データ転送速度についての考察を述べる。
本発明は、〔背景技術〕の文頭で示したデータ多重の一つであるＱＡＭの利用でないので、１ｂｐｓ／ｂａｕｄの変調として考察する。
図１１で、ノード３７がデータ３２＿３７の送信を開始し、
同時に、トークンを隣接ノード３８に送り出し、
ノード３８がデータ３５＿３８の送信を開始した場合、
ノード３７がデータ３５＿３８の受信を開始するまでに、
信号は　３７→３８→３７と伝送路を往復する。
１パケットが５０バイト（５０×８ｂｉｔ）で１００Ｍｂｐｓの伝送速度ならば、
１パケットが伝送路を占有する帯域長は、
帯域長＝５０×８×光速／１００Ｍ＝１２００ｍ　　である。
従って、ノード間距離＝帯域長／２＝６００ｍ　　であるならば、
ノード３７のＦＩＦＯに、データ３５＿３８を蓄えることなく、
データ３５＿３８がノード３６に転送される。
以上の計算から、
１００Ｍｂｐｓの伝送速度で、ノード間距離６００ｍの場合、
１パケットが５０バイトより多ければ、
サブネットの伝送路利用効率は、最大１００％まで高めることが可能である。
しかし、１パケットが５０バイトより少なければ、
サブネットの伝送路利用効率に限界が生じる。
もちろん、ノード間距離を短くしたり、低い伝送速度でサブネットを利用するならば、１パケットのデータ長を５０バイトより少なくしても、伝送路利用効率を、最大１００％まで高めることが可能である。
なお、図８、図１０、図１１において、１パケットのデータ長を同じにして記載したが、図６でのフレームフォーマット６１、６２、６３の様に、データ長が異なる多種類のパケットを、本発明のサブネット伝送路で通信しても、上記の伝送路利用効率に影響するだけであることを、述べておく。
【図面の簡単な説明】
図１は、先行技術のサブネットである。
図２は、本発明が適用される階層構造の多重通信ネットワークの例である。
図３は、図２の２０を拡大した広域伝送路付近の接続実施例である。
図４は、リング型伝送路をバス型伝送路にする機構図である。
図５は、ルーティング調査用フレーム・フォーマットである。
図６は、サブネット接続調査用フレーム・フォーマットである。
図７は、サブネットと広域伝送路の接続実施例である。
図８は、図７でのデータの流れを時間経過で表現した機構説明図である。
図９は、図７の接続実施例の拡大図である。
図１０は、図９でのルーティング機構を時間経過で表現した図である。
図１１は、データの流れを時間経過で表現した総合説明図である。
図１２は、ＭＡＲＳ−ＩＦをＬＳＩで実現した場合のブロックである。
符号の説明
１０、２０：本発明の実施形態であるサブネット
１１〜１４：広域伝送路
１５〜１９、２１〜２８：サブネット内の伝送路
２９：データおよびトークンを論理で非接続にしたノード
３１〜３９：ノード
４１：トークン
４２〜４６：送受信データまたは転送中データ
４３：一時記憶ができるＦＩＦＯメモリに蓄えられたデータ
５０、６０、７０：特定番号不要な終端器、または、特定番号不要な分岐器
５１〜５３：ルーティング調査用フレーム・フォーマット
６１〜６３：サブネット接続調査用フレーム・フォーマット
７６〜９９：ＭＡＲＳ−ＩＦのブロック回路Technical field
The present invention does not use a multiplexing device as a method for realizing multiplexed transmission of data, and a large number of nodes constituting a subnet satisfy a predetermined condition processing for data transmission, transfer and reception, thereby achieving wide area transmission. The present invention relates to a mechanism for realizing data multiplexing in a transmission path.
Background art
Data multiplexing includes a method of using a carrier such as phase shift called QAM (quadrature amplitude modulation), and a time division multiplexing method of storing data in a time slot called TDM (time division multiplexer). The latter TDM is developed as an ATM (Asynchronous Transfer Mode) and is asynchronous in a method of storing data in a time slot. However, as a condition for using time slots for time division multiplexing, it is necessary to keep the packet length constant. If the method does not use a time slot, there is no limit on the packet length. The conventional time-division multiplexing method that does not use time slots is constructed as a star network, connected in a star topology centering on one host computer, and transmitting data sent from many terminal computers to the host computer. Multiplexing was realized internally.
The network structure is roughly divided into a point-to-point channel and a broadcast channel. Point-to-point channels are classified into star, ring, and tree types, and broadcast channels are classified as bus types. Methods that combine the point-to-point channel and the broadcast channel are IEEE standard 802.5 Token Ring and PCT / JP99 / 04284 (network with arbitration mechanism by token).
The IEEE standard 802.4 token bus circulates tokens using a logical ring on a bus-type broadcast channel. Conversely, in a conventional ring-type point-to-point channel, since a logical bus is not realized, the ring-type node computer has a poor measure against failure.
Communication networks have congestion problems that degrade performance. As a method of controlling congestion between multiplexing apparatuses, there are a method using a large number of permits circulating in a subnet, a method for generating a choke packet, a method using flow control, and the like. In any case, control data must be transmitted between all multiplexing devices and processed by the control data.
The prior art on which the present invention is based is PCT / JP99 / 04284 (network having an arbitration mechanism using tokens) described above. However, at the time of the international filing of the present invention, it has not been published and the period for claiming the priority has passed, so that the prior art will be briefly described.
Although the terms “node”, “branch device” and “terminal device” are used in the above-mentioned specification, they are referred to as “node”, “branch device” and “terminator” in this specification.
Further, the “node” will be schematically described by dividing it into a “bus node” having two terminals and a “terminal node” having one terminal. In this specification, the configuration of a “node” is divided into a CPU unit and a MARS-IF (Message Access and Repeat Server interface). The concept of MARS is described in detail in International Publication No. WO 96/31968 (PCT / JP95 / 00653, a network system for computers having a transfer identification and arbitration mechanism). In brief, the concept of MARS is that a message such as a destination header is transmitted to a server called an IMP (Interface Message Processor), a data exchange, a relay system, a hub node, or the like that constructs a network. The operation can be prompted by specifying the transfer destination of the data, and the computer need only give the operation instruction in the header of the data string.
Overview of prior art
FIG. 1 is a network obtained by partially changing FIG. 1 of PCT / JP99 / 04284.
The "node", "branch device", and "termination device" are connected by a point-to-point channel. The data flow realizes a broadcast channel bus type. Therefore, data transmitted from an arbitrary node is transferred by a "bus node" having two terminals, is transferred by a "branch device" to all of a plurality of transmission paths, and is "terminator" or "terminal node having one terminal.""Will disappear. The only member capable of transmitting transmission data is the node computer, and the "branch unit" merely transfers the received data.
Because of the point-to-point channel connection, if data can be transmitted between adjacent members at any time, if the transmission path is full-duplex communication, there is no data contention on the transmission path. Causes data race. In order to avoid data conflict, in the prior art, transmission of transmission data is controlled using a token. The token passes through the “bus node”, is transmitted to a predetermined transmission path by the “branch device”, and is returned after being returned by the “terminator” or “terminal node”. Thus, the token will continue to travel forever.
The "bus node" or "terminal node" from which the token has arrived obtains the transmission right of the transmission data. Cycle the token after sending the transmission data.
The functions are summarized as follows for each member.
Functions of "bus node" (code numbers 31, 32, 33, 34, 35):
When the token arrives, transmission data can be sent.
Received data is received by the node computer and sent out as transfer data.
Send the received token to another transmission path.
Functions of “terminal node” (code number 29):
When the token arrives, transmission data can be sent.
The received data is received by the node computer.
Return the receiving token.
The functions of the "branch device" (reference numeral 70):
Sends received data to all other transmission paths.
Send the received token to a predetermined transmission path.
Functions of “Terminal” (reference numerals 50 and 60):
Delete received data.
Return the receiving token.
The above is the outline of the prior art. In addition, since it discloses that a broadcast channel can be realized in a point-to-point network, it emphasizes that it is not necessary to number all members. However, it is advisable to number the node members in order to use the network effectively. In FIG. 1, numerals 31 to 35 and 29 indicate the numbers in the frames, and numerals 50, 60, and 70 are indicated by using lead lines because there is no need to add the numbers.
Finally, in order to clarify the difference from the prior art in disclosing the present invention, the functions of the conventional nodes are extracted from the prior art [disclosure of the invention] and described.
The number at the beginning of the text will be used in the following description.
◆ Function of preceding node
1-1. It has two transmission line connection terminals for cutting and inserting the transmission line.
1-2. If a token signal is received from any one of the transmission path connection terminals, the token signal is transmitted from the other transmission path connection terminal immediately if there is no transmission data, immediately after the transmission data is transmitted if there is transmission data.
1-3. When a data signal is received from one transmission path connection terminal, it is transmitted from the other transmission path connection terminal.
1-4. The transmission of its own transmission data is started when the token signal is received, and is transmitted from two transmission path connection terminals.
1-5. Except during transmission of a data signal or a token signal, signals from two transmission path connection terminals can be received.
Disclosure of the invention
The present invention provides a method for implementing multiplexed data transmission, in which a large number of nodes constituting a subnet satisfy conditional processing for data transmission, transfer, and reception instead of one multiplexer. And a mechanism for realizing data multiplexing in a wide area transmission path. Further, a header message of transmission / reception data is processed by a MARS-IF (Message Access and Repeat Server interface), so that the range of application is further expanded.
The present invention is disclosed in the following order.
First, the multiplexing to which the present invention is applied will be briefly described, the implementation in a subnet will be described, and the invention will be disclosed regarding the network topology.
Next, the invention will be disclosed with respect to the function of the “node”.
Lastly, the invention will be disclosed with respect to the processing of the MARS-IF for reducing the load on the CPU by dividing the configuration of the "node" into the CPU and the MARS-IF.
■ Network topology
FIG. 2 shows a hierarchical network. If 10 or 20 is viewed as a computer or an IMP (Interface Message Processor), it becomes a network in which the star type is connected in a tree structure. In addition, because of the hierarchical structure, as shown on the left side of FIG.
Looking mainly at 20, the number of transmission paths connected to the lower layer is 11, which is four. The number of transmission lines connected to the further upper layer is 14, one. Therefore, 20 means that four 10 Mbps communication data are multiplexed to 100 Mbps communication data. The present invention provides 20 as a subnet instead of a single host computer or IMP.
Since 20 is a subnet, the nodes are internally connected to a subnet transmission line and have a function of connecting the lower layer transmission line 11 and the wide area transmission line 14.
The above is the multiplexed hierarchical structure network to which the present invention is applied. Hereinafter, the subnet will be disclosed in detail.
The subnet 20 is described as a bus type in FIG. 2 since the prior art is a bus type. However, in the present invention, the connection shape of the nodes may be a ring shape. The first aspect of the present invention provides a method of substantially forming a bus even if the members constituting the subnet are arranged in a ring shape.
In the conventional token bus, although the topology is a bus type, tokens are circulated using a logical ring. Conversely, the present invention provides a method in which the topology is a ring type, and the token is formed in a logical ring different from the actual connection, so that the token is made equivalent to the bus type of the prior art, and the bus type is logically used. It is.
■ “Node” function
Prior art nodes have been described in [Prior Art Summary]. The functions performed by the node according to the present invention will be described with numbers following [Overview of Prior Art].
◆ Node function in subnet
2-1. As soon as the token signal is received from one of the transmission path connection terminals, the token signal is transmitted from the other transmission path terminal.
However, if the data is being transmitted from the token signal transmission scheduled terminal, the token is reserved until the data transmission is completed, and the token signal is transmitted after the data transmission.
2-2. Unless otherwise specified, the transmission of its own transmission data is from the transmission path connection terminal that has received the token signal, and immediately after the token signal is received.
2-3. When data is received from any one transmission path connection terminal, the data is stored in its own FIFO memory (first in, first out memory), and the data stored in the FIFO memory is transmitted from the other transmission path terminal.
However, if the self transmission data is being transmitted from the data transmission scheduled terminal of the FIFO memory, the data transmission of the memory is started after the completion of the data transmission. If the data is divided into a plurality of packets, it is assumed that one packet has been transmitted.
◆ Data processing node function to avoid congestion by flow control
3-1. A number is added as control data to the token signals described in 2-1 and 2-2 above, and the number is increased only once before transmitting a series of self-transmission data, and is decreased only once after transmitting a series of self-transmission data. .
3-2. In 2-2, the following proviso is added.
However, if the number of the control data added to the token signal is equal to or greater than a predetermined value, the control unit does not transmit its own transmission data.
The subnet of the present invention differs from the conventional and prior arts in how tokens are captured. The token of the present invention serves as a trigger for obtaining the transmission right of transmission data, and immediately sends a token signal to the next node. Therefore, there is no need to hold a token signal during transmission of transmission data as in the prior art, and it is not the concept of transmission data with a token as used in the prior art. The necessary condition for making the token signal a simple trigger signal is to transmit the transmission data to the incoming terminal of the token signal and transmit the token signal to the other terminal. By using the token as a trigger for obtaining the transmission data transmission right, it is possible to transmit a signal at the same time, and the transmission path efficiency is improved.
Since it is not necessary to manage tokens by numbers in the prior art, a simple signal sequence is sufficient for the token. The control data numbered token of the present invention has the same signal form as the numbered token like the prior art token, but the meaning of the attached number is a number that avoids the congestion state by flow control, and the identification of the node. Not for numbers.
■ MARS-IF processing
The “node” includes a CPU and a MARS-IF. In order to reduce the load on the CPU unit, the MARS-IF performs various types of processing regarding data transfer.
Basic data for prompting various types of processing is added to the head of transmission data transmitted from the CPU unit through the MARS-IF. The basic data that prompts the process attached to the head of the transmission data is named ACM (Access Control Message), and will be described below.
The ACM is divided into a plurality of routing bits and a plurality of transmission data property bits.
◆ The routing bit is for selecting the transmission path to which the receiving node is connected. The CPU that transmits the transmission data uses a routing bit designating a transmission terminal. The MARS-IF selects a transmission terminal in accordance with the routing bit and transmits the transmission data, but the routing bit of the transmission data is erased by the MARS-IF to be transmitted. The MARS-IF that has received the data sets a routing bit in accordance with the receiving terminal and transfers the bit to the CPU. The CPU unit that has received the data can know the transmission path to which the transmission source node is connected by checking the ACM at the head of the data. By using the routing bits by the CPU unit, the transmission line use efficiency of the subnet can be improved.
◆ The transmission data property bit is for determining whether or not to urge a large number of nodes constituting the subnet to receive. MARS-IF performs four types of processing according to data property bits. For four types, the processing of the MARS-IF that received the data is listed.
* Data for broadcast channel:
The data is transferred to its own CPU unit, which receives it.
The data is also transferred to an adjacent node so that other CPU units can receive the data.
* Data for broadcast channel with destination:
It is determined whether to transfer data to its own CPU unit.
If the destination described in the data matches the own node number,
The data is transferred to its own CPU unit, which receives it.
If the destination described in the data does not match the own node number,
Does not transfer data to its own CPU.
The data is also transferred to an adjacent node so that the other nodes can judge the reception.
* Data specifying the destination:
It is determined whether to transfer data to its own CPU unit.
If the destination described in the data matches the own node number,
The data is transferred to its own CPU unit, which receives it.
It is not transferred to the adjacent node.
If the destination described in the data does not match the own node number,
It does not transfer data to its own CPU, but transfers it to an adjacent node.
* Data addressed to an adjacent node:
The data is transferred to its own CPU unit, which receives it.
It is not transferred to the adjacent node.
The above four types will be described in association with the related art.
"Broadcast channel data" is the same as the conventional bus type, and data is distributed to nodes connected to all transmission paths.
“Destination broadcast channel data” corresponds to broadcast by group in a conventional network.
“Data specifying a destination” is a usage method that is frequently used in a general network. However, in the conventional bus-type network, since the transmission path is directly branched, it cannot be transferred to an adjacent node, and the transmission path is not highly efficient. In the prior art, the transmission path is not directly branched, but is selected by the CPU using "broadcast channel data with destination".
“Data destined for an adjacent node” has the same processing of the MARS-IF that has received the data, but is further divided into two according to the processing of the MARS-IF that sends out the transmission data. One starts sending when a token arrives, and the other immediately sends when transmission data is received from the CPU unit irrespective of the token arrival. The former is a network-specific data transmission of the prior art, and it is possible to investigate the topology of the network. The latter is data transmission that does not require a token, and thus corresponds to conventional point-to-point data transmission and reception.
BEST MODE FOR CARRYING OUT THE INVENTION
First, it will be described that the best mode for carrying out the present invention is to divide a node and a MARS-IF into members constituting a node.
Next, the best mode based on the individual claims will be described.
Finally, the best mode in which the claims are integrated will be described.
2 has been described as a subnet for implementing the present invention. FIG. 3 shows an example in which the vicinity of the wide area transmission line 14 is enlarged.
In the enlarged view of FIG. 3, there are nodes 35, 36, and 37. The wide area transmission line 14 and the lower layer transmission line 11 are at the node 35, the lower layer transmission line 12 is at the node 36, and the lower layer transmission line 13 is at the node 37. The subnet transmission line 25 connects the nodes 35 and 36, and the subnet transmission line 26 connects the nodes 36 and 37. In FIG. 2, the lower-layer transmission line is described as the representative code 11, but in FIG. The subnet transmission lines 24 and 27 are connected to other nodes and terminators. The node 35 includes a CPU 35 and a MARS-IF 35, and the node 36 includes a CPU 36 and a MARS-IF 36. Similarly, the configuration of the node 37 includes a CPU 37 and a MARS-IF 37. The detailed operation circuit of the MARS-IF will be described in [Industrial Applicability].
In FIG. 2, it has been described that 10 Mbps data of the lower layer transmission line 11 is multiplexed by 20 and sent to the wide area transmission line 14. 3, the 10 Mbps data of the lower-layer transmission lines 12 and 13 is received by the CPU units 36 and 37, passes through the MARS-IFs 36 and 37, is transmitted by the transmission lines 25 and 26, and is transmitted by the MARS-IFs. The data is multiplexed as a result of being transmitted from the IF 35 to the CPU unit 36 and transmitted as 100 Mbps data over the wide area transmission path 14. The wide area transmission line 14 and the subnet transmission lines 24, 25, 26, and 27 have a transmission speed of 100 Mbps, and the lower layer transmission lines 11, 12, and 13 have a transmission speed of 10 Mbps.
In FIG. 3, the MARS-IF is only daisy-chain connected in terms of the subnet transmission path. CPU units capable of transmitting and receiving data are cascaded to each MARS-IF. Conversely, from the viewpoint of the CPU unit, the only operation is to pass or receive the data of the wide area transmission line or the lower layer transmission line to the MARS-IF which is cascade-connected. Therefore, dividing a node into a CPU and a MARS-IF and performing data processing will be the best mode for carrying out the invention.
In the following description, when it is not necessary to clearly separate the functions of the CPU unit and MARS-IF, it is described as “node”, and when it is necessary to divide the functions, it is described as “CPU unit” or “MARS-IF”. Describe.
The best mode based on the individual claims will be described along with [Disclosure of the Invention].
■ Embodiment of network topology
FIG. 4 is a diagram illustrating a mechanism in which a ring-type transmission line is changed to a bus-type transmission line.
In FIG. 4, nodes 31, 32, 33, 34, 35, and 29 are connected by transmission lines 16, 17, 18, 19, 21, and 22, respectively, and the nodes are connected by rings. Further, referring to FIG. 4 in detail, the node 29 logically disconnects the transmission path. It can be seen that the disconnection in this logic is disconnected because the contents of the node 29 are separated into the terminator 50 and the terminal node 29. The nodes 31 to 35 have the same logical connection and are not disconnected by logic. The nodes 31 to 35 have the same mechanism as that of FIG. 1 showing the prior art, and have a two-terminal bus node as described as the node 33 as a representative. As for the node 29, the terminal node 29 and the terminator 50 in FIG. Therefore, the ring transmission line of FIG. 4 is represented by a straight line such as “terminator” ─ “node 31” ─ “node 32” ─ “node 33” ─ “node 34” ─ “node 35” ─ “terminal node 29” It has the same value as the connected bus type.
The terminator 50 has the same structure as that already shown in the prior art. However, the terminal node 29 is not disclosed in the prior art and will be described in detail.
A “terminal node” having one terminal has the following mechanism.
* It has one transmission line connection terminal to connect to the end of the transmission line.
* If all token signals are received from the transmission line connection terminal,
If there is no transmission data, immediately after sending the transmission data if there is transmission data,
Returns the token signal.
* If a data signal is received from the transmission path connection terminal,
If it is data addressed to itself, it will be received data,
If it is not self-addressed data, discard it.
The above mechanism results in a terminal node that has the characteristics of both a prior art terminator and a node. The terminal node is desirably assigned a node number in order to determine whether or not the data is addressed to itself. However, the terminators need not be numbered. In order to clarify the necessity and unnecessaryness of the numbers, in the description of the node 29 in FIG. 4, the configuration is shown by the terminal node 29 and the terminator in the broken line frame, and 50 draws a leader line from the terminator.
In FIG. 1, if the terminal node 29 and the terminator 50 are removed and connected at the new node 29, a ring is formed by 33, 29, 31, 32 and the branch unit 70, and a daisy chain connection is established from the branch unit 70. 35 and 34 and the terminator 60 are connected. The “e” node connection topology can be a “ε” connection or a “C” connection topology depending on where it is cut off by logic. That is, in general, even if the members constituting the subnet of the present invention are arranged in a ring shape, the method of substantially forming a bus type is, when constructing an actual network, after constructing to freely connect nodes, In operation, it indicates that a bus can be made by logical disconnection. This not only provides a new method for dealing with node failures, but may also be a method for dealing with transmission line congestion in subnets. If the disconnection logic is arranged inside the MARS-IF, it does not affect the CPU section.
■ Embodiment of the "node" function
First, in the following description, [2-3] and the like described at the end of the sentence are described in [Summary of Prior Art] and [Disclosure of the Invention], respectively, in which the functions of the nodes are individually described. It is clear that this is the number at the beginning of the item and indicates the related item.
FIG. 7 shows an embodiment of connection between a subnet and a wide area transmission line. Nodes 34, 35, 36, 37, 38 are connected point-to-point between terminators 60 and 50. [1-1] A subnet is formed by the transmission lines 23, 24, 25, 26, 27, 28, the nodes 34 to 38, and the terminators 60, 50, and the wide area transmission line 14 is laid from the node 35. In addition, a lower layer transmission line is connected to each node as shown in FIG.
FIG. 8 is an explanatory diagram of a mechanism in which the flow of data is expressed with the passage of time at the nodes and the terminations in FIG. Time flows from the upper side to the lower side in FIG. That is, the time axis is indicated by the arrow shown on the left side of FIG. The scales of T0 and T1 are written on the time axis. Reference numerals 60, 34, 35, 36, 37, 38, and 50 surrounded by broken lines show the reservation status of tokens and data inside the nodes and terminators shown in FIG. 7, respectively. The token is shown as a horizontally long small rectangle, and the reference number of the token is 41. The data is shown as a vertical rectangle and the reference numbers are 42 and 43. 35_36A, 35_34, and the like written in a rectangle indicating data indicate “destination_transmission source” and distinguish data. Regarding the data of the nodes 34 to 38 surrounded by the broken line, the data on the left side in the broken line, for example, 35_38Y in 38, 35_37 in 37, 35_36A in 36, etc. are data transmitted from the transmission line terminal on the left side. . Data on the right side in the broken line, for example, 35_34 in 34 is data transmitted from the transmission line terminal on the right side. The data in the center of the broken line, for example, 35_38Y in 37, 35_38Y in 36, 35_34 in 35, etc. are data temporarily stored in the FIFO memory.
As described above, the preparation for describing the node function for realizing the data multiplexing in the wide area transmission path 14 of the present invention is completed.
Further, in FIG. 8, t0, t1, t2, t3, and t4 written between the adjacent tokens 41 denote the numbers of the control data added to the tokens flowing on the transmission path constituting the subnet. It was done. The flow control data processing function for avoiding the congestion state according to the present invention by combining with [+1], [0], and [-1] in the nodes 34, 36, 37, and 38 Ready to explain.
In FIG. 8, it will be described that if the nodes 35 to 38 satisfy the functions of [2-1] to [2-3], the data multiplexed on the wide-area transmission path 14 is obtained.
◆ Embodiment of node function in subnet
In FIG. 8, at time T0, the nodes 34, 36 and 38 have a desire to transmit data 35_34, 35_36A and 35_36B, 35_38Y to the node 35 having the connection terminal of the wide area transmission line 14, and further, at time T1, It is assumed that 37 and 38 have a desire to transmit data 35_37, 35_38Z to the node 35 having the wide area transmission line connection terminal. The details will be described below with the passage of time.
At time T0, if the token is in the terminator 60, the token is sent to the node 34 through the transmission path 23.
Node 34 receives the token from the left neighbor. Since the data to be sent is 35_34, data cannot be sent to the right. [2-2]
The token is sent to the node 35 on the right. [2-1]
The node 35 receives the token coming from the left side. Since there is no data to send, the token is sent to the node 36 on the right. [2-1]
Node 36 receives the token from the left neighbor. Since the data to be sent is 35_36A, the data 35_36A is sent to the left. [2-2]
Further, the token is transmitted to the node 37 on the right side. [2-1]
Note: Data transmission to the left and token transmission to the right are performed at the same time.
The node 37 receives the token from the left neighbor. Since there is no data to send, the token is sent to the node 38 on the right. [2-1]
At the same time, the node 35 starts receiving data 35_36A from the right side. The received data is stored in the FIFO memory and sent to the node 34 on the left. [2-3]
Node 38 receives the token from the left neighbor. Since the data to be transmitted is 35_38Y, the data 35_38Y is transmitted to the left. [2-2]
At the same time, the token is sent to the terminator 50 on the right. [2-1]
The terminator 50 receives the token from the left neighbor and sends the token to the left neighbor.
At the same time, the node 37 starts receiving data 35_38Y from the right side. The received data is stored in the FIFO memory and sent to the node 36 on the left. [2-3]
Node 38 receives the token from the right neighbor. However, since the data 35_38Y is being transmitted to the left side, the transmission token to the left side is reserved. [2-1]
At the same time, the node 36 starts receiving data 35_38Y from the right side. The received data is stored in a FIFO memory. [2-3]
Note: The self-transmission data 35_36A is being transmitted to the left side, and the data 35_38Y from the right side is also received at the same time.
After a while, the node 36 completes the transmission of the data 35_36A sent to the left. The transmission of the transfer data 35_38Y stored in the FIFO memory is started. [2-3]
Remark: The FIFO memory has a capacity enough to store all the transfer data 35_38Y. This is because there is a time at which 35_38Y is transmitted from the left while 35_38Y is received from the right.
After receiving all the data of 35_36A from the right side, the node 35 starts receiving the transfer data of 35_38Y from the right side, stores it in the FIFO memory, and sends it to the node 34 on the left side. [2-3]
Since the data is addressed to itself, it can be transmitted to the wide area transmission path 14.
Note: The capacity of the FIFO memory of the node 35 may be very small. This is because while receiving 35_36A and 35_38Y from the right side, it is transferred from the left side almost at the same time while receiving. Also, if it is determined that 35_36A or 35_38Y is data addressed to itself, it can be transmitted to the wide area transmission line 14 in FIG.
After the completion of the transmission of the transmission data 35_38Y to the left side, the node 38 holding the token transmits the token to the left side. [2-1]
After receiving all the data of 35_38Y from the right side, the node 37 receives the token from the right side. After transferring all the data of 35_38Y to the node 36 on the left, the token is transmitted to the node 36 on the left. [2-1, 2-3]
After receiving all the data of 35_38Y from the right side, the node 36 receives the token from the right side. After transferring all the data of 35_38Y to the node 36 on the left, the token is transmitted to the node 36 on the left. [2-1, 2-3]
After receiving all the data of 35_38Y from the right side, the node 35 receives the token from the right side. After transferring all the data of 35_38Y to the node 34 on the left, the token is sent to the node 34 on the left. [2-1, 2-3]
Since the data is addressed to itself, it can be transmitted to the wide area transmission path 14.
The node 34 has requested transmission of the data 35_34 rightward from the time T0. After T0, the token is received from the right side for the first time, so that the data 35_34 is transmitted to the right side. [2-2]
At the same time, the token is transmitted to the node 60 on the left. [2-1]
The terminator 60 receives the token from the right side and sends the token to the right side.
At the same time, the node 35 starts receiving data 35_34 from the left side. The received data is stored in the FIFO memory and sent to the node 36 on the right. [2-3]
Since the data is addressed to itself, it can be transferred to the wide area transmission path 14 as well.
Node 34 receives the token from the left neighbor. However, since the data 35_34 is being transmitted to the right side, the sending token to the right side is reserved. [2-1]
Since the time T1 has come, the node 37 wishes to transmit the data 35_37 and the node 38 transmits the data 35_38Z to the node 35 having the wide area transmission line connection terminal.
The node 34 holding the transmission token to the right neighbor sends the token to the right after completing the transmission of the transmission data 35_34 to the right. [2-1]
After receiving all the data of 35_34 from the left side, the node 35 receives the token from the left side. After transferring all the data of 35_34 to the node 36 on the right, the token is sent to the node 36 on the right. [2-1, 2-3]
Since the data is addressed to itself, it can be transmitted to the wide area transmission path 14.
Node 36 receives the token from the left neighbor. Since the data to be sent is 35_36B, the data 35_36B is sent to the left side. [2-2]
At the same time, the token is transmitted to the node 37 on the right. [2-1]
Remark: Generally, transmission data is packetized. Packetized data groups are 35_36A and 35_36B.
The node 37 has requested the transmission of the data 35_37 to the left from time T1. After T1, for the first time, a token is received from the left neighbor, so data 35_37 is sent to the left neighbor. [2-2]
At the same time, the token is transmitted to the node 38 on the right. [2-1]
At the same time, the node 35 starts receiving data 35_36B from the right side. The received data is stored in the FIFO memory and sent to the node 34 on the left. [2-3]
Since the data is addressed to itself, it can be transmitted to the wide area transmission path 14.
Node 38 receives the token from the left neighbor. Since the data to be sent is 35_38Z, the data 35_38Z is transmitted to the left. [2-2]
At the same time, the token is sent to the node 35 on the right. [2-1]
The terminator 50 receives the token from the left neighbor and sends the token to the left neighbor.
At the same time, the node 37 starts receiving data 35_38Z from the right side. The received data is stored in a FIFO memory. [2-3]
Remarks: The self transmission data 35_37 is being transmitted to the left side, and the data 35_38Z from the right side is received at the same time.
Node 38 receives the token from the right neighbor. However, since the data 35_38Z is being transmitted to the left side, the transmission token to the left side is reserved. [2-1]
After a while, the node 36 completes the transmission of the data 35_36B sent to the left. The transmission of the transfer data 35_37 stored in the FIFO memory is started. [2-3]
Remark: The FIFO memory has a capacity enough to store all the transfer data 35_37. This is because there is a time at which 35_37 is being transmitted from the left while 35_37 is being received from the right.
After a while, the node 37 completes the transmission of the data 35_37 sent to the left. The transmission of the transfer data 35_38Z stored in the FIFO memory is started. [2-3]
After receiving all the data of 35_36B from the right side, the node 35 starts receiving the transfer data of 35_37 from the right side, stores it in the FIFO memory, and sends it to the node 34 on the left side. [2-3]
Since the data is addressed to itself, it can be transmitted to the wide area transmission path 14.
The node 38 that has reserved the token transmits the token to the left side after the transmission of the transmission data 35_38Z to the left side is completed. [2-1]
After receiving all the data of 35_38Z from the right side, the node 37 receives the token from the right side. After transferring all the data of 35_38Z to the node 36 on the left, the token is transmitted to the node 36 on the left. [2-1, 2-3]
The node 36 starts receiving 35_38Z from the right side after receiving all the data of 35_37 from the right side, and receives the token from the right side after receiving all. After transferring all the data of 35_37 and 35_38Z to the node 35 on the left, the token is transmitted to the node 35 on the left. [2-1, 2-3]
The node 35 starts receiving 35_37 from the right side after receiving all the data of 35_36B from the right side, and starts receiving 35_38Z from the right side after receiving all. Receive. The token is sent to the node 35 on the left. [2-1, 2-3]
Since the data is addressed to itself, it can be transferred to the wide area transmission path 14 as well.
Remarks: The FIFO memories of all nodes have a capacity enough to store one packet of transfer data. This is because there is a time when 35_36B is transmitted from the left side while receiving 35_36B from the right side, and similarly, 35_37 is transmitted from the left side while receiving 35_37 from the right side, and 35_38Z is received from the right side. This is because 35_38Z is transmitted from the left side inside.
Node 34 receives the token from the right neighbor. Since there is no data to send, the token is sent to the terminator 60 on the left. [2-1]
The terminator 60 receives the token from the right side.
As described above, if the node performs the functions described in [2-1] [2-2] [2-3], the data 35_36A and 35_38Y, the token passage gap, the data 35_34, It will be appreciated that token multiplexing, data 35_36B, 35_37, 35_38Z, token passing variance, and data multiplexing are implemented.
If a large number of nodes are connected to the subnet and it is desired to transmit data to the nodes 35 having the wide area transmission path terminals at the same time, the data stream becomes almost continuous, and the token passing time ratio decreases. This means that the transmission efficiency of the wide area transmission path 14 approaches 100%.
However, if voice data is transmitted and received over the wide area transmission line 14, if all nodes desire data transmission at the same time, the voice may be interrupted when demodulating the data distributed from the connected switching device and converting it to voice. Happens. Therefore, in a data multiplexing device used for a telephone line or the like, the maximum line capacity is determined.
The maximum line capacity in the data multiplexing mechanism based on subnets of the present invention depends on the number of packets transmitted or received in one cycle time when a token passes through its own node. That is, in the example of FIG. 8, the node 35 is “4”, and the other nodes are “1”. This is because the node 35 receives four packets 35_34, 35_36B, 35_37, and 35_38Z during the cycle of the token that has arrived from the right neighbor again. This is because, for the other nodes, only one packet is transmitted during the cycle of the token.
◆ Embodiment of node function of data processing to avoid congestion state by flow control
In FIG. 8, at time T0, the nodes 34, 36 and 38 have a desire to transmit data 35_34, 35_36A and 35_36B, 35_38Y to the node 35 having the connection terminal of the wide area transmission line 14, and further, at time T1, It is assumed that 37 and 38 have a desire to transmit data 35_37, 35_38Z to the node 35 having the wide area transmission line connection terminal. In the following, the processing of the control numeral added to the token for each node will be described in detail over time.
The transition of the control numeral of the node 34 will be described with the number over time.
1. At time T0, transmission of self-transmission data is desired. [+1] is described and prepared as a control number to be added to the token. [3-1]
This procedure is performed only once before transmitting a series of self-transmission data 35_34.
2. Since the token t0 comes from the left side, [+1] is added to t0 to be t1, and [+1] is rewritten to [0]. The token t1 is transmitted to the right. [3-1]
This procedure is a program for changing a control numeral added to a token. However, since it is a program, the numbers do not always change as described below.
3. Since the token t3 comes from the right side, [0] is added to t3 to be t3, and [0] is rewritten to [0]. The data 35_34 is transmitted to the right and the token t3 is transmitted to the left. [3-1, 2-2]
4. Since the token t3 comes from the left side, [0] is added to t3 to be t3, and [0] is rewritten to [0]. Since the data 35_34 is being transmitted to the right side, the token t3 is reserved, and after transmitting the data 35_34, the token t3 is transmitted to the right side. [2-1, 3-1]
5. After the transmission of the data 35_34, [0] is rewritten to the reduced number [−1] in order to reduce the control number, in preparation for the next token.
This procedure is performed only once after transmitting a series of self-transmission data 35_34.
6. Since the token t2 comes from the right side, [-1] is added to t2 to be t1, and [-1] is rewritten to [0]. The token t1 is sent to the left. [3-1]
The transition of the control numeral of the node 35 will be described with the number over time.
1. At time T0, transmission of self-transmission data is not desired. [0] is described and prepared as a control number to be added to the token in the initial setting. [3-1]
2. When a token comes from any one transmission line connection terminal, [0] is added to the subscript of the token, [0] is rewritten to [0], and a token signal is transmitted from the other transmission line terminal. [2-1, 3-1]
The transition of the control numbers of the node 36 will be described with numbers over time.
1. At time T0, transmission of self-transmission data is desired. [+1] is described and prepared as a control number to be added to the token. [3-1]
This procedure is performed only once before transmitting a series of self-transmission data 35_36A and 35_36B.
2. Since the token t1 comes from the left side, [+1] is added to t1 to be t2, and [+1] is rewritten to [0]. The data 35_36A is transmitted to the left and the token t2 is transmitted to the right. [3-1, 2-2]
3. Since the token t3 comes from the right side, [0] is added to t3 to be t3, and [0] is rewritten to [0]. The token t3 is sent to the left. [3-1, 2-1]
4. Since the token t3 comes from the left side, [0] is added to t3 to be t3, and [0] is rewritten to [0]. The data 35_36B is transmitted to the left, and the token t3 is transmitted to the right. [3-1, 2-2]
5. After the transmission of the series of packet data 35_36B, [0] is rewritten to the reduced number [−1] in order to reduce the control number, in preparation for the next token. This procedure is performed only once after transmitting a series of self-transmitted packet data 35_36A and 35_36B.
6. Since the token t3 comes from the right side, [-1] is added to t3 to be t2, and [-1] is rewritten to [0]. The token t2 is transmitted to the left. [3-1, 2-1]
The transition of the control numeral of the node 37 will be described with time.
1. At time T0, transmission of self-transmission data is not desired. [0] is described and prepared as a control number to be added to the token in the initial setting. [3-1]
2. When a token comes from any one transmission line connection terminal, [0] is added to the subscript of the token, [0] is rewritten to [0], and a token signal is transmitted from the other transmission line terminal. [2-1, 3-1]
3. At time T1, it is desired to transmit self-transmission data. As control numbers to be added to the token, [0] is increased and [+1] is described and prepared. [3-1]
This procedure is performed only once before transmitting a series of self-transmission data 35_37.
4. Since the token t3 comes from the left side, [+1] is added to t3 to be t4.
If the value to avoid the congestion state is “4”,
As described above, [+1] is rewritten to [0]. The data 35_37 is transmitted to the left and the token t4 is transmitted to the right. [3-1, 2-2]
If the value to avoid the congestion state is “3”,
The token to be transmitted to the right is t3. Therefore, the control number added to the token is not changed by [+1].
Since the control digit is not [0], even if a token comes from the left side, the data 35_37 cannot be sent to the left side. [3-2]
5. If the value to avoid the congestion state is “4”,
Since the state is described in FIG. 8, after the data 35_37 is transmitted, [0] is rewritten to the decreased number [−1] in order to decrease the control number, and the next token is prepared.
This procedure is performed only once after transmitting a series of self-transmission data.
If the value to avoid the congestion state is “3”,
Since the data 35_37 is not transmitted in a state not described in FIG. 8, the control numeral holds [+1].
6. Token came from right next.
If the value to avoid the congestion state is “4”,
The token coming from the right side is t4, and [-1] is added to t4 to be t3, and [-1] is rewritten to [0]. Sends the token t3 to the left side [3-1, 2-1]
If the value to avoid the congestion state is “3”,
The token coming from the right side is t3, and when [+1] is added to t3, it becomes t4, so the control number added to the token is not changed by [+1]. Therefore, the token t3 is transmitted to the left. [3-2, 2-1]
The transition of the control numeral of the node 38 will be described with time.
1. At time T0, transmission of self-transmission data is desired. [+1] is described and prepared as a control number to be added to the token. [3-1]
This procedure is performed only once before transmitting a series of self-transmission data 35_38Y. If the self-transmitted data 35_38Y and 35_38Z are a series of data, the process is the same as the transition of the control number of the node 36. However, 35_38Y and 35_38Z are not a series of data but separate data into separate packets.
2. Since the token t2 comes from the left side, [+1] is added to t2 to be t3, and [+1] is rewritten to [0]. The data 35_38Y is sent to the left and the token t3 is sent to the right. [3-1, 2-2]
3. Since the token t3 comes from the right side, [0] is added to t3 to be t3, and [0] is rewritten to [0]. Since the data 35_38Y is being transmitted to the left side, the token t3 is reserved, and after transmitting the data 35_38Y, the token t3 is transmitted to the left side. [2-1, 3-1]
4. After the transmission of the self-transmission data 35_38Y, [0] is rewritten to the reduced number [-1] in order to reduce the control number, in preparation for the next token.
5. At time T1, transmission of the next self-transmission data 35_38Z is desired. As a control number to be added to the token, [-1] is increased and described as [0] and prepared. [3-1]
This procedure is performed only once before transmitting a series of self-transmission data 35_38Z.
6. If the value to avoid the congestion state is “4”,
Since the token t4 comes from the left side, [0] is added to t4 to be t4, and [0] is rewritten to [0]. The new data 35_38Z is transmitted to the left and the token t4 is transmitted to the right. [3-1, 2-2]
If the value to avoid the congestion state is “3”,
Since the token t3 comes from the left side, [0] is added to t3 to be t3, and [0] is rewritten to [0]. The new data 35_38Z is transmitted to the left and the token t4 is transmitted to the right. [3-1, 2-2]
When the value for avoiding the congestion state is “3”, the node 37 cannot transmit the data 35_37. However, the node 38 can transmit new data 35_38Z. This difference is due to the desire to send new data within the time that the token passing through the node 38 makes one round, so that the distinction between the previous data 35_38Y and the current data 35_38Z is performed in the processing within the node 38, This is because there was no opportunity to cause a change in the control number added to the token. If the token has made one round or more, when the value for avoiding the congestion state is “3”, 35_38Z cannot be transmitted.
7. Since the token t4 (or t3) comes from the right side, [0] is added to t4 (or t3) to be t4 (or t3), and [0] is rewritten to [0]. Since the data 35_38Z is being transmitted to the left side, the token t4 (or t3) is reserved, and after transmitting the data 35_38Z, it is transmitted to the left side. [3-1, 2-1]
8. After transmitting the series of self-transmission data 35_38Z, [0] is rewritten to the reduced number [−1] in order to reduce the control number, in preparation for the next token.
This procedure is performed only once after transmitting a series of self-transmission data 35_38Z.
As described above, it will be understood that the mechanism of [3-1] [3-2] for avoiding the congestion state can be realized by the procedure processing (program) of each node.
■ Embodiment of MARS-IF processing
The MARS-IF performs processing according to an ACM (Access Control Message) bit added to the head of transmission data transmitted from the CPU unit through the MARS-IF. The ACM is divided into a plurality of routing bits and a plurality of transmission data property bits. As described in FIGS. 5 and 6, the first two bits are used for routing, and the next four bits are used as transmission data property bits. As a preparation for explanation, FIGS. 9 and 10 will be described first.
FIG. 9 is an enlarged view of the connection near the node 36 in FIG. In the nodes 35, 36, and 37 in FIG. 9, the left adjacent terminal is X, and the right adjacent terminal is Y. Therefore, the terminal Y on the right side of the node 35 is connected to the terminal X on the left side of the node 36 via the transmission line 25. The terminal Y on the right side of the node 36 should be connected to the terminal X on the left side of the node 37 via the transmission line 26, but is erroneously connected to the terminal Y on the right side of the node 37 via the transmission line 26. The terminal X for the left side was connected to the transmission line 27 to the right side.
FIG. 10 is an explanatory diagram expressing the flow of data over time in the nodes and terminators in FIG. The time flows from the upper side to the lower side in FIG. That is, the time axis is indicated by the arrow shown on the left side of FIG. T2 and T3 scales are written on the time axis. The tokens are indicated by small horizontal rectangles, and the movement of the tokens on the transmission path is indicated by arrows. The data is indicated by a vertically long rectangle, and the movement of the data on the transmission path is indicated by a thick arrow. Reference numerals 60, 35, 36, 37, and 50 surrounded by broken lines indicate the reservation states of the transmission line terminals X and Y, the token, and the data inside the node and the terminator shown in FIG. 9, respectively. Regarding the transmission line connection terminals X and Y, in FIG. 10, 35 and 36, X is on the left and Y is on the right. However, for 37, Y is on the left and X is on the right, and vice versa.
In FIG. 10, the numbers “01” on the right and center of 35 and on the left of 37, “10” on the left of 36, and “11” on the center of 36 and 37 indicate the routing bits. Note that “11” and “01” described in the center of the broken line frame are indicated by routing bits added at the time of data transmission, and “1” and “10” described at the right or left side are routing bits added at the time of data reception. The bits are shown to make it easier to understand.
◆ Embodiment of bit processing for routing
The formats 51, 52, and 53 in FIG. 5 are exactly the same except that the head is different from “11”, “nn”, and “01”. Reference numeral 51 denotes a routing state described when the CPU unit transmits data to its own MARS-IF. Reference numeral 52 denotes data flowing between the MARS-IF, that is, the data flowing through the transmission paths 24-27 in FIG. Reference numeral 53 denotes data transferred by the MARS-IF to its own CPU unit, that is, data received by the CPU unit. Since the routing bits are two bits, four states occur. According to the four states, the first table is used as routing bits for the CPU unit 51 to indicate the transmission terminal to the MARS-IF, and the second table is used as the routing bits for the MARS-IF 53 indicating the reception terminal in the CPU unit. Raise.
In FIG. 8, the node 34 sends out its own transmission data 35_34 to the right, and the nodes 36, 37, 38 send out its own transmission data 35_36A, 35_37, 35_38Y, etc. to the left. Whether to transmit the self-transmission data to the right or to the left is treated as what each node already knows. Of course, as in the prior art, the data transmission direction can be determined by examining the data transmission direction in advance, creating a routing table, and referring to the table when transmitting data. The routing of the present invention will be described with reference to examples of bit transitions 51, 52, and 53.
At time T2 in FIG. 10, it is assumed that the terminator 60 sends a token toward the node 35, and the nodes 36 and 37 have a desire to transmit to the node 35. However, it is assumed that the nodes 36 and 37 do not know the status of the transmission line connection terminal.
[Table 1] Routing bit for CPU unit to indicate transmission terminal to MARS-IF

[Table 2] Routing bits for MARS-IF indicating receiving terminal in CPU

At time T2, the CPU unit of the node 36 prepares to transmit the data 35_36S. The first two bits of 35_36S are “11”. [Table 1]
Note: Since the node 36 does not know the connection direction of the node 35, it transmits data to all connection terminals as "11".
After a while, the MARS-IF of the node 36 receives the token from the terminal X.
According to the instruction from the CPU unit, 35_36S is transmitted from the X and Y terminals. [1-4, Tables 1, 2-2]
Remarks: The CPU unit 36 sends the data “11 + 35_36S” to the MARS-IF 36, and the MARS-IF 36 converts the data “11 + 35_36S” into “nn + 35_36S” and sends out the data to the transmission lines 25 and 26. Therefore, “11” created by the CPU unit 36 is deleted from the data flowing through the transmission paths 25 and 26. [1-4] is a node function of the prior art. In the MARS-IF function, "11" of the first two bits corresponds to "designation" described in [2-2].
The nodes 35 and 37 have started receiving the data 35_36S.
Since the MARS-IF 35 determines that the data 35_36S received from the terminal Y is addressed to itself, the MARS-IF 35 transmits “01 + 35_36S” to the CPU unit 35. [Table 2]
At the time T2, the node 37 prepares for transmission by setting the first two bits of the first data 35_37S to “11”. [Table 1]
However, since data 35_36S is received instead of the token, 35_36S is transmitted as transfer data from the terminal X to the transmission path 27. [1-3, 2-3]
Since the transmission of the first data 35_36S has been completed, the node 36 sends out the token from the terminal Y. [1-2, 2-1]
The node 37 has received the token from the terminal Y. According to the instruction from the CPU unit, 35_37S is transmitted from the X and Y terminals. [1-4, Tables 1, 2-2]
The node 36 has started receiving the data 35_37S. Since it is not data addressed to itself, it is transmitted from terminal X to transmission line 25 as transfer data. [1-3, 2-3]
Since the MARS-IF 35 determines that the data 35_37S received from the terminal Y is addressed to itself, the MARS-IF 35 transmits “01 + 35_37S” to the CPU unit 35. [Table 2]
Since the transmission of the first data 35_37S has been completed, the node 37 sends the token from the terminal X. [1-2, 2-1]
The CPU unit 35 has completed the reception of the data 35_36S with “01.” As the reply data, “01 + 36_35R” is created.
Subsequently, the reception of “01 + 35_37S” is completed. "01 + 37_35R" was created as the reply data.
Remark: Since the data receiving terminal indicates the connection direction of the transmission source node, the reception routing bit state is set as the return routing bit.
The MARS-IF 35 receives “01 + 36_35R” and “01 + 37_35R” from the CPU unit 35 and knows that the data is to be transmitted to the terminal Y. It waits until a token comes from terminal Y. [Table 1, 2-2]
Note: In the example of FIG. 8, the node 35 is “4” and the other nodes are “1” as the maximum line capacity in the data multiplexing mechanism using subnets. Here, it is assumed that two different packets of data 36_35R and 37_35R are to be continuously transmitted, and that the number to be added as the token control data is set to [+2] and the process stands by. [3-1]
The node 35 continuously sends data 36_35R and 37_35R since the token comes from the terminal Y. [Table 1, 2-2]
The MARS-IF 36 has started receiving the self-addressed data 36_35R. Since it has been received from the terminal X, "10 + 36_35R" is transmitted to the CPU section 36. [Table 2]
The MARS-IF 37 has started receiving the self-addressed data 37_35R. Since it has been received from the terminal Y, “01 + 37 — 35R” is transmitted to the CPU unit 37. [Table 2]
As described above, routing is performed by performing the processing of Tables 1 and 2. Therefore, return routing can be ensured only by referring to the received data without creating and referring to the routing table.
The routing bits received from the own MARS-IF by the CPU units of the nodes 36 and 37 are different from "10" and "01". This is because the transmission line on the right and the transmission line on the left are connected in reverse in the transmission line connection of the node 37 in FIG. However, in any case, the node that has transmitted the data is located at the terminal that has received the data, so that even if the transmission paths are connected in reverse, the routing is performed correctly.
◆ Embodiment of data property bit processing
According to the Disclosure of the Invention, it was stated that four types of data properties, token use status at the time of transmission, and network topology can be investigated. These data properties are illustrated and summarized in Table 3 as data property bits. Although four bits are used as the data property bits in Table 3, the bits described in "x" can be freely defined as "0" or "1". Note that, like “x”, “y” in FIGS. 5 and 6 can be freely defined as “0” or “1”.
[Disclosure of the Invention] In the MARS-IF process, a terminal node having one terminal in FIG. 3 is a two-terminal bus node such as the nodes 35, 36, and 37 in FIG. 3 or the node 33 in FIG. Since the data processing of the terminator and the branching device is not clearly shown, the processing for data reception will be described together according to the classification in Table 3.
[Table 3] Correspondence between data property bits and data properties

The branching unit transmits the data to all the transmission lines except for the transmission line from which the data arrives, except for the reception data of the data property bit “1100”.
“00xx”: Reception of data for broadcast channel
Two-terminal bus node: receives all data and sends it out to another transmission path.
Terminal node: receives all data and does not transmit.
Terminator: Not sent.
“01xx”: Reception of broadcast channel data with destination
Two-terminal bus node: receives data addressed to itself and sends it out to other transmission paths.
Terminal node: Receives data addressed to itself and does not send it out to other transmission paths.
Terminator: Not send all.
“10xx”: Reception of data specifying destination
2-terminal bus node: receives data addressed to itself,
If the data is not addressed to itself, it is transmitted to another transmission path.
Terminal node: Receives data addressed to itself and does not send it out to other transmission paths.
Terminator: Not send all.
“1100”: Receiving topology investigation data using token
Two-terminal bus node: All signals are received, and the CPU determines transmission.
Terminal node: All are received, and the CPU determines transmission. .
Branching device: Transmits to one predetermined transmission line.
Terminator: Only a code indicating a data property bit of “1100”, or
All the received data is transmitted when the token arrives.
The token is transmitted after transmitting the code or the data.
“1101”: data addressed to an adjacent node using the token, and
“111x”: Reception of data addressed to an adjacent node ignoring the token
Two-terminal bus node: receives all and does not transmit to other transmission paths.
Terminal node: Receives all and does not transmit to other transmission paths.
Terminator: Not send all.
The above is the processing of the members that construct the received subnet based on the data property bits. The best embodiment will be described for data of data property bits “1101” and “111x”.
The data multiplexing network mechanism of the present invention uses a network having a token arbitration mechanism as a prior art. If the token disappears, the system will go down and will not function as a network. Usually, the disappearance of the token occurs when a failure occurs in a part of a network construction member or a transmission path, or when the token signal changes due to noise during transmission of the token signal. Thus, even though the token has expired, most building components are normal, and sometimes all are normal. Therefore, if the network has a communication mode that does not require a token to check for a network failure, the failed network can be used to find the cause of the failure. It should be noted that data of the data property bit "111x" can be transmitted when the token is circulating normally, but data competition occurs and the network system goes down.
Hereinafter, it is assumed that the token is circulating normally. Therefore, the description will be given except for a communication mode for investigating a fault location for transmitting data of the data property bit “111x”.
“1100” and “1101” classified by the data property bits have an adjacent node as a destination. The form of communication to an adjacent node is a basic form for a point-to-point channel, and may be RS-232-C (CCITT recommendation V2.4) which is generally well known. Although the data multiplexing network mechanism of the present invention uses the prior art to change from a point-to-point channel to a broadcast channel, the basis of the channel configuration is a point-to-point channel. Therefore, when using a point-to-point channel, the destination of the node is not necessarily required. The invention part of the data property bit processing is a method for investigating the subnet topology using a point-to-point channel. Therefore, an embodiment relating to the processing of the data property bit “1100” will be described in detail.
In the reception processing of "1100" in accordance with the classification in Table 3 described above, it is described that the CPU unit determines whether to transmit data for a node. Table 4 shows the transmission judgment criteria of the CPU unit. Note that the branch unit and the terminator do not originally require a CPU unit, so Table 4 is not applied.
In the case where the subnet is as shown in FIG. 1, an example in which the node 32 desires a topology check using the data property bit “1100” will be described along time.
The node 32 sets the investigation node number of the frame 61 to “32”, writes an appropriate binary number in “yy”, sets the number to “011100yy / 32”, and sends it to the MARS-IF32. [Table 4-1]
Remarks: The investigation node number of the node 32 is “32”. If the node number is undetermined, “32” may be changed to an unused node number such as “00”. It suffices to be able to distinguish nodes that have requested a connection survey.
The MARS-IF 32 sends “nn1100yy / 32” from the terminal Y of the node 32 because the routing bit is “01”. [Table 1]
Note: The transmission path connected to the terminal Y of the node 32 in FIG. 1 may be connected to the node 31 or to the branching device 70, but the terminal Y of the node 32 is connected to the node 31. Suppose it had been.
Table 4 Processing of CPU when data property bit is "1100"

The MARS-IF 31 receives “nn1100yy / 32” and sends “011100yy / 32” to the CPU unit 31. [Table 2]
Note: The routing bit received by the CPU unit 31 from the MARS-IF 31 is “10” when received from the terminal X and “01” when received from the terminal Y. It is assumed that the signal is received from the terminal Y. [Table 2]
The CPU unit 31 has received the data “011100yy / 32”. The routing bit is rotated to “101100yy / 32”. [Table 4-3]
Since the initial transmission of the topology investigation has not been performed, the CPU unit 31 passes “101100yy / 32/31” as the value of the frame 62 to the MARS-IF 31 as transmission data. [Table 4-4]
Note: If the bit string "100" is rotated right, it becomes "010", and if it is rotated left, it becomes "001". However, if the two bits are “01” or “10”, both the right rotate and the left rotate have the same value, which is the same as the inverted two bits. [Table 4-3]
The MARS-IF 31 transmits “nn1100yy / 32/31” to the terminator 50 from the terminal X. [Table 1]
The terminator 50 has received “nn1100yy / 32/31”. “Nn1100yy” is transmitted.
Remarks: When receiving "1100" in accordance with the classification in Table 3 above, the terminal sends a code indicating the data property bit of "1100" or all received data, so only a simple code Was sent.
The MARS-IF 31 receives “nn1100yy” from the terminal X and sends “101100yy” to the CPU unit 31. [Table 2]
Remarks: The connection terminal of the CPU section 31 has the terminal Y on the node 32 side. Therefore, reception from the terminator 50 is reception from the terminal X.
The CPU unit 31 has received the data of “101100yy”. Since only the code is received, the previously transmitted data “101100yy / 32/31” is regarded as received. [Table 4-2]
Since the initial transmission of the topology investigation is not performed, the CPU unit 31 passes “011100yy / 32/31/31” as the value of the frame 63 to the MARS-IF 31 as transmission data. [Table 4-3, Table 4-4]
The MARS-IF 31 transmits “nn1100yy / 32/31/31” from the terminal Y to the node 32. [Table 1]
The MARS-IF 32 receives “nn1100yy / 32/31/31” from the terminal Y and sends “011100yy / 32/31/31” to the CPU unit 32. [Table 2]
The CPU unit 32 receives the data “011100yy / 32/31/31” and rotates the routing bits to “101100yy / 32/31/31”. [Table 4-2]
The CPU unit 32 sets the initial transmission of the topology investigation to “011100yy / 32”. When a match is determined for the ACM part, the leading routing bit is different, so that the data “101100yy / 32/31/31/32” is transmitted as the value of the frame 63. [Table 4-4]
The MARS-IF 31 transmits “nn1100yy / 32/31/31/32” to the branching unit 70 from the terminal X. [Table 1]
The branching device 70 has received “nn1100yy / 32/31/31/32”.
If it is general data, the data received by the branching device is transmitted from all other terminals. However, since the data property bit is "1100", the data is cyclically transferred like a token and transmitted to the node 33.
Remarks: It is stated that, upon receiving "1100" data summarized according to the classification in Table 3 above, as a treatment of the branching device, the data is transmitted to one predetermined transmission line. Originally, the branching device previously determined the transmission destination of the token signal. Of course, the sending terminal corresponding to the receiving terminal of the “1100” data may be different from the sending destination of the token signal, but they are the same.
The node 33 receives “?? 1100yy / 32/31/31/32” from the branching unit 70 and sends out the transmission data to the terminal node 29 as “nn1100yy / 32/31/31/32/33”. [Table 1, Table 4-4]
Note: Although the routing bit is described as “??” in 63 of FIG. 6 and above, it is actually “01” or “10”.
The terminal node 29 receives “?? 1100yy / 32/31/31/32/33” from the node 33 and sets the transmission data as “nn1100yy / 32/31/31/32/33/29” to the node 33. Send out. [Table 1, Table 4-4]
Hereinafter, similar processing is performed in the node 33, the branching device 70, the node 35, the node 34, the terminal device 60, the node 34, the node 35, and the branching device 70. The data of the frame 63 in which the routing bit of the data received by the CPU unit 32 from the MARS-IF 32 is rotated in [Table 4-2] is “011100yy / 32/31/31/32/33/29/33/35”. / 34/34/35 ". The node 32 sets the initial transmission of the topology investigation to “011100yy / 32”. When the match is determined for the ACM unit, since they are the same, the subnet connection status is obtained by analyzing the received data. Data to be analyzed excludes codes. [Table 4-4]
Therefore, the data string of “32/31/31/32/33/29/33/35/34/34/35” expresses the connection status of the subnet centering on the node 32. An analysis method of this data string will be briefly described.
1) If the same number follows, the neighbor is a "terminator".
Example: "31/31", "34/34" are continued.
Therefore, “31” and “34” are in contact with “terminator”.
2) If there is only one number in the connection data string, it is "terminal node".
Example: Since there is only one "29", it is "terminal node 29".
3) Straight connection with left and right target numbers centering on the terminator and terminal node.
Example: From "32/31 (terminal) 31/32", {Node 32} Node 31} Terminator.
From “33 / terminal node 29/33”, {node 33} terminal node 29.
From “35/34 (terminal) 34/35”, {Node 35} Node 34} Terminator.
4) Considering the straight-line connection obtained in 3), the investigation start node, and the branching device connection, the investigation start node can obtain the subnet topology.
The description of the best mode based on the individual claims is completed above, and the best mode based on the claims is shown and described.
FIG. 11 is an explanatory diagram of the mechanism expressed with the passage of time, similarly to FIGS. 8 and 10. Time flows from the upper side to the lower side in FIG. 11, and is indicated by an arrow shown on the left side in FIG. 11 as a time axis. Points different from FIG. 8 and FIG. 11 will be described.
The terminator 60 was changed to the terminal node 29. Nodes 31, 32, 33 and 39 have been added. "Destination_source" is described only for the node 35 in order to make FIG. 11 fit on one sheet of paper. The data of the reference numeral 44 is transmitted and received between the nodes 31 and 33. "X" is written to represent data 31_33 or 33_31 and to clearly indicate them. The data of the reference numeral 45 represents the data 32_37 transmitted and received between the nodes 37 and 32, and “/” is written to clearly show the data 32_37 and 37_32.
In the node 35 of FIG. 8, the data 35_36A, 35_38Y transmitted to the left, the data 35_34 transmitted to the right, and the data 35_36B, 35_37, 35_38Z transmitted to the left are described as "omitted". Further, the data described in FIG. 10 is also described as “omitted”. However, since the destination is indicated, if the data property bit is set to “10xx (data specifying the destination)”, the reception processing according to the classification in Table 3 is not “omitted” but “disappeared”. ", And is deleted in FIG.
The data 44 in which “X” is written in FIG. 11 simply passes through the node 32 between the nodes 33 and 31. The data 45 in which “/” is written simply passes through the nodes 36, 35, 34 and 33 between the nodes 37 and 32. Other data is transmitted and received to and from the node 35. Therefore, the nodes other than the destination and the source nodes simply pass through intermediate nodes, similarly to the data of “X” and “/”.
Paying attention to the transmission data of the node 35, the data 34_35, 33_35, and 69_35 are transmitted with the token signal arriving from the left 34 as a trigger, and the data 39_35, 36_35, and 38_35 are generated with the token signal arriving from the right 36 as a trigger. Then, the data 34_35, 33_35, and 69_35 were transmitted using the token signal coming from the left 34 as a trigger. The transmission time interval between the data 33_35 and 69_35 differs depending on the presence / absence of the data 45 of “/”. Of course, if the transmission time interval between the data 33_35 and 69_35 is minimized, the data 45 of “/” is transferred by the node 35 after the transmission of the data 69_35. However, as a condition for minimizing the transmission time interval between the data 33_35 and 69_35, the capacity of the FIFO memory 43 increases.
In order to minimize the capacity of the FIFO memory, it is the best embodiment to give the sending terminal priority in the following order.
1. Transfer data: data that the MARS-IF receives from the transmission line terminal
2. Transmission data: data received from the CPU subordinate to the MARS-IF
3. Token signal
Since the transmission of the token signal is the lowest priority, the transmission of the token is reserved if the transmission path terminal to be transmitted is used as described in [2-1]. The node 35 transmits six packets of 34_35, 33_35, 69_35, and 39_35, 36_35, 38_35 during one cycle of the token. This means that the object of the present invention is to realize the multiplexing in the wide area transmission line in FIG. 2 and to immediately transfer the data received from the upper wide area transmission path 14 to the lower area wide transmission path 11. is there.
If the terminal node 29 and the terminator 50 in FIG. 11 have the same value as the ring-connected node 29 shown in FIG. 4, as a measure to increase the amount of data “/” that simply passes through the node 35, the terminal node 29 Is a two-terminal bus node, and logically disconnected at the node 36 or 37, the data "/" that simply passes through the node 35 can be extremely reduced, and the transmission line utilization efficiency of the subnet can be increased.
The node was divided into a CPU section and a MARS-IF. If the congestion control portion where the processing change is large is realized by a program of the CPU unit and the MARS-IF is a small LSI, it can be realized by a partial modification of the circuit described in the prior art.
Therefore, a block diagram serving as a guideline for manufacturing an interface LSI satisfying the function of the MARS-IF will be described in [Industrial Applicability].
Industrial applicability
FIG. 12 shows that the present invention can be used in industry by causing the CPU unit to mainly handle congestion control and making the MARS-IF an interface LSI 76. Hereinafter, FIG. 12 will be described.
The CPU unit constructs a network through the upper wide area transmission path 14 and the lower wide area transmission path 11, and connects to a subnet through the interface LSI 76. Names of RT, ST, CTR, CTW, SD, and RD are given as signal transmission lines for the CPU unit and 76.
RT, ST: RT is a signal line for transmitting a token with control data from 76 to the CPU unit, and ST is a signal line for transmitting a token with control data from the CPU unit to 76.
CTR, CTW: CTR is a signal line for transmitting a control signal from 76 to the CPU unit, and CTW is a signal line for transmitting a control signal from the CPU unit to 76.
SD, RD: SD is a signal line for transmitting data to be transmitted to the subnet from the CPU unit to 76, and RD is a signal line for transmitting data obtained from the subnet from 76 to the CPU unit.
The interface LSI 76 has a transmission line connection terminal X78 and a transmission line connection terminal Y79. The internal block 76 separates the received signal from the transmission line on the X side into a token signal and a data signal 80, separates the received signal from the transmission line on the Y side into a token and data 81, and transmits to the transmission line on the X side A parallel / serial converter 82 for supplying a transmission signal, a parallel / serial converter 83 for supplying a transmission signal to the Y-side transmission path, a buffer 84 for temporarily storing a received token, and FIFO memories 85, 84 and 85 for storing received data. A controller 86 for controlling the stored signals, a buffer 87 for temporarily storing a token with control data transmitted by the CPU unit, a buffer 88 for temporarily storing data transmitted by the CPU unit, a controller 89 for controlling signal transmission between blocks, And a gate 90 for sending the token with control data in the buffer 84 to the CPU unit, A gate 91 for sending data to the CPU unit, a gate 92 for sending transfer data in the FIFO 85 to the X side, a gate 93 for sending to the Y side, and a token without control data in the buffer 84 is directly sent to the X side. A gate 94, a gate 95 for sending directly to the Y side, a gate 96 for sending the transmission data in the buffer 88 to the X side, a gate 97 for sending to the Y side, and a gate for sending the token with control data in the buffer 87 to the X side 98, and a gate 99 for sending out to the Y side. Although the control lines of the gates 90 to 99 are connected to 89, they are omitted to avoid complicated figures.
In the configuration of FIG. 12, in order to set the terminator to the X side, the “Data” output of 80 is stopped if it is data other than the code of FIG. 6, and 80 informs 89 of the arrival of the token signal. Control to open the gate. If the code 80 in FIG. 6 is notified to 89 that the code has been received, data is stored in 85, but this can be handled by passing only the first code at 92. Of course, all of the 85 data may be sent from 92. Since there is only one token signal in the subnet, when a token arrives from 79 other than the terminator, the data and the token are logically disconnected even if the transmission / reception control is performed. This is because it is apparent in FIG. 11 that data is not received from both the transmission line terminals X and Y at the same time.
In the configuration of FIG. 12, in order to set the terminator to the Y side, the “Data” output of 81 is stopped if it is data other than the code of FIG. Control opening. If the code 81 shown in FIG. 6 is notified to 89 that the code 81 has been received, the data is stored in 85 and all the stored data or only the code at the head of the data can be passed through 93 to cope with it.
In the configuration of FIG. 12, if the controller 89 controls [1-1] to [1-5] shown in [Overview of Prior Art] without controlling to open the gates 90, 98, and 99, the “preceding The "node shown by technology" can be realized. Since the FIFO 85 is a first-in first-out memory, it serves as a simple buffer in the node shown in the prior art.
In the configuration of FIG. 12, if the controller 89 controls [2-1] to [2-3] shown in [Disclosure of the Invention] without controlling the gates 90, 98 and 99 to open, the two-terminal bus node Can realize the MARS-IF function. In order to realize the MARS-IF function used for the terminal node, the function as the terminator may be prioritized in 80 or 81.
In the configuration of FIG. 12, the token with control data is sent to the CPU through the gate 90 without performing the control to open the gate 94, and the control to open the gates 98 and 99 is performed by 89, whereby the CPU [3-1] to [3-2] described in [Disclosure of the Invention], a node function for processing control data for avoiding a congestion state can be realized.
In the processing of the routing bit, to add the receiving terminal to the received data, 80 or 81 may be added to the head of the data at the time of data reception. The received data is sent to the FIFO 85.
In the processing of the routing bits, the controller 89 examines the transmission data buffer 88 and opens the gate 96 or 97 for selecting the transmission terminal and changing the data at the first two bits of the data transmitted from the CPU unit. Use to select the sending terminal. In the data change processing, when the parallel / serial converters 82 and 83 convert the data into serial data, the routing bits may be deleted.
In the reception process using the data property bits, the description in Table 3 may be applied to the gates 91, 92, and 93.
As described above, the MARS-IF can be realized by the interface LSI by controlling each gate in FIG.
A main object of the present invention is to replace a star-type host computer, which is a multiplexing device, with a subnet. However, no problem occurs even if the subnet is used as a LAN (Local Area Network) for industrial use. This is because, in the hierarchical structure shown in FIG. 2, there is no equipment corresponding to the host computer that connects the wide-area transmission path 14 in the uppermost layer in the real world. Conversely, a sensor of a personal computer, an industrial machine, or the like is connected to the lowermost wide-area transmission line. However, since the signal transmission speed cannot exceed the speed of light, the size of the subnet is limited.
Finally, a consideration is given of the distance between the nodes constituting the subnet and the transmission data rate.
Since the present invention does not use QAM, which is one of the data multiplexes shown at the beginning of [Background Art], it is considered as 1 bps / baud modulation.
In FIG. 11, the node 37 starts transmitting data 32_37,
At the same time, send the token to the adjacent node 38,
When the node 38 starts transmitting the data 35_38,
By the time the node 37 starts receiving the data 35_38,
The signal reciprocates on the transmission path from 37 to 38 to 37.
If one packet is 50 bytes (50 × 8 bits) and the transmission speed is 100 Mbps,
The bandwidth length that one packet occupies the transmission path is
Band length = 50 × 8 × light speed / 100M = 1200 m.
Therefore, if the distance between nodes = bandwidth / 2 = 600 m,
Without storing the data 35_38 in the FIFO of the node 37,
The data 35_38 is transferred to the node 36.
From the above calculations,
At a transmission speed of 100 Mbps and a distance of 600 m between nodes,
If one packet is more than 50 bytes,
The transmission path utilization efficiency of the subnet can be increased up to 100%.
However, if one packet is less than 50 bytes,
There is a limit to the transmission path utilization efficiency of the subnet.
Of course, if the distance between nodes is shortened or the subnet is used at a low transmission rate, the transmission path utilization efficiency can be increased up to 100% even if the data length of one packet is less than 50 bytes. is there.
In FIGS. 8, 10 and 11, one packet has the same data length. However, as in the frame formats 61, 62, and 63 in FIG. It should be noted that communication on the subnet transmission line of the present invention only affects the above-mentioned transmission line utilization efficiency.
[Brief description of the drawings]
FIG. 1 is a prior art subnet.
FIG. 2 is an example of a multiplex communication network having a hierarchical structure to which the present invention is applied.
FIG. 3 shows a connection example in the vicinity of a wide area transmission line obtained by enlarging 20 in FIG.
FIG. 4 is a diagram illustrating a mechanism in which a ring-type transmission line is changed to a bus-type transmission line.
FIG. 5 shows a frame format for routing investigation.
FIG. 6 shows a frame format for checking a subnet connection.
FIG. 7 shows an embodiment of connection between a subnet and a wide area transmission line.
FIG. 8 is an explanatory diagram of the mechanism in which the flow of data in FIG. 7 is represented by the passage of time.
FIG. 9 is an enlarged view of the connection example of FIG.
FIG. 10 is a diagram showing the routing mechanism in FIG. 9 over time.
FIG. 11 is a comprehensive explanatory diagram expressing the flow of data over time.
FIG. 12 is a block diagram when the MARS-IF is realized by an LSI.
Explanation of reference numerals
10, 20: subnet according to an embodiment of the present invention
11 to 14: Wide area transmission path
15 to 19, 21 to 28: transmission path in the subnet
29: Node where data and token are logically disconnected
31 to 39: node
41: Token
42 to 46: transmission / reception data or data being transferred
43: Data stored in FIFO memory that can be temporarily stored
50, 60, 70: Terminator not requiring a specific number or branching device not requiring a specific number
51-53: Frame format for routing investigation
61 to 63: Subnet connection investigation frame format
76 to 99: MARS-IF block circuit

[0007]
If the data is divided into a plurality of packets, it is after the transmission of one packet.
◆ Node function of data processing to avoid congestion state by flow control 3-1. A number is added as control data to the token signals described in 2-1 and 2-2 above, and the number is increased only once before transmitting a series of self-transmission data, and is decreased only once after transmitting a series of self-transmission data. .
3-2. In 2-2, the following proviso is added.
However, if the number of the control data added to the token signal is equal to or greater than a predetermined value, the control unit does not transmit its own transmission data.
The subnet of the present invention differs from the conventional and prior arts in how tokens are captured. The token of the present invention serves as a trigger for obtaining the transmission right of transmission data, and immediately sends a token signal to the next node. Therefore, there is no need to hold a token signal during transmission of transmission data as in the prior art, and it is not the concept of transmission data with a token as used in the prior art. The necessary condition for making the token signal a simple trigger signal is to transmit the transmission data to the incoming terminal of the token signal and transmit the token signal to the other terminal. By using the token as a trigger for obtaining the transmission data transmission right, it is possible to transmit a signal at the same time, and the transmission path efficiency is improved. The token of the present invention is called a trigger token.
Since it is not necessary to manage tokens by numbers in the prior art, a simple signal sequence is sufficient for the token. The control data numbered token of the present invention has the same signal form as the numbered token like the prior art token, but the meaning of the attached number is a number that avoids the congestion state by flow control, and the identification of the node. Not for numbers.
(3) MARS-IF processing The "node" is composed of a CPU and a MARS-IF. In order to reduce the load on the CPU unit, the MARS-IF performs various types of processing regarding data transfer.
Basic data that prompts various types of processing is transmitted from the CPU unit through MARS-IF.

[0037]
The description of the best mode based on the individual claims is completed above, and the best mode based on the claims is shown and described.
FIG. 11 is an explanatory diagram of the mechanism expressed with the passage of time, similarly to FIGS. 8 and 10. Time flows from the upper side to the lower side in FIG. 11, and is indicated by an arrow shown on the left side in FIG. 11 as a time axis. Points different from FIG. 8 and FIG. 11 will be described.
The terminator 60 was changed to the terminal node 29. Nodes 31, 32, 33 and 39 have been added. "Destination_source" is described only for the node 35 in order to make FIG. 11 fit on one sheet of paper. The data denoted by the reference numeral 44 represents the data 31_33 or 33_31 transmitted and received between the nodes 31 and 33, and “X” is written therein for clarity. The data of the reference numeral 45 represents the data 32_37 transmitted and received between the nodes 37 and 32, and “/” is written to clearly show the data 32_37 and 37_32.
In the node 35 of FIG. 8, the data 35_36A, 35_38Y transmitted to the left, the data 35_34 transmitted to the right, and the data 35_36B, 35_37, 35_38Z transmitted to the left are described as "omitted". Further, the data described in FIG. 10 is also described as “omitted”. However, since the destination is indicated, if the data property bit is set to “10xx (data specifying the destination)”, the reception processing according to the classification in Table 3 is not “omitted” but “disappeared”. ", And is deleted in FIG.
The data 44 in which “X” is written in FIG. 11 simply passes through the node 32 between the nodes 33 and 31. The data 45 in which “/” is written simply passes through the nodes 36, 35, 34 and 33 between the nodes 37 and 32. Other data is transmitted and received to and from the node 35. Therefore, the nodes other than the destination and the source nodes simply pass through intermediate nodes, similarly to the data of “X” and “/”.
Focusing on the transmission data of the node 35, data 34_35, 33_35, 69_35 is transmitted by the trigger token arriving from the left side 34, and data 39_35, 36_35, 38_35 is transmitted by the trigger token arriving from the right side 36. , Followed by data 34_35, 33_35, 69_35 by the trigger token coming from the left 34.

[0038]
Sent. The transmission time interval between the data 33_35 and 69_35 differs depending on the presence / absence of the data 45 of “/”. Of course, if the transmission time interval between the data 33_35 and 69_35 is minimized, the data 45 of “/” is transferred by the node 35 after the transmission of the data 69_35. However, as a condition for minimizing the transmission time interval between the data 33_35 and 69_35, the capacity of the FIFO memory 43 increases.
In order to minimize the capacity of the FIFO memory, it is the best embodiment to give the sending terminal priority in the following order.
1. 1. Transfer data: data that the MARS-IF receives from the transmission line terminal 2. Transmission data: data received from the CPU subordinate to the MARS-IF Token signal (including trigger token signal)
Since the transmission of the token signal is the lowest priority, the transmission of the token is reserved if the transmission path terminal to be transmitted is used as described in [2-1]. The node 35 transmits six packets of 34_35, 33_35, 69_35, and 39_35, 36_35, 38_35 during one cycle of the token. This means that the object of the present invention is to realize the multiplexing in the wide area transmission line in FIG. 2 and to immediately transfer the data received from the upper wide area transmission path 14 to the lower area wide transmission path 11. is there.
If the terminal node 29 and the terminator 50 in FIG. 11 have the same value as the ring-connected node 29 shown in FIG. 4, as a measure to increase the amount of data “/” that simply passes through the node 35, the terminal node 29 Is a two-terminal bus node, and logically disconnected at the node 36 or 37, the data "/" that simply passes through the node 35 can be extremely reduced, and the transmission line utilization efficiency of the subnet can be increased.
The node was divided into a CPU section and a MARS-IF. If the congestion control portion where the processing change is large is realized by a program of the CPU unit and the MARS-IF is a small LSI, it can be realized by a partial modification of the circuit described in the prior art.
Therefore, a block diagram serving as a guideline for manufacturing an interface LSI satisfying the function of the MARS-IF will be described in [Industrial Applicability].

[0007]
If the data is divided into a plurality of packets, it is after the transmission of one packet.
◆ Node function of data processing to avoid congestion state by flow control 3-1. A number is added as control data to the token signals described in 2-1 and 2-2 above, and the number is increased only once before transmitting a series of self-transmission data, and is decreased only once after transmitting a series of self-transmission data. .
3-2. In 2-2, the following proviso is added.
However, if the number of the control data added to the token signal is equal to or greater than a predetermined value, the control unit does not transmit its own transmission data.
The subnet of the present invention differs from the conventional and prior arts in how tokens are captured. The token of the present invention serves as a trigger for obtaining the transmission right of transmission data, and immediately sends a token signal to the next node. Therefore, there is no need to hold a token signal during transmission of transmission data as in the prior art, and it is not the concept of transmission data with a token as used in the prior art. The necessary condition for making the token signal a simple trigger signal is to transmit the transmission data to the incoming terminal of the token signal and transmit the token signal to the other terminal. By using the token as a trigger for obtaining the transmission data transmission right, it is possible to transmit a signal at the same time, and the transmission path efficiency is improved. The token of the present invention is called a trigger token.
Since it is not necessary to manage tokens by numbers in the prior art, a simple signal sequence is sufficient for the token. The control data numbered token of the present invention has the same signal form as the numbered token like the prior art token, but the meaning of the attached number is a number that avoids the congestion state by flow control, and the identification of the node. Not for numbers.
(3) MARS-IF processing The "node" is composed of a CPU and a MARS-IF. In order to reduce the load on the CPU unit, the MARS-IF performs various types of processing regarding data transfer.
Basic data that prompts various types of processing is transmitted from the CPU unit through MARS-IF.

[0037]
The description of the best mode based on the individual claims is completed above, and the best mode based on the claims is shown and described.
FIG. 11 is an explanatory diagram of the mechanism expressed with the passage of time, similarly to FIGS. 8 and 10. Time flows from the upper side to the lower side in FIG. 11, and is indicated by an arrow shown on the left side in FIG. 11 as a time axis. Points different from FIG. 8 and FIG. 11 will be described.
The terminator 60 was changed to the terminal node 29. Nodes 31, 32, 33 and 39 have been added. "Destination_source" is described only for the node 35 in order to make FIG. 11 fit on one sheet of paper. The data denoted by the reference numeral 44 represents the data 31_33 or 33_31 transmitted and received between the nodes 31 and 33, and “X” is written therein for clarity. The data of the reference numeral 45 represents the data 32_37 transmitted and received between the nodes 37 and 32, and “/” is written to clearly show the data 32_37 and 37_32.
In the node 35 of FIG. 8, the data 35_36A, 35_38Y transmitted to the left, the data 35_34 transmitted to the right, and the data 35_36B, 35_37, 35_38Z transmitted to the left are described as "omitted". Further, the data described in FIG. 10 is also described as “omitted”. However, since the destination is indicated, if the data property bit is set to “10xx (data specifying the destination)”, the reception processing according to the classification in Table 3 is not “omitted” but “disappeared”. ", And is deleted in FIG.
The data 44 in which “X” is written in FIG. 11 simply passes through the node 32 between the nodes 33 and 31. The data 45 in which “/” is written simply passes through the nodes 36, 35, 34 and 33 between the nodes 37 and 32. Other data is transmitted and received to and from the node 35. Therefore, the nodes other than the destination and the source nodes simply pass through intermediate nodes, similarly to the data of “X” and “/”.
Focusing on the transmission data of the node 35, data 34_35, 33_35, 69_35 is transmitted by the trigger token arriving from the left side 34, and data 39_35, 36_35, 38_35 is transmitted by the trigger token arriving from the right side 36. , Followed by data 34_35, 33_35, 69_35 by the trigger token coming from the left 34.

[0038]
Sent. The transmission time interval between the data 33_35 and 69_35 differs depending on the presence / absence of the data 45 of “/”. Of course, if the transmission time interval between the data 33_35 and 69_35 is minimized, the data 45 of “/” is transferred by the node 35 after the transmission of the data 69_35. However, as a condition for minimizing the transmission time interval between the data 33_35 and 69_35, the capacity of the FIFO memory 43 increases.
In order to minimize the capacity of the FIFO memory, it is the best embodiment to give the sending terminal priority in the following order.
1. 1. Transfer data: data that the MARS-IF receives from the transmission line terminal 2. Transmission data: data received from the CPU subordinate to the MARS-IF Token signal (including trigger token signal)
Since the transmission of the token signal is the lowest priority, the transmission of the token is reserved if the transmission path terminal to be transmitted is used as described in [2-1]. The node 35 transmits six packets of 34_35, 33_35, 69_35, and 39_35, 36_35, 38_35 during one cycle of the token. This means that the object of the present invention is to realize the multiplexing in the wide area transmission line in FIG. 2 and to immediately transfer the data received from the upper wide area transmission path 14 to the lower area wide transmission path 11. is there.
If the terminal node 29 and the terminator 50 in FIG. 11 have the same value as the ring-connected node 29 shown in FIG. 4, as a measure to increase the amount of data “/” that simply passes through the node 35, the terminal node 29 Is a two-terminal bus node, and logically disconnected at the node 36 or 37, the data "/" that simply passes through the node 35 can be extremely reduced, and the transmission line utilization efficiency of the subnet can be increased.
The node was divided into a CPU section and a MARS-IF. If the congestion control portion where the processing change is large is realized by a program of the CPU unit and the MARS-IF is a small LSI, it can be realized by a partial modification of the circuit described in the prior art.
Therefore, a block diagram serving as a guideline for manufacturing an interface LSI satisfying the function of the MARS-IF will be described in [Industrial Applicability].

Claims (5)

Many nodes are laid in a ring by point-to-point connection,
Furthermore, regarding a token ring type network that uses tokens to avoid data races,
As a way to make a bus type network using tokens,
From many nodes with two transmission path connection terminals,
Select any one node,
For one terminal of the selected node,
When the token signal is received, the token signal is returned,
By receiving the data signal and discarding it,
With the function of a terminator,
For the other terminal of the selected node,
When the token signal is received, the token signal is returned,
By making it possible to receive and transmit data signals,
To have the function of terminal node,
Depending on the logic of the selected node internal element,
Different functions of terminator and terminal node are realized for each connection terminal,
Even in a network where many nodes are ring-shaped with real connections,
A token ring network having a mechanism equivalent to a bus type in which nodes are connected in a straight line. In networks that use tokens to avoid data races,
As a data transmission mechanism of a node member having two transmission path connection terminals,
If a token signal is received from any one transmission line connection terminal,
Send a token signal from the other transmission path connection terminal,
If there is transmission data originating from the own node,
Sends the token signal to the transmission path connection terminal that received it,
A data transmission mechanism of a node member capable of transmitting a token signal and transmission data at the same time by transmitting a token signal and transmission data to different transmission path connection terminals. In networks that use tokens to avoid data races,
Cycle the token signal with the number added as a token,
As the process of the node that received the token,
If the number is greater than the predetermined value,
Disables transmission of data originating from its own node,
If the number is less than the predetermined value,
Allow transmission of data originating from its own node,
When sending data originating from the own node,
Tokens arrived at the time of data transmission,
Send a token with a number according to the amount of data to be sent,
In the token that arrives after data transmission,
Sending a token with a number reduced according to the amount of data sent,
By
Flow control method for avoiding network congestion. Regarding nodes connected to the network, and data to be transmitted and received,
Divide the node into one CPU part and one interface part,
As processing of the interface unit having a plurality of transmission path connection terminals,
When data is received from the transmission line,
A bit corresponding to the received transmission path connection terminal is added to the head of the data,
Transfer the data to the CPU,
If data is received from the CPU,
Select the transmission path connection terminal corresponding to the bit described at the head of the received data,
Transferring data to the selected transmission path connection terminal,
By
An interface unit having an effect that the CPU unit can grasp the received transmission line connection terminal by analyzing the first bit of the data, and select the transmission line connection terminal to be transmitted by writing the instruction state in the first bit of the data. Regarding a switch or a terminator that is a member connected to a bus-type network,
When a data stream arrives to understand the network connection status,
As processing of a branching device having a plurality of transmission line connection terminals,
If a data string is received from any one transmission line connection terminal,
From among all other transmission line connection terminals,
To one predetermined transmission line connection terminal,
Sending out all incoming data streams,
As processing of the terminator having only one transmission line connection terminal,
A code for grasping the network connection status, or
Sending out all incoming data streams,
A branching device and a terminator capable of transferring a data sequence for grasping a network connection status to an adjacent member.

Images