Self-routing multipath switching network for switching cells in the asynchronous transfer mode
40 claims: 16 independent, 24 dependent
- 1Réseau de commutation à auto-acheminement par cellules et à trajets multiples pour la commutation de cellules à multiplexage temporel asynchrone comprenant :- des ports d'entrée, - des ports de sortie, - des commutateurs élémentaires (TSi1..., AS1..., TSr1...) arrangés en plusieurs étages de commutateurs élémentaires interconnectés, - chaque commutateur élémentaire ayant des entrées (I1...) et des sorties (o1..) et étant agencé pour transférer une cellule de longueur fixe ou variable reçue sur l'une de ses entrées, sur une ou plusieurs de ses sorties, en fonction d'informations d'acheminement (CCH) associées à ladite cellule, - les ports d'entrée du réseau correspondant aux entrées des commutateurs élémentaires d'un premier étage et les ports de sortie du réseau correspondant aux sorties des commutateurs élémentaires d'un dernier étage, caractérisé en ce que - chaque commutateur élémentaire d'un étage au moins du réseau possède au moins trois sorties, - ces sorties sont arrangées en groupes de sorties, - un groupe de sorties,comprend une ou plusieurs sorties déterminées, - en fonction de données d'acheminement (IRN, RCA, RCC) associées à une cellule reçue sur l'une quelconque de ses entrées, le commutateur élémentaire est agencé pour identifier (RMD) un ensemble comprenant un ou plusieurs desdits groupes de sorties, - le commutateur élémentaire est agencé pour transférer ladite cellule reçue, sur une sortie sélectionnée parmi les sorties du groupe unique dudit ensemble, ou des sorties, une par groupe dudit ensemble, chacune sélectionnée parmi les sorties du groupe auquel elle appartient.
- 2Réseau de commutation conforme à la revendication 1, caractérisé en ce que les commutateurs élémentaires de deux étages au moins (TS1..;AS1..) possèdent les caractéristiques indiquées dans la partie caractérisante de la revendication 1, en ce que l'arrangement des sorties en groupes n'est pas nécessairement le même dans chacun de ces deux étages au moins et en ce que les commutateurs de chacun de ces étages sont agencés pour détenir chacun des paramètres d'acheminement propres (RCCTM), dérivés d'une donnée de position dépendant de l'étage dans lequel il se trouve, ces paramètres différents permettant d'appliquer des arrangements différents dans les commutateurs élémentaires des deux étages considérés.
- 3Réseau de commutation conforme à la revendication 1, caractérisé en ce que les commutateurs élémentaires de deux étages au moins (TS1...;AS1...) possèdent les caractéristiques indiquées dans la partie caractérisante de la revendication 1, en ce que l'arrangement des sorties en groupes n'est pas nécessairement le même dans chacun de ces deux étages au moins et en ce que les commutateurs de chacun de ces étages sont agencés pour détenir chacun des paramètres d'acheminement propres (RCCTM), dérivés d'une donnée de position dépendant de l'étage dans lequel il se trouve et de son emplacement dans l'étage, ces paramètres différents permettant d'appliquer des arrangements différents dans les commutateurs élémentaires des deux étages considérés.
- 4Réseau de commutation conforme à la revendication 2 ou 3, caractérisé en ce que lesdites informations d'acheminement (CCH) sont interprétées dans chaque étage pour déterminer le mode de transfert (MT) d'une cellule reçue vers les sorties du commutateur élémentaire, et en ce que ladite interprétation s'effectue en fonction desdits paramètres d'acheminement.
- 5Réseau de commutation conforme à l'une quelconque des revendications 1 à 4, caractérisé en ce que les commutateurs élémentaires de certains étages du réseau au moins comprennent des moyens pour interpréter des informations d'acheminement, contenues dans une étiquette d'acheminement d'une cellule (CCH), qui comprennent un code de commande d'acheminement (RCC) définissant la séquence de transfert requis pour la cellule, à travers le réseau de commutation, une adresse de port de sortie (RCA), pour un acheminement de point à point, et/ou un numéro de référence interne d'arbre de diffusion (IRN), pour un acheminement de point à multipoint.
- 6Réseau de commutation conforme à la revendication 5, caractérisé en ce que les commutateurs élémentaires considérés, interprétant ledit code de commande d'acheminement (RCC) en fonction desdits paramètres d'acheminement, sont prévus pour mettre en oeuvre en conséquence un mode d'acheminement (MT) qui peut être, entre autres, un acheminement de point à point (RS) ou un acheminement de point à multipoint (MC).
- 7Réseau de commutation conforme à l'une quelconque des revendications 1 à 6, caractérisé en ce que les commutateurs élémentaires d'un étage au moins sont dissymétriques et réalisent chacun une expansion du trafic entrant de l'étage, réduisant la charge de trafic de cellules des sorties de ces commutateurs élémentaires par rapport à leurs entrées.
- 8Réseau de commutation conforme à l'une quelconque des revendications 1 à 6, caractérisé en ce que les commutateurs élémentaires d'un étage au moins sont dissymétriques et réalisent chacun une concentration de trafic sortant de l'étage, augmentant la charge de trafic de cellules des sorties de ces commutateurs élémentaires par rapport à leurs entrées.
- 9Réseau de commutation conforme aux revendication 7 et 8, caractérisé en ce que l'expansion est réalisée par au moins l'un des premiers étages (TSM...) et la concentration par au moins l'un des derniers étages (TSo1...), assurant ainsi une réduction de charge de trafic à l'intérieur du réseau de commutation entre ces deux types d'étates à commutateurs dissymétriques.
- 10Réseau de commutation conforme à la revendication 9, caractérisé en ce que le taux d'expansion dans au moins un des premiers étages est exactement compensé par le taux de concentration dans au moins un des derniers étages, réalisant ainsi un réseau de commutation symétrique ayant le même nombre de ports d'entrée et de sortie.
- 11Réseau de commutation conforme à l'une quelconque des revendications 1 à 6, caractérisé en ce que les commutateurs élémentaires de tous les étages, sont symétriques, ayant le même nombre d'entrées et de sorties, et en ce que le réseau de commutation est par suite également symétrique, ayant le même nombre de ports d'entrée et de ports de sortie.
- 12Réseau de commutation conforme à l'une quelconque des revendications 5 à 11, caractérisé en ce que les commutateurs élémentaires (ISE, Fig. 3) de certains des étages aux moins d'un réseau de commutation à plusieurs étages acheminent des cellules appartenant à deux courants de trafic opposés, en ce que, dans chacun de ces commutateurs élémentaires dits commutateurs bidirectionnels, les entrées sont réparties en deux ensembles d'entrées (I1..., I16 ;I17...I32) et les sorties en deux ensembles de sorties (01...,016 ;017..., 032), un courant de trafic étant normalement acheminé d'un premier ensemble d'entrées à un premier ensemble de sorties et l'autre courant de trafic d'un deuxième ensemble d'entrées à un deuxième ensemble de sorties, en ce que l'interprétation des informations d'acheminement dans le commutateur élémentaire tient compte du courant de trafic (I0, Fig. 4) auquel elles appartiennent.
- 13Réseau de connexion conforme à l'une quelconque des revendications 4 à 11, caractérisé en ce que les commutateurs élémentaires de tous les étages du réseau de commutation sont unidirectionnels, acheminant des cellules appartenant à un seul courant de trafic acheminées des entrées de chacun de ces commutateurs élémentaires à leurs sorties.
- 14Réseau de connexion conforme à la revendication 13, caractérisé en ce qu'il comprend trois étages au moins, chaque commutateur élémentaire d'un étage, sauf le dernier, étant connecté par une ou plusieurs mailles à chaque commutateur élémentaire de l'étage suivant et chaque commutateur élémentaire d'un étage, sauf le premier, étant connecté par une ou plusieurs mailles à chaque commutateur élémentaire de l'étage précédent.
- 15Réseau de commutation conforme à la revendication 13, caractérisé en ce qu'il comprend au moins une unité de sélection d'entrée (USi, Fig. 6) et au moins une unité de sélection de sortie (USr), formées chacune de commutateurs élémentaires agencés en deux étages au moins et dans chacune desquelles chaque commutateur d'un étage, sauf le dernier, est connecté par une ou plusieurs mailles à chaque commutateur élémentaire de l'étage suivant et chaque commutateur élémentaire d'un étage, sauf le premier, est connecté par une ou plusieurs mailles à chaque commutateur élémentaire de l'étage précédent, les unités de sélection d'entrée étant disposées en tandem avec les unités de sélection de sortie, les ports d'entrée étant connectés aux entrées des unités de sélection d'entrée et les ports de sortie aux sorties des unités de sélection de sortie.
- 16Réseau de commutation conforme à la revendication 15, caractérisé en ce que des unités de sélection dites plans de sélection (PS1..., PS16, Fig. 11 ou 12) connectent chacune des sorties d'unité de sélection d'entrée à des entrées d'unité de sélection de sortie.
- 17Réseau de commutation conforme à l'une quelconque des revendications 5 à 16, caractérisé en ce que, dans les commutateurs élémentaires d'un étage au moins, l'un desdits groupes de sorties comprend toutes les sorties du commutateur élémentaire, dans le cas d'un commutateur élémentaire unidirectionnel, ou alors toutes les sorties de l'un des deux ensembles de sorties pour l'une des deux directions de transfert, dans le cas d'un commutateur élémentaire bidirectionnel, pour une distribution général (DI, Fig 4) du trafic entrant sur toutes les sorties de ce commutateur dans la direction de transfert entrante.
- 18Réseau de commutation conforme à la revendication 12, caractérisé en ce qu'il est configuré en réseau replié extensible, possédant au moins deux étages dont l'un au moins est bidirectionnel (Fig. 7) et, lorsque c'est le cas pour le premier étage, les ports d'entrée étant connectés aux entrées dudit premier ensemble d'entrées des commutateurs élémentaires d'un premier étage et les ports de sortie étant connectés aux sorties dudit deuxième ensemble de sorties de ces mêmes commutateurs élémentaires du premier étage, le dernier étage étant composé de commutateurs élémentaires unidirectionnels qui acheminenent un courant de trafic entrant vers les commutateurs de l'étage précédent, réalisant ainsi une réflexion de trafic.
- 19Réseau de commutation conforme à la revendication 12, caractérisé en ce qu'il est configuré en réseau replié extensible, possédant au moins trois étages (fig 9), dont l'un au moins est bidirectionnel et, lorsque ce n'est pas le cas pour le premier étage, alors constitué de deux ensembles de commutateurs homologues unidirectionnels, l'un entrant, l'autre sortant, les ports d'entrée étant connectés aux entrées dudit ensemble de commutateurs unidirectionnels entrant du premier étage et les ports de sortie étant connectés aux sorties dudit ensemble de commutateurs unidirectionnels sortant du même premier étage, le dernier étage étant composé de commutateurs élémentaires unidirectionnels qui acheminent un courant de trafic entrant vers les commutateurs de l'étage précédent, réalisant ainsi une réflexion.
- 20Réseau de commutation conforme à la revendication 18, caractérisé en ce que chaque commutateur élémentaire d'un étage, sauf le dernier, est connecté par une ou plusieurs mailles à chaque commutateur élémentaire de l'étage suivant et chaque commutateur élémentaire d'un étage, sauf le premier, est connecté par une ou plusieurs mailles à chaque commutateur élémentaire de l'étage précédent.
- 21Réseau de commutation conforme à la revendication 18 ou 19, caractérisé en ce qu'il comprend trois étages au moins (fig. 10) et en ce que les deux premiers étages au moins sont constitués d'unités de sélection d'entrée et de sortie formées chacune de commutateurs élémentaires agencés en deux étages au moins et dans chacune desquelles chaque commutateur élémentaire d'un étage, sauf le dernier, est connecté par une ou plusieurs mailles à chaque commutateur élémentaire de l'étage suivant et chaque commutateur élémentaire d'un étage, sauf le premier, est connecté par une ou plusieurs mailles à chaque commutateur élémentaire de l'étage précédent.
- 22Réseau de commutation conforme à la revendication 21, caractérisé en ce que les unités de sélection d'entrée et de sortie sont interconnectées par au moins deux plans de sélection (PS1..., PS16, fig. 11,12), chacun comprenant un arrangement d'un ou plusieurs étages selon la capacité requise pour le réseau de commutation.
- 23Réseau de commutation conforme à la revendication 22, caractérisé en ce que l'augmentation de la capacité dudit réseau extensible est réalisée par addition d'étages successifs et sans modification de câblage entre étages, chaque configuration intermédiaire de moins grand nombre d'étages utilisant comme dernier étage équipé un étage de commutateurs élémentaires bidirectionnels qui peuvent effectuer une réflexion de dernier étage du réseau replié.
- 24Réseau de commutation conforme à l'une quelconque des revendications 4 à 12 et 18 à 23, caractérisé en ce que les commutateurs élémentaires d'un ou plusieurs des premiers étages du réseau de commutation sont agencés pour que, dans la direction d'acheminement correspondant au trafic entrant et lorsque les données d'acheminement spécifient un acheminement de point à point et/ou un acheminement de point à multipoint, et compte tenu desdits paramètres d'acheminement, il soit procédé à la distribution générale (DI) du trafic entrant.
- 25Réseau de commutation conforme à l'une quelconque des revendication précédente, caractérisé en ce que, dans un commutateur élémentaire, toute cellule devant être transférée à l'une des sorties d'un groupe de sorties sélectionné, est fournie sur l'une quelconque des sorties de ce groupe.
- 26Réseau de commutation conforme à la revendication 25, caractérisé en ce que ladite sélection, entre les sorties d'un groupe d'acheminement, s'effectue de manière à équilibrer la charge de cellules sur les différentes sorties de ce groupe.
- 27Réseau de commutation conforme à la revendication 24 ou 26, caractérisé en ce que ladite sélection s'effectuant de manière à équilibrer la charge de cellules sur les différentes sorties de ce groupe est basée sur un processus de distribution quasi ou pseudo-aléatoire visant à une décorrélation entre la répartition des cellules sur les entrées et la répartition des cellules sur les sorties du commutateur élémentaire.
- 28Réseau de commutation conforme à la revendication 13, caractérisé en ce qu'il comprend deux réseaux unidirectionnels dissymetriques à trois étages homologues (fig. 8), le premier interconnectant N1 ports d'entrée à N2 ports de sorties, le second interconnectant N2 ports d'entrée à N1 ports de sortie, juxtaposés de façon à réaliser l'équivalent d'un réseau de commutation interconnectant dans les deux sens de trafic, d'un côté N1 ports d'entrée et de sortie et, de l'autre, N2 ports de sortie et d'entrée.
- 29Réseau de commutation conforme à la revendication 12, caractérisé en ce qu'il comprend trois étages au moins dont l'un au moins est bidirectionnel (fig 9), arrangés pour constituer un réseau bidirectionnel non replié interconnectant un premier ensemble de N1 ports d'entrée et de sortie et un deuxième ensemble de N2 ports de sortie et d'entrée, en ce que chaque commutateur élémentaire d'un étage interne est connecté aux commutateurs élémentaires des étages précédent et suivant par une ou plusieurs mailles et chaque commutateur élémentaire d'un étage d'extrémité est connecté d'un côté aux commutateurs élémentaires de l'étage adjacent et, de l'autre, à l'un des ensembles ports d'entrée et de sortie, et en ce que les cellules sont transférées entre un port d'entrée et un port de sortie appartenant soit aux deux ensembles de ports différents, par transfert à travers les trois étages, soit appartenant au même ensemble de ports par transfert réfléchi dans d'un des commutateurs élémentaires d'un étage bidirectionnel permettant l'interconnexion de ces deux ports.
- 30Réseau de commutation conforme à la revendication 29, caractérisé en ce qu'il comprend au moins quatre étage (fig. 10), en ce que les deux premiers étages au moins, proches de l'ensemble des N1 ports d'entrée et de sortie, sont constitués d'unités de sélection d'entrée et de sortie (TSU1..., TSUn) formées chacune de commutateurs élémentaires agencés en deux étages au moins, en ce qu'un étage au moins, proche de l'ensemble des N2 ports d'entrée et de sortie interconnecte, d'une part, cet ensemble de N2 ports et, d'autre part, les unités de sélection donnant accès audits N1 ports, et en ce que les commutateurs élémentaires d'étages successifs sont interconnectés par une ou plusieurs mailles.
- 31Réseau de commutation conforme à l'une quelconque des revendications précédentes, caractérisé en ce que l'ordre des cellules, susceptible d'être modifié après transfert dans le réseau de commutation suivant des trajets différents, est rétabli par un circuit de rétablissement d'ordre de cellules prévu dans chaque port de sortie du réseau de commutation.
- 32Réseau de commutation conforme à la revendication 31, caractérisé en ce que des liaisons externes à débit relativement élevé sont connectées à plusieurs ports du réseaux de commutation, ces ports fonctionnant à débit moins élevé, en ce que les cellules entrantes d'une telle liaison à haut débit sont distribuées sur les différents ports d'entrée auxquels elle est raccordée, et en ce que les cellules destinées à une telle liaison à haut débit sortante sont acheminées dans le réseau de commutation vers le groupe de ports de sortie auxquels elle est raccordée, les cellules de ces ports de sortie étant multiplexées au débit plus élevé de la liaison sortante après rétablissement de l'ordre des cellules sur l'ensemble des ports de sortie de la liaison sortante considérée.
- 33Réseau de commutation conforme à la revendication 16 ou 22, caractérisé en ce que le nombre desdits plans de sélection est dimensionné en fonction de la plus élevée des charges de trafic de cellules moyennée sur l'ensemble des ports d'entrée et de sortie de chaque unité de sélection, la prise en compte uniquement de ces valeurs moyennes étant rendu possible par la distribution du trafic sur une multiplicité de trajets dans chaque unité de sélection.
- 34Réseau de commutation conforme à la revendication 6, caractérisé en ce qu'il comprend des commutateurs élémentaires disposant d'une mémoire de points de branchement (MCM) et en ce que les identités des groupes de sorties correspondant aux branchements à effectuer dans un commutateur élémentaire, lorsqu'il doit effectuer un mode d'acheminement de point à multipoint est obtenu par la lecture de la mémoire de points de branchement à l'aide dudit numéro de référence interne d'arbre de diffusion (IRN).
- 35Réseau de commutation conforme à la revendication 24 ou 34, caractérisé en ce que le contenu des mémoires de points de branchement (MCM) des commutateurs élémentaires des différents étages est marqué de telle façon que seules celles des commutateurs élémentaires effectuant un acheminement sélectif contiennent des indications de branchement nécessaires à l'étage considéré pour l'arbre de diffusion considéré, en ce que, en conséquence, aucune indication de branchement n'est marquée dans les commutateurs élémentaires des étages effectuant une distribution générale du trafic, et en ce que les différents commutateurs élémentaires qui appartiennent à un ensemble de commutateurs élémentaires équivalents en ce qui concerne les trajets internes entre étages, possèdent le même contenu de points de branchement pour chaque arbre de diffusion, permettant ainsi d'éviter la génération de copies de cellules inutiles dans le réseau de commutation.
- 36Réseau de commutation conforme à la revendication 34, caractérisé en ce que ledit arbre de diffusion (IRN) ne dépend pas du port d'entrée, mais uniquement de l'ensemble des ports de sortie vers lesquels une cellule provenant d'un port d'entrée quelconque doit être transférée.
- 37Réseau de commutation conforme à l'une quelconque des revendications précédentes, caractérisé en ce qu'une partie au moins des commutateurs élémentaires est remplacée par des modules de commutation chacun d'eux étant composé de plusieurs commutateurs élémentaires agencés de telle façon que ledit module de commutation présente au niveau de ses d'accès d'entrée et de sortie les caractéristiques et performances d'un commutateur élémentaire virtuel unique de plus grosse taille en nombres d'entrée et de sortie et dispose des fonctionalités précédement citées.
- 38Réseau de commutation conforme à l'une quelconque des revendications précédentes, caractérisé en ce que lesdites cellules sont des paquets.
- 39Réseau de commutation conforme à l'une quelconque des revendications précédentes, caractérisé en ce que lesdites cellules sont des cellules de longueur fixe ou variable composées d'un certain nombre de sous-cellules de longueur fixe.
- 40Réseau de commutation conforme à l'une quelconque des revendications précédentes, caractérisé en ce que lesdites informations d'acheminement (CCH) associées à la cellule sont contenues dans la cellule même.
Independent claims40
185 paragraphs, as filed
The present invention relates to a self-routing, multi-path switching network for switching cells with asynchronous time multiplexing.
A switching network, in its broadest structural definition comprises input ports to which input links are connected, output ports to which output links are connected, elementary switches arranged between these input ports and these output ports, arranged in one or more stages and interconnected by meshes. The input ports of the switching network are associated with the elementary switch inputs of a first stage, while the output ports are associated with the elementary switch outputs of a last stage. The outputs of the elementary switches of said first stage are coupled by meshes, possibly by means of intermediate stages of elementary switches, to the inputs of the elementary switches of said last stage. The meshes, in general, include one or more autonomous links, each connecting an output of a switch to an input of another switch, or the same.
Such a switching network is said to be multipath when it offers more than one switching path between any input port and any output port. In this case, the designation of an input port and an output port is not sufficient to determine a path going from one to the other. There is still a choice to be made between the multiple paths offered by the switching network between this input port and this output port.
Such a network is also said to be self-routing, in the broad sense of the term, when it is such that the determination of a switching path is made, in the switching network, not only with the aid of information. routing containing the implicit or explicit designation of an input port and that of an output port of the switching network, but also by a routing decision specific to the switching network.
The self-routing capability introduced into the switching network thus provides an autonomous means for solving the routing problem posed by multipath networks, when routing is only defined by the designation of an input port. and an output port.
Multi-path and connection-based self-routing switching networks are well known in circuit switching or synchronous time multiplex channel switching. The routing or connection establishment process is accomplished there once, at the start of a communication, and determines a switching path whose elements, initially free, are then occupied by the communication, and therefore become unavailable for calls. other communications. The established switching paths therefore cause a risk of blockage due to lack of channels in the search for any new switching path. The switching process applied in these known switching networks generally embraces the whole or a significant part of the switching network. It is therefore complex and requires a relatively long period of time, in terms of elementary control operations to establish the connection. On the other hand, this period of time is very short, compared to the duration of the communication and is therefore not detrimental, as regards the communication efficiency of the switching network.
The cells, also called packets, are digital information units comprising in particular a label containing information making it possible to identify the destination of the cell, as well as communication data.
The cells are either fixed length or variable length. In the latter case, the cell is transferred in the form of an uninterrupted sequence of a certain number of sub-cells of fixed length of relatively small size, which makes it possible to optimize the internal transfer, the storage in memory or in the registers, and processing functions at the sub-cell level.
In an asynchronous time division multiplex cell transmission system, the cells of several communications, transmitted on the same link, follow each other in any order and cells relating to the same communication are transmitted at irregular intervals.
In view of the characteristics of transmission of asynchronous time-division multiplex cells, a switching network of asynchronous time-division multiplex cells is generally a switching network arranged to individually switch such cells, so that a switching path is sought for each of them, to route it from a network input port to at least one output port. In this case, a "self-routing by cell" switching network makes it possible to perform a path search and selection for each individual cell. In addition, it is also commonly accepted that, in this type of network, provision should be made not only for conventional, so-called point-to-point routing, from an input port to an output port, but also at least the so-called point-to-multipoint routing, between an input port and several output ports.
However, a prior connection establishment process, such as that which has just been mentioned for circuit switching networks or synchronous time-division multiplex channels, if it is in principle also applicable to cell switching with asynchronous time multiplexing, however presents a certain number of drawbacks, such as the complexity of managing the speed of the connections established on each internal mesh of the network and too long a period of time to establish data connections.
Multipath switching networks and connection auto-routing known in circuit switching or synchronous time multiplex switching are therefore not an optimal solution for switching cells with asynchronous time multiplexing.
The technique has therefore turned to specific switching networks in which the switching process, including the path search, takes place stage by stage for each individual cell, therefore without establishing and marking prior connection at the start of the communication.
Examples of such a switching network can be found in the articles "Fast Packet Technology for Future Switches" by JJ Degan et al, AT & T Technical Journal, No. 2 March / April 1989, and "A 2x2 Switching Element for Broadband ISDN ", from GB Mund IEEE Pacific Rim Conference on Communications, Computers and signal Processing", 1-2 June 1989, and more particularly in the article "Design of a Broadcast Packet Network", by JS Turner, published in "Proceedings of IEEE INFOCOM'86", Fifth Annual Conference, "Computers and Communications Integration Design, Analysis, Management", pages 668 to 673.
A network such as that of this last document comprises, from input ports to output ports, stages specialized in the multiplication of cells necessary for point-to-multipoint communications, stages specialized in the mixing of cells, for mixing the cells from the different inputs and obtain equal and stable average flow rates per mesh, as far as statistically possible, and stages specializing in selective routing to exit ports. This network is constructed using elementary switches with two inputs and two outputs only. A small buffer for two cells is provided at each input of the basic switch. When a cell must take an exit which is not available, because another cell is then retransmitted on this exit, the entry can temporarily keep the cell not retransmitted.
Such a network has drawbacks and limitations such as, in particular:<ul id="ul0001" list-style="dash" compact="compact"><li>a large number of stages, when the number of input and output ports is high,</li><li>an efficiency limited by the retention of cells at the entrances in the event of occupation of the exits,</li><li>a certain difficulty in the realization of extensions, due inter alia to the large number of stages,</li><li>the existence of cell multiplication stages, causing an additional cost,</li><li>a certain sensitivity to the process of arrival of cells at each input port, which is not completely controlled by the use of patch stages and which therefore affects the performance of the switching network, etc.</li></ul>
The subject of the present invention is a self-routing asynchronous time multiplex cell switching network per cell not suffering, or at least suffering only to a much lesser extent, from the above-mentioned drawbacks and limitations.
The switching network of the present invention is characterized in that:<ul id="ul0002" list-style="dash" compact="compact"><li>each switch of at least one stage of the network has at least three outputs,</li><li>these exits are arranged in groups of exits called routing groups,</li><li>a group of outputs includes one or more specific outputs,</li><li>according to routing information associated with a cell of fixed or variable length received on any one of its inputs, the switch is designed to identify a set comprising one or more of said groups of outputs, a copy of the cell being transferred to each of the identified output groups,</li><li>the switch is arranged to transfer said received cell, to an output selected from the outputs of the single group of said set, or outputs, one per group of said set, each selected from the outputs of the group to which it belongs.</li></ul>
This network is therefore composed, in at least one of its stages, of elementary switches which, by the constitution of groups of determined outputs, between which each time an output is selected, not only provide a selective routing, but also bring a mixing cells by distribution of these on the different outputs of a routing group. In addition, the possible routing on an output of each of several groups achieves the multiplication of cells necessary in the case of point-to-multipoint routing. Thus, as will appear below, such a switch lends itself to the realization of integrated switching networks in which the specialization of parts of the switching network disappears. The more the basic switch includes outputs per group, the better the shuffling is carried out, because the greater the number of paths offered to a cell, which also tends to reduce blocking due to lack of available output, or to increase the efficiency of routing of the elementary switch and therefore of the entire switching network. Similarly, the elementary switch is then less sensitive to the process of arrival of the cells, since the cells of the same input are routed by a multiplicity of different paths to each stage.
In a multistage switching network, provision may be made for elementary switches of at least two stages to have the characteristics indicated; the arrangement of the outputs in groups is then not necessarily the same in each of these two stages at least and the switches of each of these stages are arranged to hold each of the own routing parameters derived from a position datum.
Such arrangements will make it possible to take account of the respective location of the elementary switches, in the switching network, in the arrangement of their outputs in groups and in particular of the fact that the mesh between pairs of successive stages is not necessarily the same.
According to another characteristic of the invention, said routing information is interpreted in each stage to determine the mode of transfer of a received cell to the outputs of the elementary switch, and said interpretation is carried out as a function of said routing parameters derived from its position.
These provisions make it possible to use the same type of switch in the different stages of a switching network and to use a single set of routing information in a cell and nevertheless to obtain different modes of transfer from the cell in the different stages of the switching network, which provides great flexibility in the use of different routing modes to transfer a cell through the switching network according to a variety of possible transfer sequences.
In a particularly advantageous application mode, the switches of at least certain stages of the network include means for interpreting routing information, contained in a routing label of a cell, which includes a routing control code (RCC) defining the transfer sequence required for the cell, an output port address (RCA), for point-to-point routing, and / or an internal broadcast tree reference number (IRN), for point-to-multipoint routing.
The switches considered, interpreting said routing control code as a function of said routing parameters derived from its position, are provided to implement consequently a routing mode which can be, inter alia, point-to-point routing or point-to-multipoint routing.
The invention includes the case where the elementary switches of all the stages are symmetrical, having the same number of inputs and outputs, and where the switching network is consequently also symmetrical, having the same number of input ports and output ports.
According to another characteristic of the invention, the elementary switches of at least one stage can be asymmetrical and each achieve an expansion of incoming traffic, reducing the traffic load of cells of the outputs of these switches compared to their inputs.
Such arrangements make it possible to reduce the load on the outputs of the switches and therefore the retention of the cells or their loss in the elementary switches, or the corresponding buffer memory requirements in these elementary switches.
Conversely, the elementary switches of at least one stage can be asymmetrical and each achieve a concentration of outgoing traffic by increasing the traffic load of cells of the outputs of these switches compared to their inputs.
Each of these two types of use of asymmetrical elementary switches in at least one stage makes it possible to produce asymmetrical switching networks, that is to say those in which the numbers of input ports and of output ports are different, in the first case to break down traffic over a greater number of outgoing ports and in the second case to concentrate traffic over a smaller number of outgoing ports.
In addition, the combined use of asymmetrical stages, but of different directions, in the same switching network is also possible. In particular, it is even possible to produce a symmetrical switching network, having the same number of input ports and output ports, in which at least one of the first stages is asymmetrical and realizes an expansion of traffic and one of the last floors at least is asymmetrical and achieves a concentration of traffic which compensates for the prior expansion. There is then obtained, between these two asymmetrical stages, a reduction in the traffic load of the internal cells of the switching network with the advantages mentioned above.
The invention also relates to a switching network in which the switches of some of the stages at least route cells belonging to two opposite traffic streams, and in which, in each of these switches called bidirectional switches, the inputs are distributed in two sets of inputs and outputs into two sets of outputs, one traffic stream normally being routed from a first set of inputs to a first set of outputs and the other traffic stream from a second set of inputs to a second set of outputs; the interpretation of the routing data in the elementary switch also takes account of the set of inputs to which they belong to determine whether the routing must be "normal", that is to say to the set of outputs associated (same traffic flow) or "reflected", that is to say to the other set of outputs (opposite traffic flow).
The invention also extends however to the case where the elementary switches of all the stages of the switching network are unidirectional, routing cells belonging to a single traffic stream routed from the inputs of each of these elementary switches to their outputs.
According to one embodiment of the invention, the connection network comprises at least three unidirectional stages, each switch of a stage, except the last, being connected by one or more meshes to each switch of the following stage and each switch of a stage, except the first, being connected by one or more meshes to each switch of the preceding stage.
According to another embodiment of the invention, the switching network comprises at least one input selection unit and at least one output selection unit, each formed of elementary switches arranged in at least two stages and in each of which each switch of a stage, except the last, is connected by one or more meshes to each switch of the following stage and each switch of a stage, except the first, is connected by one or more meshes to each switch of the previous stage, the input selection units being arranged in tandem with the output selection units, the input ports being connected to the inputs of the selection units input and output ports at the outputs of the output selection units.
In this latter embodiment, a single input selection unit can be connected directly to a single output selection unit, by mixing between the outputs of the first and the inputs of the second.
In the case of several input and output selection units, according to the invention, the switching network can further comprise selection units called selection planes connecting each of the input selection unit outputs to output selection unit inputs.
According to another characteristic of the invention, in the elementary switches of at least one stage, one of said groups of outputs of an elementary switch comprises all the outputs of the elementary switch, in the case of a unidirectional switch, or well all the outputs of one of the two sets of outputs, pc a given transfer direction, in the case of a bidirectional switch, for a general distribution of incoming traffic on all the outputs of this basic exchange, in the incoming transfer direction.
According to another embodiment of the invention, the switching network is configured as an expandable folded network, having at least two stages, at least one of which is bidirectional, the input ports being connected to the inputs of said first set of inputs of the elementary switches of a first stage and the output ports being connected to the outputs of said second set of outputs of these same elementary switches of the first stage, when the latter is bidirectional, the last stage being made up of unidirectional switches which convey a current of incoming traffic towards the switches of the preceding stage, thus realizing a traffic reflection.
When the first stage is not composed of bidirectional switches, but of two homologous sets of unidirectional switches (one for each transfer direction), the input ports are connected to the inputs of the incoming switches and the output ports to the outputs of the outgoing switches.
According to this latter embodiment, the folded switching network comprises at least two stages, at least one of which is bi-directional and each switch of a stage, except the last, is connected by one or more meshes to each switch of the next stage and each switch of a stage, except the first, is connected by one or more meshes to each switch of the preceding stage.
According to an alternative of this latter embodiment, the folded switching network comprises at least three stages and at least the first two stages consist of input and output selection units each formed of elementary switches arranged in two stages at less and in each of which each switch of a stage, except the last, is connected by one or more meshes to each switch of the following stage and each switch of a stage, except the first, is connected by one or more meshes to each switch of the previous stage.
In the latter case, in order to produce large-capacity switching networks which are extensible in number of stages, said input and output selection units can be interconnected by a number of selection units known as folded selection planes , each comprising an arrangement of one or more floors depending on the capacity required.
These different types of folded switching networks have at least one bidirectional stage offer the advantageous property of being extensible in number of stages, according to the required capacity in number of ports of the network, this without requiring modifications of wiring between stages . Indeed, any bidirectional stage can temporarily constitute the last stage equipped with an intermediate configuration thanks to its property of possible reflection of traffic from the incoming transfer direction to the outgoing transfer direction.
In the last embodiments envisaged, advantageously, the first stage or stages of the switching network carry out a general distribution of the incoming traffic over all the possible paths, that is to say over all the meshes, between these first stages , thus performing a mixing of the cells received on the input ports over a multiplicity of paths. In a basic unidirectional switch of such a stage, the single routing group therefore includes all the outputs of the switch; in a bidirectional elementary switch of such a stage, the set of outputs in the incoming transfer direction constitutes the single routing group by which the incoming traffic is distributed. In order to achieve such shuffling of incoming traffic, according to the invention, each cell must be able to choose any of the outlets available in such a distribution routing group.
Then, advantageously, the elementary switches of one or more of the first stages of the switching network are arranged so that, when the routing data specifies a point-to-point transfer and / or a point-to-multipoint transfer, and taking into account from said position data, a general distribution of the traffic entering the first stage or floors is carried out.
Advantageously also, the selection of one of the outputs of a group is carried out so as to balance the cell charge on the different outputs of this group.
Advantageously also, said selection being made so as to distribute the cell charge on the different outputs of a group is based on a quasi or pseudo-random distribution process aimed at decorrelation between the flows of cells on the inputs and the flow of cells on the outputs of the elementary switch.
Quasi or pseudo-random distribution is performed when selecting an output from the different outputs belonging to the selected routing group. This mechanism can intervene before registering the identities of cells in queues which are then individually assigned to the outputs. In addition, the two mechanisms described above, load balancing and random distribution, can be advantageously combined.
The multipath and self-routing switching networks according to the invention also have the following characteristics:<ul id="ul0003" list-style="dash" compact="compact"><li>The combined use of a multiplicity of paths for transferring the different cells of a communication to the output port or ports concerned and of elementary switches effecting a storage of the cells for an indefinite time has the consequence of routing these cells to the output port (s) with a variable transfer time capable of modifying the order of successive cells. This feature involves the introduction of cell order recovery circuits at each output port of the switching network.</li><li>The mixing of incoming cell traffic on a multiplicity of possible paths in the first distribution stages provides the characteristic property that the speed of the internal links of such switching networks is no longer conditioned by the speed of the external transmission links which are assigned to it. connected, or even by the speed of the services transmitted on these external links. Indeed, the principle of distribution by a multiplicity of paths inside the network makes it possible to distribute the flow of cells entering from an external link at high speed over several input ports of the switching network at low speed; for example, an external 2.4 Gbit / s link can be connected to 16 input ports at 150 Mbit / s. After transfer of the cells in the switching network on all possible paths, the cells intended for an outgoing external link at the same speed of 2.4 Gbit / s are routed to a group of 16 output ports at 150 Mbit / s where cell order restoration and asynchronous muultiplexing are performed on the outgoing 2.4 Gbit / s link.</li></ul>
Analogous reasoning shows that a service which would require a cell speed equivalent to 200 Mbit / s can be transferred in the switching network over a multiplicity of paths, each corresponding to an internal link at 150 Mbit / s.<ul id="ul0004" list-style="dash" compact="compact"><li>The characteristic consequence of mixing the traffic entering the first stages of distribution is also to achieve an averaging of the load of the external links on the switches of the internal stages of the switching network. It is then possible, for example, to equip a variable number of selection plans as a function of the average traffic load on all of the external links of the selection unit most loaded with traffic.</li><li>With regard to point-to-multipoint transfers according to pre-established broadcast trees, the multiplicity of possible paths for transferring a cell through the multi-path and self-routing switching network requires a characteristic organization of the contents of the memory memories. diffusion trees of the elementary switches of the different stages. According to the invention, it is possible to define the corresponding connection points so that no unnecessary copy is generated on any stage, which avoids any internal overload of the internal meshes between stages.</li></ul>
The various objects and characteristics of the invention will be explained in more detail in the course of the following description of an embodiment of the invention, provided without limitation, with reference to the appended figures which represent:<ul id="ul0005" list-style="dash" compact="compact"><li>FIG. 1, the known diagram of the circuits of an example of an ISE elementary switch used in the switching network of the present invention,</li><li>FIG. 2, the format of a cell, provided for the implementation of the present invention,</li><li>FIG. 3, the diagram of an application mode of the switch of FIG. 1, for routing the cells in two opposite directions,</li><li>FIG. 3bis, the detailed diagram of the logic for managing the memories of SBML sub-cells of FIG. 1,</li><li>FIG. 4, the diagram of the routing circuit RL according to the invention, applicable in the elementary switch of FIG. 1, making it possible to use the latter in accordance with FIG. 3,</li><li>FIG. 4bis, the detailed diagram of the cell management logic and selection of COQML output of FIG. 1,</li><li>FIG. 5, a symmetrical unidirectional switching network according to the present invention,</li><li>FIG. 6, a symmetrical unidirectional switching network with four stages arranged in two back-to-back selection units each having two stages,</li><li>FIG. 7, a folded symmetrical switching network corresponding to that of FIG. 6,</li><li>FIG. 8, an asymmetrical unidirectional switching network with two times three stages,</li><li>FIG. 9, a three-stage asymmetrical bidirectional switching network corresponding to the previous one,</li><li>FIG. 10, an asymmetrical bidirectional switching network with several two-stage selection units,</li><li>FIG. 11, a symmetrical unidirectional switching network with several two-stage selection units interconnected by several three-stage selection planes.</li><li>FIG. 12, a folded symmetrical switching network corresponding to that of FIG. 11.</li></ul>
In these figures, for simplicity, various connections are shown as single wires, although they can incorporate a plurality of such wires. In addition, the figures do not represent all of the control circuits, their realization evidently arising, for those skilled in the art, from the content of the description.
The elementary switch represented in FIG. 1 with X inputs I1 / IX and Y outputs O1 / OY (X and Y not being equal to 1 simultaneously) is arranged to switch digital signals grouped in cells or packets of fixed or variable lengths . Such a cell, represented in FIG. 2, is for example constituted by a series of successive sub-cells, comprising a first FSC sub-cell, an intermediate ISC sub-cell and a final LSC sub-cell, all of equal length. , for example 162 bits, that is 2 bits and 20 characters at 8 bits. Each of these sub-cells contains a SCH (2-bit) sub-cell control field and a DB1 - DBs data block, the first FSC sub-cell additionally containing a CCH cell label which, for example, contains routing information enabling the elementary switch to determine to which group (s) of RG1 / RGY outputs all the successive sub-cells belonging to the same cell must be successively transferred, this transfer takes place on the same or the same outputs. In the present description, the SCH sub-cell control field is assumed to have an explicit binary value 11, 00 or 01 indicating that the sub-cell is respectively the first FSC sub-cell, an ISC intermediate sub-cell or the last LSC sub-cell of the cell.
The CCH label itself comprises three parts, an RCC routing control field, a destination indication in the form of an RCA network exit address and an internal IRN broadcast tree reference number.
The RCC command field, which may include 5 bits, contains routing mode data designating a point-to-point routing mode or a broadcasting routing mode, or any other mode provided, some of which will be explained more far. If, for a switch, the RCC control field designates the point-to-point routing mode, analysis of the RCA network output address provides the identity of the selected output group. If the RCC command field designates the broadcast routing mode, the IRN broadcast tree reference number is used to read a memory which provides the identities of the output groups corresponding to the connections to be made for this tree in the elementary switch. .
The RCA network output address, which will include up to 14 bits, for example, is the identity of the switch network output port (or group of output ports) to which the received cell should be addressed . When the selective routing is carried out by more than one stage of the switching network, only part of this destination data is necessary in each of the elementary switches for the routing of the cell.
The internal reference number IRN, which will also include for example 14 bits, is a number used in the switching network to identify the broadcasting tree according to which an incoming cell must be transferred to a certain number of given output ports. It is interesting to note that, according to the invention, a diffusion tree in a self-routing and multipath network is not a point-to-multipoint connection, because, being independent of the input port, it is characterized only by all destination output ports; in addition, in such a multipath network, it corresponds to a multiplicity of potential point-to-multipoint paths between all of the input ports and all of the destination output ports of the broadcasting tree considered. In fact, a given IRN broadcast tree is not necessarily specific to a single communication, but can be used by all communications from any input ports requiring to transfer each cell to the set of output ports. recipients of this mailing tree.
FIG. 3 now represents, for reasons of writing convenience, a mode of use of the ISE elementary switch of FIG. 1 in the case of bidirectional routing, with the possibility of reflection, which will be explained later by referring in Figures 7, 9, 10 and 12.
The ISE elementary switch, in the example considered, has 32 inputs I1 to I32 and 32 outputs O1 to O32. The inputs I1 to I32 are divided into two sets of inputs I1 to I16 and I17 to I32. The outputs are divided into two sets of outputs O1 to O16 and O17 to O32. Internally, except in the case of reflection, the elementary switch is arranged to allow normal routing from left to right, of inputs I1 to I16 towards outputs O1 to O16, as well as, in parallel, but from right to left due external wiring mode, from inputs I17 to I32 to outputs O17 to O32. In the event of reflection, the switch allows the routing of inputs I1 to I16 to outputs O17 to O32, or else of inputs I17 to I32 to outputs O1 to O16. In such a switch, the assignment of the inputs to the routing directions is predetermined. It can be indicated by an IO bit attached to each entry and which indicates whether it belongs to an "incoming" direction (I1 to I16, for example, carrying traffic normally intended for all the outputs O1 to O16) or to the direction opposite "outgoing" (I17 to I32, according to the same example, carrying traffic normally intended for all the outputs O17 to O32).
The set of 16 exits of each direction can for example be divided into 8 groups at most of at least two exits and the routing of any cell on the exits of one of the groups simply requires identification, by a word 8 bits (one bit per group) of the group or groups to which the cell must be retransmitted, it being understood that the cell is retransmitted on a single output of each group thus identified.
Referring again to FIG. 1, the inputs I1 / IX of the elementary switch which are represented therein are connected to the respective data inputs of a multiplexer circuit MX by means of the cascade connection of series-parallel converter circuits. respective SPR1 / SPRX and respective locking circuits IL1 / ILX. The data output CI of the MX multiplexer is coupled to the data input, also CI, of a buffer of sub-cells BM, of RAM type, while the selection input XI of the MX multiplexer is controlled by a input clock circuit XC capable of successively connecting each of the X inputs of the multiplexer to the output of multiplexer CI, during a sub-cell period. Such a sub-cell period is the time interval during which a sub-cell is received in a series-parallel converter circuit SPR1 / SPRX.
The buffer memory BM is fully shared and its data output is connected to the data input of a DX demultiplexer, the Y data outputs of which are coupled to respective outputs O1 / OY by means of parallel-series converter circuits. respective PSR1 / PSRY. The selection input YJ of the demultiplexer DX is controlled by an output clock circuit YC capable of successively connecting the input of the demultiplexer to the Y outputs of the demultiplexer during a sub-cell period.
It should be noted that, for sub-cells having a length of 162 bits and for the same bit rate of 50 Mbit / s at the inputs and outputs, a sub-cell period is equal to:<maths id="math0001" num=""><math display="block"><mrow><mtext>162 / 50 = 3.24 µs.</mtext></mrow></math><img file="EP0446540B1_D0001.tif" /></maths>
More precisely, when, for example, the elementary switch has X = 32 inputs and Y = 32 outputs, 32 write operations and 32 read operations, i.e. 64 operations, must be carried out in the buffer memory BM during the same period of 3.24 µs sub-cell. Consequently, each of these operations must be carried out by:<maths id="math0002" num=""><math display="block"><mrow><mtext>3.24 / 64 = 50.62 ns.</mtext></mrow></math><img file="EP0446540B1_D0002.tif" /></maths>
Furthermore, when for example X = 16 and Y = 32, 48 operations must be carried out during the same sub-cell period. This means that each of these operations must be carried out by:<maths id="math0003" num=""><math display="block"><mrow><mtext>3.24 / 48 = 67.50 ns.</mtext></mrow></math><img file="EP0446540B1_D0003.tif" /></maths>
The buffer memory BM is subdivided into C, 512 for example, sub-cell buffer memory locations, each capable of recording a sub-cell, for example of 162 bits; it has an address input AC, as well as a read / write selection input RW, respectively coupled to the outputs of the same names of an SBML sub-cell buffer memory management logic.
The elementary switch further comprises a sub-cell logic SL and a routing logic RL, which are both coupled to the data output CI of the multiplexer MX.
The SL sub-cell logic is mainly a detector intended to detect and check the SCH sub-cell control field of each sub-cell and to provide active output signals LS, FO or NF, depending on whether the sub-cell is respectively a last LSC sub-cell, a first FSC sub-cell, or is not a first sub-cell.
The routing logic RL analyzes the routing information of the cell label CCH of each first sub-cell FSC of a cell and provides active output signals RMD and RC, as a function of l routing information. More particularly, the signal RMD provides the identity of one or more groups of selected outputs to which the sub-cells of the cell must be transferred, while the signal RC indicates the number of these groups of selected outputs, it is ie 1 for point-to-point transfer and a value greater than 1 for point-to-multipoint transfer. The type of routing information and the corresponding analysis process performed by the routing logic RL to generate the RMD and RC output signals depends on the routing mode used for the cell. The cell label CCH may, for example, contain Y bits of routing information, each of these bits corresponding to a group of outputs to which the cell is to be transferred.
COQML output cell queue management logic performs both cell queue management and output selection functions, registering the address of the first input sub-cell WISA in an appropriate queue, depending on the routing mode and the RMD data supplied by the RL logic, and by transmitting the address of the first FSAO output sub-cell to the SBML logic, together with the identity of the selected output YS.
The SBML sub-cell buffer management logic is coupled to the previously mentioned outputs LS, NF of the logic SL, RC of the logic RL, XI of the input clock circuit XC, FSAO of the management logic of COQML, and YJ cells of the YC output clock circuit. It manages the use of the buffer memory locations of the BM memory, by providing the address of free locations, making them occupied when they are used and freeing them when they are no longer used. Under the control of signals applied to its inputs, it also controls, via the read / write selection signal RW, the read and write operations in the buffer memory BM, while constituting linked lists chaining the addresses of buffer memory sub-cells of the same cell. This is necessary, since the sub-cells of the same cell are saved in uncorrelated locations of the buffer memory BM, whereas they must be routed on the same or the same selected outputs O1 / OY, and this in the same order and without interruption as they were when they arrived on one of the I1 / IX entrances.
We will now briefly describe the operation of the elementary switch considered. When a sub-cell of a variable-length cell, such as that represented in FIG. 2, appears on one of the inputs I1 / IX, I1 for example, of the elementary switch, it is received by the circuit corresponding serial-parallel converter SPR1. Supplied by this converter circuit SPR1, the parallel version of the sub-cell is transferred to the corresponding locking circuit IL1, by which it is supplied to the multiplexer MX. Under the control of the clock signal XI supplied on the input of the same name by the input clock circuit XC, the sub-cell is, at a certain moment, corresponding to this input I1, supplied to the input CI data from the buffer BM, as well as the sub-cell logic SL and the routing logic RL. It is then determined whether the sub-cell is a first FSC sub-cell, a last LSC sub-cell, or is not a first sub-cell, and for which group or groups of RG1 / RGY outputs this sub-cell - and the following sub-cells belonging to the same cell - must be transferred respectively. The resulting output signals LS, NF and RC are applied to the SBML sub-cell buffer management logic and the output signal RMD to the COQML output cell queue management logic.
Under the control of the clock signal XI, the SBML logic supplies the address of a free buffer memory location, WISA for example, to the address address AC of the buffer memory BM, consequently what the sub -cell present at the data entry CI of the buffer memory BM is recorded in the location of the latter having the address WISA. This WISA address is made busy and is added to the linked list of addresses of all the sub-cells already received from the same cell (in this case the signals NF, LS, RC and RMD are used). In this list, the addresses are arranged in the same order as that of the sub-cells of the cell.
During a read operation, under the control of the clock signal Yj supplied on the input of the same name by the output clock circuit YC, the address of a sub-cell, ROSA for example, is supplied to the address address AC of the buffer memory BM and the sub-cell contained in the corresponding memory location is read and transferred to the data output of the buffer memory BM. From there, it is supplied, via the DX demultiplexer to the output or to one of the outputs initially indicated by the signal YS of the COQML management logic.
We will now refer to FIG. 3bis, which represents the buffer management logic of the SBML sub-cells of FIG. 1, in more detail.
As already mentioned, this sub-cell buffer management logic has LS, NF, RC, XI, YJ, FSAO, YS inputs, and AC, L, RW and FSAI outputs. It includes a circuit for managing the free memory locations FMLMC, a sub-cell chaining memory SLM, an incoming sub-cell pointer memory ISPM and an outgoing sub-cell pointer memory OSPM.
The FMLMC circuit of the SBML logic is constituted by a queue memory for free locations FQ which is for example a queue memory of FIFO type (first input - first output) recording the addresses of all the locations BM buffer free. The FMLMC circuit has a ROSA input, a WISA output and QC and RW control terminals.
The SLM sub-cell chaining memory comprises C memory locations corresponding to the C buffer memory locations of the BM sub-cell buffer memory and stores, for each of them:<ul id="ul0006" list-style="dash" compact="compact"><li>the chaining address to the next sub-cell (NCB),</li><li>the number of sub-cell copies to be read (NC),</li><li>a last sub-cell indicator of the cell (L).</li></ul>
The SLM memory field is associated with a retrograde counter DC, so that the value NC is decremented by one at each reading operation from the SLM memory; then the new value is saved in place of the previous one. Of course, when the new value reaches zero, all the sub-cell copies have been read and the DC counter generates a QC signal allowing the recording of the address of the sub-cell read (ROSA), which becomes free, in the FMLMC free buffer location management circuit.
The ISPM incoming sub-cell pointer memory has X locations corresponding to the X inputs and it operates in synchronism with the input clock signals XI defining the time-division multiplex operation of the X inputs. For each entry, it records:<ul id="ul0007" list-style="dash" compact="compact"><li>the buffer address of the last received sub-cell (LCB),</li><li>the number of sub-cell copies to be read later (LC),</li><li>a last sub-cell indicator of cell (B).</li></ul>
The memory of pointers of outgoing sub-cells OSPM has Y locations corresponding to the Y outputs and it operates in synchronism with the output clock signals YJ defining the time multiplex operation of the Y outputs. For each input, it records the address of the location of the next sub-cell waiting to be transmitted on the output considered (WCB).
The write control circuit provides the various circuits with the appropriate control signals corresponding to the alternating operation of the SBML circuits associated with the alternate read and write operations in the buffer memory BM, consequently interleaved clock signals relating to the inputs. (XI) and exits (YJ). In agreement with this, the signal RW result is active during each write operation in the buffer memory BM, for the input of a sub-cell, and inactive during a read operation of the buffer memory BM for transmission in exit from a sub-cell.
The following description of principle characterizes the operation accomplished by these functions of the management logic SBML during a writing or reading phase and for each of the three types of sub-cells of a cell: first sub-cell FSC, sub- ISC intermediate cell and last LSC sub-cell.
We will first consider the write phase in the buffer memory BM, in the case of a first sub-cell FSC. In such a case, the signal FO is supplied and the management logic SBML receives logic SL and RL:<ul id="ul0008" list-style="dash" compact="compact"><li>NF = O, to indicate a first sub-cell,</li><li>LS = O, to indicate that it is not a last sub-cell,</li><li>RC = 2, assuming for example the case of a point-to-multipoint transfer on two groups of outputs.</li></ul>
Since the signal RW is active, a WISA sub-cell write address is supplied by the management circuit FMLMC, which is the free location of the selected buffer memory in which the received sub-cell is recorded. The WISA address is also stored in the ISPM pointer memory, for the entry XI, in order to save it as the address of last sub-cell received, for the next cycle relating to the same entry. Furthermore, the WISA address is also supplied to the COQML logic which will register it as the reference identity of this new cell received, because this logic receives the signal FO of value 1.
As regards the chaining memory SLM, since it is a first FSC sub-cell (NF = 0), the WISA address is not recorded in the NCB field, since this new sub -cell does not need to be chained with the last of the previous cell. In addition, the other data fields are used for this preceding sub-cell by selecting its address which is supplied by the LCB field of the ISPM pointer memory and by recording LC and B coming from the ISPM pointer memory in the NC fields. and L of the SLM memory respectively. The PC and LS control signals are stored respectively in the LC and B fields of the ISPM memory for input XI.
In the case of an intermediate sub-cell, the signal FO is inactive and the management logic for buffer memory of sub-cells SBML receives logic SL and RL from the signals:<ul id="ul0009" list-style="dash" compact="compact"><li>NF = 1,</li><li>LS = O,</li><li>the RC signal is not used with NF = 1.</li></ul>
As before, the signal RW is active and another WISA address is supplied by the management circuit FMLMC, address of buffer memory location which is used for:<ul id="ul0010" list-style="dash" compact="compact"><li>address the buffer memory BM and write therein the intermediate sub-cell ISC,</li><li>be recorded in the LCB field of the ISPM pointer memory as the new last sub-cell address received from the cell,</li><li>be recorded in the NCB field of the SLM memory addressed by the content of the LCB field of the ISPM memory, in order to register there that this new WISA address is the address of the next sub-cell linked with the previous one which is the memory location selected in the SLM memory.</li></ul>
Simultaneously, the data from the LC and B fields of the ISPM memory are transferred to the NC and L fields of the SLM memory, before B is replaced in the ISPM memory by a new value coming from the LS signal.
In the case of a last LSC sub-cell, the signal FO is inactive and the management logic SBML receives logic SL and RL:<ul id="ul0011" list-style="dash" compact="compact"><li>NF = 1,</li><li>LS = 1,</li><li>whereas RC is not used since NF = 1.</li></ul>
Again the signal RW is active and another WISA address is supplied by the FMLMC circuit, and the corresponding buffer location address is used in the buffer memory BM, and the ISPM and SLM memories exactly as in the previous case the registration of an ISC intermediate sub-cell.
Simultaneously, the LC and B values of the ISPM memory are transferred to the NC and L fields of the SLM memory before B is replaced, in the ISPM memory by the new value coming from the LS signal, consequently indicating that, now, the last sub-cell of a cell has just been received.
However, as was underlined in the description relating to the recording of a first FSC sub-cell, during the following cycle relating to the input XI, the values LC and B = 1 will be transferred into the fields NC and L of the SLM memory at the address of the previous cell (the last) provided by the LCB field of the ISPM memory.
We will now consider the reading phase of the buffer memory BM during which the signal RW is inactive. First, we will particularly consider the case of reading a first FSC sub-cell.
It must be assumed that, at the time of sending the first sub-cell, the content WCB of the memory of pointers of outgoing OSPM sub-cells of the considered output YS was initialized with the address of the first sub-cell of the cell to be transmitted. This will appear later, when reading the last sub-cell of the cell.
The OSPM pointer memory consequently supplies the address of the outgoing sub-cell to be read which is used for:<ul id="ul0012" list-style="dash" compact="compact"><li>address the buffer memory BM for reading the first corresponding sub-cell FSC,</li><li>select the SLM memory for reading, which provides:<ul id="ul0013" list-style="dash" compact="compact"><li>an NCB indication which is transferred to the OSPM memory for recording as a new WBC address, for the next cycle relating to the YJ output,</li><li>an NC indication which is decremented by one unit and re-registered as a new NC indication, if it is not zero; if the value zero is obtained, which means that the desired number of read operations for this sub-cell (providing the desired number of copies thereof) has been carried out, the DC circuit generates a QC signal allowing the circuit FMLMC to record that the ROSA address buffer location can be freed and included in the set of free buffer locations;</li><li>a value L which is equal to zero since it is not the last cell and controls the aforementioned transfer of the value NCB of the memory SLM to the field WCB of the memory OSPM, via the multiplexer SO.</li></ul></li></ul>
In the case of an ISC intermediate cell, the same operations take place in the OSPM and SLM memories and in the FMLMC circuit as for the reading of a first FSC sub-cell.
In the case of a last LSC sub-cell, the OSPM memory again supplies the ROSA address of the sub-cell to be transmitted, a last LSC sub-cell in this case, which is used for:<ul id="ul0014" list-style="dash" compact="compact"><li>address the buffer memory BM for reading the last sub-cell,</li><li>select the SLM memory for reading, which provides:<ul id="ul0015" list-style="dash" compact="compact"><li>an NC value which is decremented and processed exactly as in the other sub-cell read cases,</li><li>a value L, now equal to 1, indicating that there is a last sub-cell LSC, which, in this particular case, prohibits the transfer of the value NCB from the memory SLM to the location WCB OSPM memory, since there is no chaining to a next sub-cell supplied by the NCB value of the SLM memory in the case of a last LSC sub-cell; instead, L = 1 is supplied to the COQML logic to indicate that the output considered YZ becomes available for sending a next cell, as of the next cycle, a last sub-cell of a cell being 'sending.</li></ul></li></ul>
Then, after selection by the COQML logic of the appropriate cell which must be transmitted on the considered output YJ, the COQML logic initializes the value WBC in the OSPM memory for the YJ output, by writing there the address of location buffer of the first FSAO sub-cell of the newly selected cell, before the next cycle concerning the output YJ. Since this initialization process is not carried out during the clock period YJ of the last sub-cell LSC, an asynchronous access to the OSPM memory is used, by means of the output address YS supplied. by COQML logic.
While this exemplary embodiment of the SBML sub-cell buffer management logic has been described to illustrate the principles of sub-cell buffer management employed for the transfer of cells composed of sub-cells between any input and any output (or outputs) of the elementary switch, other embodiments of the functions of this SBML management logic, for example as regards the management circuit for free buffer memory locations FMLMC, are also included in types of elementary switches according to the present invention.
FIG. 4 represents an embodiment of the routing circuit RL arranged for use in the elementary switch ISE of FIG. 1, and also allowing bidirectional operation in accordance with what is illustrated in FIG. 3.
The circuits of FIG. 4 receive, in an IR register, the cell label CCH mentioned in FIG. 1, to output the information marking the control link RMD which provides the cell management logic and selection of outputs information specifying the selected routing mode (RS, MC, DI, ES, PH), as well as routing data (RG, PO)
The routing mode information is as follows:<ul id="ul0016" list-style="dash" compact="compact"><li>an RS "group" mode signal, which is present when the cell is to be retransmitted on one of the outputs of a group of outputs, in the case of point-to-point routing,</li><li>a "broadcast" mode signal MC, which is present when the cell is to be routed on one of the outputs of each of several groups of outputs, in the case of point-to-multipoint routing,</li><li>a “distribution” mode signal DI, which is present when the cell must be retransmitted on one of the outputs of a set of outputs, in the sense explained with reference to FIG. 3, in the case of an elementary switch bidirectional, or on one of all the outputs of the elementary switch, in the case of a unidirectional elementary switch, thereby achieving a general distribution, aimed at mixing the cells received by the switching network,</li><li>an "service" mode signal ES, which indicates that the cell received is intended for a particular control output,</li><li>a "directed transfer" mode signal PH, which indicates that the cell must be retransmitted on a predetermined output, for test reasons, for example.</li></ul>
The RMD link routing data includes:<ul id="ul0017" list-style="dash" compact="compact"><li>group identity signals RG which identify the group or groups on an output from which the received cell is to be retransmitted, for the RS and MC routing modes,</li><li>PO individual exit identity signals used with PH routing mode.</li></ul>
The circuits of FIG. 4 also receive, according to the input on which a received cell arrived, an incoming direction indicator IO which is for example provided by the reception circuit supplying the cell in question, on the multiplexer entry of figure 1, specifying the incoming direction concerned, in the sense mentioned with reference to figure 3.
The circuits of Figure 4 include the following:<ul id="ul0018" list-style="dash" compact="compact"><li>the IR register already mentioned, to receive the CCH label of each cell received, which includes, as indicated, the RCC, RCA and IRN information,</li><li>an RCCTM command translation memory, recording 32 words of 16 bits, called routing parameters, each comprising a three-bit MT mode mode code, an EF reflection indicator or bit, a routing group field "incoming" RPI, 6-bit and an "outgoing" RPO routing group field, also 6-bit,</li><li>a broadcast memory MCM, recording a plurality of 8-bit MSK mask words, one bit per group of outputs, each identifying the different routing groups to which a copy must be sent,</li><li>a routing mode decoder TD, decoding the routing mode code MT and consequently supplying one of the five mode signals mentioned above,</li><li>a direction selector RD selecting either the "incoming" RPI routing group field or the "outgoing" RPO "routing group field of the command translation memory RCCTM, as a function of the reflection bit EF and of the incoming direction indicator IO,</li><li>an output group selector MS having two 8-bit inputs in parallel and which supplies the group identity signals RG, also 8-bit, each of the bits of which corresponds to a distinct group among the 8 possible routing groups,</li><li>a 14-bit shift register SR having a five-conductor PO output; in the case where the routing mode is the "physical" PH mode, this output identifies the output to which the received cell must be routed,</li><li>a GD routing group decoder,</li><li>an OR-exclusive XOR door,</li><li>and two AND gates, AN1 and AN2.</li></ul>
The logic routing circuits of FIG. 4 operate as explained below, when the header of a received cell is present on the input multiplexer (FIG. 1), the label CCH of which is supplied to the register IR, while the IO bit indicates the incoming direction of routing. As indicated above, a clock punctuates the operation of the circuits, in an appropriate manner, in accordance with current practice in the matter.
The RCC control information, characteristic of a transfer sequence through the switching network, does not directly indicate the routing mode to be applied in the elementary switch considered. This routing mode depends on the type of switching network and the position of the basic switch in it.
The control information, to be interpreted, is used as the address for reading, in the command translation memory RCCTM, the routing parameters comprising the elements MT, EF, RPI and RPO defined above.
The routing mode code to be applied MT is decoded by the routing mode decoder TD which, accordingly, provides one of the mode signals RS, MC, ES, DI or PH.
The incoming direction indicator IO is applied to one of the inputs of the XOR exclusive OR gate, while the reflection bit EF is applied to its other input. The output of gate XOR provides the control signal for the outgoing direction selector RD. The latter selects either the "incoming" RPI routing group field or the "outgoing" RPO routing group field, specifying, for both sets of outputs, a specific part of the address RCA destination intended to provide the identity of a routing group on an output of which the received cell must be retransmitted. Each of these fields includes a 4-bit POS position indicator and a two-bit RGS dimension indicator. The position indicator POS controls the shift register SR so that the RCA information is shifted there and that a part of three bits which it contains comes in the three stages on the left, in the figure, of this register SR , or that a part of five bits which it contains comes in the five stages on the left, in the figure, of the register SR. The RGS dimension indicator indicates how many of the three bits mentioned first should be used to define the identity of a routing group. Thus, the left bit of these three bits is transmitted directly from the shift register SR to the group number decoder GD, the next bit is transmitted by the AND gate AN1 conditioned by one of the signals RGS and the third bit by the AND gate AN2 conditioned by the other of the RGS signals. The group number decoder GD supplies an 8-bit word which constitutes the identity of a routing group, applied to the selector MS. In this word, only one bit is at 1, for example, all the others being at O.
Simultaneously, the internal 14-bit broadcast tree reference number IRN is applied by the IR register to the broadcast memory MCM, where it serves as an address for reading an 8-bit mask word MSK. As indicated previously, this mask word identifies one or more routing groups in an 8-bit word, one or more of which is at 1, and the others at 0. It is also applied to the selector MS.
If the mode signal supplied by the decoder TD is the "group" mode signal RS, the selector MS supplies at its output a group identity signal RG which is the signal supplied by the decoder GD; if it is the "broadcast" mode signal MC, the signal RG transmitted by the selector MS is the signal MSK.
Furthermore, the five stages on the left of the shift register SR, after the offset caused by the position indicator POS, directly provide the identity PO of an output to which the received cell must be retransmitted, in the case of the mode "physical" routing PH.
In the particular case of DI distribution mode, no group of outputs needs to be identified, since it is a single group comprising all the outputs of the set considered in the direction of transfer concerned .
In the case of the "service" ES mode, the output concerned is directly known, since the cell received is intended for a particular control output shown in FIG. 1.
It is thus seen that the information contained in the command translation memory RCCTM defines, in each elementary switch, the interpretation that this switch must make of the 32 possible transfer sequences designated by the command information RCC to determine the mode d routing to be applied to the routing data contained in the label of the received cell. This amounts to combining the routing information of a cell, unchanged while the cell crosses switches of different stages of the switching network, with the routing parameters of the switch, derived from its position in the network, for example its own on each floor and leading to a particular routing mode in each floor and for each routing sequence.
The information contained in the RCCMT command translation memory is semi-permanent and can be entered when each elementary switch is put into service. On the other hand, the information contained in the broadcasting memory MCM must be modified during operation, for the establishment of each broadcasting tree.
FIG. 4bis represents the general diagram of the cell management logic and selection of COQML output of FIG. 1.
When the routing decision of a new cell received on an input of the elementary switch is taken by the routing circuit RL, this circuit provides the COQML logic with the routing mode information, as well as the data of associated routing, on the control link RMD, this command being validated by the control signal FO received from the circuit SL, which indicates the presence of a first FSC sub-cell containing the information intended for the routing of the cell during reception. In addition, simultaneously, the management logic of sub-cell memories SBML supplies the logic COQML with the WISA address of the buffer memory BM, in which this first sub-cell FSC has been recorded.
When an output YJ of the elementary switch transmits the last sub-cell LSC of a cell and will therefore become available to transmit a next one, the logic for managing memories of sub-cells SBML indicates a request for a next cell by means of the signal L then active, as explained above, in the description of the SBML logic. Then, the COQML logic selects the next cell to be transmitted on this output, by supplying the SBML logic with the FSAO address of the first FSC sub-cell of the cell to be transmitted on the output YJ, this latter indication being notified by the YS output address also supplied by COQML logic to SBML logic, in order to be able to carry out this operation outside the synchronous clock time YJ relative to the YJ output.
The cell management and COQML output selection logic includes the following circuits:<ul id="ul0019" list-style="dash" compact="compact"><li>BQ1 / BQZ queues whose respective inputs come from a BI demultiplexer and the respective outputs are connected to a BO multiplexer, and which temporarily store the identities of cells awaiting output, respecting the discipline of first entry / first out, said identities being characterized for example by the addresses of the first sub-cell of each cell in the buffer memory,</li><li>QICL input control logic which receives cell queuing requests,</li><li>a QOCL output control logic which selects each next cell to be transmitted on one of the outputs of the elementary switch, as soon as it becomes available.</li></ul>
In addition to managing the temporary waiting of cells to be transmitted, by storing their identities in queues, COQML logic also ensures the selection of an individual output in each selected routing group, given that for the modes point-to-point routing RS, point-to-multipoint MC, and distribution DI, the routing circuit RL identifies only the routing groups to which a copy of the cell must be transmitted, or all of the outputs in one direction in DI mode.
In a first embodiment, this individual output selection function is performed by the QICL input control logic, before the identity of the cell is queued. In this case, each queue BQ1 / BQZ is directly associated with each of the Y outputs of the elementary switch.
Another equivalent embodiment consists in performing the same output selection function by the QOCL output control logic, therefore after the cell identity has been queued. In this second case, each queue BQ1 / BQZ is associated with a routing group comprising one or more outputs and not with an individual output of the elementary switch.
In either variant, the output selection device necessary for the RS, MC and DI routing modes can be produced in a known manner on the basis of a cyclic distribution of the cells at the outputs of a routing group considered, which enables a homogeneous distribution of the traffic load of cells of the routing group on each of its outputs. Another solution proposed is to use a pseudo or quasi-random signal generator to select an output for each cell, which makes it possible to eliminate, at least in large part, any correlation between the fluxes of cells on the inputs and the outputs of each elementary switch.
In the case of ES or PH modes, an output of the elementary switch is respectively implicit or already selected and the role of the COQML logic is limited to the function of managing these cells in queues by corresponding individual output.
We will now, with reference to FIGS. 5 to 12, give several examples of embodiment of the switching network of the invention, based on FIGS. 1 to 4, as regards the specific characteristics of the elementary switches used in the different stages of these switching networks.
Indeed, the properties of a switching network come from the properties of the elementary switches which constitute it, as they are arranged, according to the invention, by the semi-permanent routing parameters derived from the position of the switches and from the configuration of the switching network, essentially comprising the number of stages, the unidirectional or bidirectional characteristic of each stage, the mesh between the elementary switches and the method of connecting the input and output ports.
The invention, as will be seen, applies in all the configurations of switching networks which will be described, as well as in the numerous variants which can easily be deduced therefrom.
It should also be pointed out that, in any configuration of a switching network, it is generally desirable, for reasons of standardization and ease of extension, that the same type of elementary switch is used in all the stages of the network. It will easily be verified that the elementary switch of FIGS. 1, 3 and 4 meets this need thanks to their ability to be initialized with specific routing parameters, for example by stage, for each cell transfer sequence across the network. of commutation.
FIG. 5 represents a switching network RC1, unidirectional and symmetrical, consisting of elementary switches such as that of FIG. 1, arranged in three stages comprising, the first, switches TSi1 to TSiT, having n inputs each, the central stage , switches AS1 to ASk, and the last, switches TSo1 to TSoT having n outputs each. In this way, the switching network has N = nT input ports connected to the inputs of the elementary switches of the first stage and N = nT output ports connected to the outputs of the elementary switches of the last stage. Cell traffic is routed from ingress ports to egress ports through all switches in one direction; this is why this network is said to be unidirectional. It is said to be symmetrical because the number of outputs is the same as that of the inputs. The T switches of the first stage have one or more (m) meshes towards each of the k switches of the central stage, ie mxk outputs. The T switches of the last stage have one or more (m) meshes coming from each of the switches of the central stage, ie mxk inputs. The switches of the central stage thus have T xm inputs and T xm outputs.
Using the figures previously advanced (FIG. 3, for example), the elementary switches of the three stages can be switches with 32 inputs and 32 outputs, with T xm = kxm = 32.
Each of the switches of the central stage being able to reach all the switches of the last stage, a cell reaching the switching network on any input port, of the elementary switch TSi1, for example, can be addressed to any of the switches of the central stage, AS1 to ASk, whether the routing of this cell is of point-to-point or point-to-multipoint type. In such a network, provision will therefore be made, according to the invention, for the outputs of the first stage switches to be arranged in a single group of outputs and for a cell received by a first stage switch to be point-to-point. point or point to multipoint, will be retransmitted on an output selected in this unique group of outputs. It suffices for this (see description relating to FIG. 4) that the command translation memory (RCCTM), in these elementary switches TSi1 to TSiT, provide the "distribution" mode signal DI in exchange for the RCC command information signifying that the cell must be distributed on any output to the next floor, for point-to-point or point-to-multipoint routing.
In other words, a position data item included in the switches of the first stage (implicitly represented by the routing parameters contained in the command translation memory) will allow an interpretation in this sense of the routing information of the cell.
On the other hand, with regard to the switches of the central stage, each of them has only one or more (m) meshes towards each switch of the last stage. The routing depends on the identity of the recipient exit. A group of one or more (m) meshes is accessible for this purpose. Each central stage switch will thus have T groups of one or m outputs. With a different position data, these switches will interpret the same cell routing information, so as to select the appropriate routing group, in the case of point-to-point routing, or the appropriate groups in the case of '' a point-to-multipoint routing interesting the outputs of several different elementary switches of the last stage. According to the embodiment of FIG. 4, the control information RCC considered above will be translated, in these elementary switches of the central stage, into a "group" mode signal RS, for point-to-point routing, or in an "MC" mode signal, for point-to-multipoint routing, while the RCA output address, or the IRN internal reference number, will be used to identify the selected group or groups of outputs.
What applies to the elementary switches of the central stage, relative to the routing cases considered, also applies to the switches of the last stage.
In the case where m = 2, for example, the switches of the central stage have two meshes with each of the switches of the first and of the last stage. In this case, the elementary switches of the central stage will have T groups of two outputs between which it will be necessary to make a selection.
The number of elementary switches of the central stage can, according to another variant, be greater than that of the switches of the two extreme stages, all other things being equal. This will reduce the traffic load on the internal mesh of the switching network. Starting from the previous example (m = 2), the central stage could comprise 64 elementary switches, and, correlatively, the switches of the two extreme stages would have 128 outputs or inputs.
It can also be envisaged that, for a given dimensioning of the network of FIG. 5, in terms of the number of elementary switches per stage and the number of links by mesh between them, certain input links with transmission speed higher than that practiced in switching network, either coupled to several input ports on which the cells of this external link are distributed; these are then transferred individually by the multiplicity of possible paths to the destination output ports. By generalizing, this can make it possible to produce a switching network whose transmission and switching speed would be lower than that of the external transmission links that it serves. Of course, what has just been said about the input links applies symmetrically to the output links where the cells of several output ports are multiplexed towards the outgoing external link.
The symmetrical unidirectional switching network of FIG. 6 comprises 4 stages. The designations of the elementary switches of the two extreme stages are the same as in FIG. 5. The elementary switches of the two central stages are respectively referenced ASi1 to ASik and ASo1 to ASok.
The elementary switches of the first two stages form an input selection unit USi, while the elementary switches of the last two stages form an output selection unit USo. These two selection units are coupled to each other by meshes joining the homologous outputs and inputs.
All that has been exposed relative to the network of FIG. 5 still applies here, considering that two homologous elementary switches of the two central stages form only one. However, from the point of view of control, the elementary switches ASi1 to ASik will advantageously be arranged to operate a distribution, whether the routing mode is point to point or point to multipoint. The switches ASo1 to ASok, for their part, from the control point of view, will be treated like the elementary switches AS1 to ASk of FIG. 5.
The switching network of FIG. 7 is similar to that of FIG. 6, but folded, therefore bidirectional, confusing the elementary switches of the extreme stages, TSi1 to TSiT and TSo1 to TSoT, into switches of the first stage TS1 to TST, while the designation of the switches of the second stage of FIG. 5, AS1 to ASk, has been used for the elementary switches of the second stage of FIG. 7.
The inputs of the first stage switches are shared between the input ports and the output ports of the connection network (by halves for a network without expansion / concentration). The internal meshes are double, comprising one or more connections for each direction of routing. The switches are bidirectional and they accomplish the reflection, either in the first stage, or in the second stage.
In the routing of a cell from an input of the elementary switch TS1 to one or more outputs of another switch TST, whether within the framework of point-to-point or point-to-multipoint routing, the switch TS1 distributes to all of the switches in the next stage, while that of the switches of the second stage which routes the cell applies the "group" or "broadcast" routing mode and the same applies to the elementary switch TST. In the case of a cell transfer of an input from the elementary switch TS1 to one or more outputs of this same switch TS1, a transfer reflection can be carried out directly at the level of the first stage, that is to say in the bidirectional switch TS1 which transfers the cell directly to the destination (s), instead of transferring it normally for distribution to the next floor.
All that has been explained concerning the network of FIG. 5 therefore also applies to the network of FIG. 7, with the necessary conversions, taking into account the superposition of the two directions of traffic routing in the switches of the two stages and the execution of a reflection in the switches of the first or second stage.
Referring to Figure 3, the first stage switches include 16 inputs and 16 outputs connected respectively to input ports and to output ports. They also include 16 inputs and 16 outputs connected to 16 / m switches of the second stage, when m is the number of internal meshes between two switches belonging respectively to each of these two stages. If the first stage comprises 16 switches (case where m = 1), 16 inputs and 16 outputs of the switches of the second stage are respectively connected to these 16 switches of the first stage, by 16 bidirectional meshes each comprising a single link in each direction of routing. The traffic coming from these 16 exits is then reflected towards these 16 exits. The other 16 inputs and outputs of the second stage switches are not used and are available for network expansion by adding a third stage. If the first stage includes up to 32 elementary switches, the 32 inputs and 32 outputs of the switches of the second stage provide the 32 meshes with two links, one in each direction of routing, which are necessary to reach them.
The operating mode of the switches in the treatment of the two directions of routing has already been explained with reference to FIGS. 3 and 4.
Reference will now be made to FIG. 8 which represents a switching network derived from that of FIG. 5, but here obtained by juxtaposition of two separate unidirectional switching networks but homologous stage by stage, each allowing the transfer in a given direction, of N1 inputs on one side to the N2 outputs on the other side, or N2 inputs on the latter side to the N1 outputs on the first side. In fact such a switching network can typically be used to constitute an asymmetrical assembly interconnecting N1 bidirectional link on one side to N2 bidirectional link on the other side, with N1> N2. The unidirectional network from N1 to N2 therefore achieves a concentration of traffic and the other unidirectional network from N2 to N1 an expansion of traffic. Each of these unidirectional networks therefore differs from that of FIG. 1 only in that it is asymmetrical due to the presence of at least one stage of asymmetrical elementary switches, that is to say of which the numbers d 'inputs and outputs are different, for example 32x16 or 16x32. Apart from this configuration variant, the routing principles described for each stage of the network in FIG. 5 remain applicable to the corresponding stages in each of these two asymmetrical unidirectional networks.
FIG. 9 represents an equivalent of the switching network of FIG. 8, in a bidirectional configuration. In this case, the asymmetric switching network interconnects the N1 input and output ports on one side to the N2 input and output ports on the other side. Assuming N1> N2, such a network is typically applied to the concentration of N1 links with relatively low traffic to N2 links with higher traffic. Cell transfers can therefore be carried out between these two sets of ports N1 and N2, ie unidirectional, from an input port of the set N1 (or N2) to an output port of the set N2 (or N1 ), or bidirectional between an input and output port of the N1 assembly and an input and output port of the N2 assembly. In addition, the presence of at least one bidirectional stage also makes it possible to transfer cells between an input port and an output port of the same assembly N1 (or N2), by performing a reflection in a bidirectional stage. From the configuration point of view, the dimensioning of this network is similar to that of FIG. 8; as in the latter, at least one of the stages is asymmetrical, so that the switching network has different numbers of input and output ports N1 and N2 on either side of the switching network.
The different transfer sequences are applied as follows:<ul id="ul0020" list-style="dash" compact="compact"><li>For an unreflected cell transfer between an entry port on one side of the network (set N1 or N2) and an exit port on the other side of the network (set N2 or N1), the first stage distributes the traffic entering all the switches of the intermediate stage. Then, the latter performs a selective routing to the last stage by transferring the cell to one or more groups of outputs leading to one or more switches of the last stage. In the latter, a selective routing makes it possible to transfer the cell to the destination output port or ports.</li><li>In the case of a reflected transfer, between an input port and an output port of the same set N1 or N2, the reflection can occur in the switch of the first stage, if the ports are connected to the same switch and s' it is bidirectional. Otherwise, the first stage switch distributes the cell to any one of the intermediate stage switches, which will advantageously be bidirectional to allow this type of transfer reflected by selective routing to the first stage switch (s). In the latter, selective routing allows the cell to be transferred to the destination outgoing port (s).</li></ul>
FIG. 10 shows a possible extension of the switching network of FIG. 9 by adding an additional selection stage. In addition, the first two stages of switches, on the side of all the N1 input and output ports, are formed into m two-stage selection units, with bidirectional routing, each similar, except for the dimensioning, to the network. of FIG. 7, the elementary switches being moreover identified by the same references as in this latter figure.
Apart from the addition of a fourth stage and the organization of the first two stages on the side of the NI port set, the types of transfer possible in such a bidirectional asymmetric switching network can be easily extrapolated from that of the figure 9:<ul id="ul0021" list-style="dash" compact="compact"><li>A transfer of cells not reflected between an input port of the set N1 and an output port of the set N2 is carried out by distribution of the first stage to any of the switches of the second stage of the selection unit , then the latter distributes to any switch on the third stage. In the latter, the switch performs a selective routing to one or more of the switches of the fourth stage, the latter performing a selective routing to one or more output ports of the set N2.</li><li>For a transfer of cells not reflected in the opposite direction, from the set N2 to the set N1, the first stage distributes each cell to one of the switches of the following stage. In this transfer direction, the latter performs selective routing to one or more of the m two-stage selection units, while freely choosing one of the switches of the third stage in each destination selection unit. In a switch of the third stage, a selective routing transfers the cell to one or more switches of the fourth stage, the latter carrying out a selective routing towards one or more output ports of the set N1.</li><li>For a reflected transfer between an input port and an output port of the same set N2 or of the same subset N'1 of a selection unit, the reflection occurs either on the first or on the second stage as in the Figure 9 network.</li><li>On the other hand, for a reflected transfer between an input port and an output port of the same set N1, but of two different selection units, the reflection is only possible at the third stage which interconnects the m selection units. In this case, the first stage distributes to the switches of the second stage; the latter also distributes to the switches of the third stage; the latter then reflects the transfer and performs selective routing to one or more destination selection units, while leaving the free choice between the switches of the second stage in each selection unit concerned. Then, the second stage switch performs selective routing to one or more switches of the first stage, the latter performing selective routing to one or more output ports of the set N'1 recipients.</li></ul>
This example of a switching network illustrates once again the fact that, according to the invention, the output groups formed in the switches of the switching network are not the same depending on the stage to which they belong and that they are required to do so. counts in the routing mode implemented, at the level of each stage, although the routing information of the cell remains the same in all the stages.
We will then turn to FIG. 11 which represents a unidirectional switching network composed of selection units which include input terminal units TSUi, selection planes PS and output terminal units TSUo. In each selection unit, there are elementary switches which can be of the type of the preceding figures, each represented by the usual sign of a switching matrix, with, on the left, the number of the inputs of the elementary switch and, on the right, the number of its outputs. These elementary switches are connected to each other by meshes.
Inside an input terminal unit, TSUi1 for example, there are two stages of elementary switches, the elementary switches TSi1 to TSi16 and the elementary switches ASi1 to ASi4. There is generally one or more meshes between an output of an elementary switch of the first stage and an input of an elementary switch of the second stage. The four outputs of a first stage switch, TSi1 for example, are then each connected to one or more inputs of each of the four elementary second stage switches. In the case of a single mesh, the 16 inputs of an elementary switch of the second stage, ASi1 for example, are each connected to an output of each of the 16 elementary switches of the first stage. The 16 times 4 inputs of the first stage elementary switches are connected to 64 input ports pi1 to pi64. The other input terminal units may be similar, except for the numerical values shown. In this example, the output terminal units are assumed to be arranged in the same way and symmetrically. Thus the output terminal unit TSUo1, for example, gives access, by the two stages of elementary switches comprising the elementary switches ASo1 to ASo4 and TSo1 to TSo16, to output ports po1 to po64.
The figure also shows TSUi128 and TSUo128 input and output terminal units, to indicate a total number of terminal units in the switching network.
The selection planes, such as the selection plan PS1, comprise three selection stages formed by the elementary switches PSi1 to PSi32, PSc1 to PSc16, PSo1 to PSo32. The arrangement of the internal meshes between a stage and the following one obeys the principle of that of the terminal selection units, by supposing in this example a number of meshes between switches equal to one; it will not be described in detail.
16 selection plans are planned PS1 to PS16. The 16 outputs of a second stage switch of an input terminal selection unit, TSUi1 for example, are individually connected, by 16 meshes, to an input of each of the 16 selection planes. The 4 outputs of the same rank of the four elementary switches of an input terminal unit, TSUi1 for example, are connected to successive inputs of the same elementary switch, PSi1, for example of a selection plane, PS1 in l 'occurrence. Therefore, the 512 inputs of a selection plan, PS1 for example, are connected, four by four, to the four elementary switches of the second stage of each of the 128 input terminal units.
The arrangement of the meshes between the outputs of the elementary switches of the third stage of the selection planes, PSo1 to PSo32 for the selection plane PS1 for example, and the inputs of the elementary switches of the first stage of the output terminal units is symmetrical with that we just described.
When the whole of the switching network is symmetrical with respect to the central stage of the selection planes, that is to say when the numbers of inputs and outputs of switches and the numbers of homologous meshes by vertical central symmetry are identical, it is possible to produce a folded equivalent switching network, with bidirectional switches at least for part of the stages, as shown in FIG. 12 and described below. Each central elementary switch, such as PSc1, is connected, by three switching stages, on each side, to all the input ports and to all the output ports. Conversely, between any input port and any output port, there are in this example more than 4000 (4 k) separate paths passing through one of the four switches ASi of the incoming selection unit, one of the 16 selection planes PS, one of the 16 central switches PSc in a plan, and one of the four switches ASo of the outgoing selection unit. Given the total accessibility of each input port to all of the central switches PSc in all the plans, the transfer to this first part of the network achieves a generalized distribution of all of the incoming traffic on the set of 16x16 central switches PSc, therefore a complete mixing of all incoming cell traffic.
Then, from the central stage PSc to the output ports, the routing is necessarily selective, in order to reach the destination output port or ports. If it is a question of reaching several distinct output ports, in a point-to-multipoint routing, this selective routing must include several branches with one or more stages.
Referring now to the preceding descriptions, relating in particular to FIGS. 1 and 3, we will define how the different routing modes are applied in the elementary switches of the network of FIG. 11.
We will first consider a point-to-point routing, for example between the input port pi1 and the output port pol. In the cell label, a routing mode data item of the RCC command field specifies point-to-point routing. The RCA output address comprises 7 bits designating the terminal unit TSUo1 and 6 bits designating the output port pol in the terminal unit.
In the elementary switches of the first stage of the switching network, such as TSi1, the routing parameters are such that the cell is retransmitted on one of the set of all the outputs of the elementary switch. We have seen the conditions of this selection previously. For example, the cell is thus retransmitted towards the elementary switch ASi1.
In the elementary switches of the second stage of the switching network, such as ASi1, the routing parameters have the same effect as in the first stage and the cell is thus retransmitted on one of the set of all the outputs of the switch elementary, for example that which leads to the plane PS1 and therefore, therein, towards the elementary switch PSi1.
The same applies in the elementary switches of the third stage of the switching network and the cell reaches for example the elementary switch PSc1.
From the central floor, the routing becomes selective, at least in part.
The routing parameters of the switches in the central stage of the switching network are such that the switch PSc1 selects a group of four terminal units in which the destination terminal unit is located, on the basis of 5 bits, from among the 7 designation bits of the terminal selection unit, which designate the routing group, in the example shown a single output, leading to that of the 32 elementary switches of the fourth stage of the switching network which, in the plane PS1, accesses the destination terminal unit. Thus a mesh is selected leading to the elementary switch PSo1.
The operating mode is similar to that which we have just seen, in the elementary switches of the fifth stage of the switching network. The routing parameters are different; they cause the selection of the 2 remaining bits of the identity of the destination terminal unit, which identifies a routing group comprising the four outputs leading to the four elementary switches ASo1 to ASo4, according to the example considered. One of these four outputs is selected in the manner mentioned above. It leads the cell, for example, to the ASo1 switch.
In the elementary switches of the sixth stage of the switching network, the routing parameters select, from the 6 bits of the RCA address designating the output port, the 4 bits identifying the elementary switch of the last stage serving this output port. The cell is thus led to the elementary switch TSo1, in accordance with the example chosen.
Finally, similarly, in the basic switch of the last stage of the switching network, the routing parameters will allow the cell to be transferred to the output port po1.
During this point-to-point transfer sequence, the routing parameters of the elementary switches of the different stages therefore first made it possible to conduct the cell in a non-selective manner towards any elementary switch of the central stage, then of the drive selectively, using successive parts of the RCA output address, to the specified destination.
It is easy to verify that, since all the elementary switches of the central stage, in a selection plane, see the 32 elementary switches of the fifth stage of this selection plane in the same way, the routing is identical in each of them. Likewise, insofar as all the selection plans see the output terminal units in the same way, it can be concluded that all the elementary switches of the central stage carry out the routing in the same way. A similar reasoning leads to the same conclusion with regard to the elementary switches of the other stages, from the fifth to the last. The conclusion is that, in such an example of a folded switching network, the routing parameters depend only on the identity of the stage in which the elementary switch is included, not on its position in the stage.
On the other hand, they could also depend on the position of the elementary switch in the stage, in certain variants of this type of switching network, for example such as all of the terminal units or all of the selection plans is not not composed of identical configuration selection units, as is the case, for example, during switching network extension operations requiring progressive switching from one configuration to another.
Furthermore, it should be emphasized that successive cells, from the same source and with the same destination can, mainly thanks to the routing routing practiced in the first stages of the switching network, take a large number of different routes, which achieves a mixing of regular and irregular traffic currents, favorable to the more homogeneous flow of the flow rates of various cells subjected to the switching network and therefore to the relative performance of cell transfer.
Point-to-multipoint routing, in such a network is carried out on the same bases, except that in some or all stages with selective routing, there is a multiplication of the cell with retransmission on groups of Different outputs in the floor switches where the diffusion tree indicates that several outgoing branches are required to the next floor.
This example illustrates how point-to-multipoint transfers according to pre-established broadcasting trees can be implemented according to the invention in multipath and self-routing networks, while retaining for this type of transfer the possibility of choosing a path. from one to several arbitrary ports among the multiplicity of possible paths through the switching network, thanks to a characteristic organization of the content of the memories of diffusion trees of the elementary switches of the different stages. According to the invention, it is possible to define the corresponding connection points so that no unnecessary copy is generated on any stage, which makes it possible to avoid any internal overload of the internal meshes between stages. This characteristic is carried out according to the following principles:<ul id="ul0022" list-style="dash" compact="compact"><li>the absence of copies in the distribution stages;</li><li>the same content (branch points) of broadcast tree memory in all the switches of a stage which belong to the same set of equivalent multipaths for routing the cells to the switch groups of the following stage;</li><li>in these "equivalent" switches which carry out selective routing on different distinct routing groups, a connection to several predetermined routing groups (in the broadcasting tree memory) is carried out by transferring a copy of the incoming cell only to the routing groups marked as required connections in these switches among the set of possible routing groups. Thus, no unnecessary copies are generated on each floor.</li></ul>
FIG. 12 represents the folded version of the network of FIG. 11, obtained by using at least part of the stages of bidirectional switches, except for the central stage PSc which remains unidirectional. From the analogy of the notations used, it is clear that the homologous switches can now be confused into bidirectional stages, namely:<ul id="ul0023" list-style="dash" compact="compact"><li>the previous stages 1 and 7 (TSi and TSo) become the first bidirectional stage (TS),</li><li>the previous stages 2 and 6 (ASi and ASo) become the second bidirectional stage (AS),</li><li>the preceding stages 3 and 5 (PSi1 and PSo1) become the third bidirectional stage (PSa).</li></ul>
On the other hand, the central stage PSc remains unidirectional and this fourth stage is called the mirror stage (compulsory reflection).
All the characteristics and properties of the unidirectional network in Figure 11 can be easily transposed to that of the folded version:<ul id="ul0024" list-style="dash" compact="compact"><li>The same number of possible paths exists, without counting the additional paths made possible by the intermediate reflections described below.</li><li>The two main steps, during the transfer of a cell, general distribution, then selective routing, are also carried out by now referring to a first transfer part in the incoming direction to the reflection stage and to a second part transfer in the outgoing direction from the latter to the first floor, the routing operations on each stage being easily transposed by symmetry and being based on the same principles as those which have been described for the network of FIG. 11.</li></ul>
However, the folded variant of FIG. 12 has the following additional characteristics derived from the intrinsic faculties of reflection possible in a bidirectional switch as described above:<ul id="ul0025" list-style="dash" compact="compact"><li>In the first stage of a point-to-point transfer, where the cell is freely distributed in the incoming direction to one of the switches of the next stage, at each bidirectional stage, potentially TS, AS and PSa, premature reflection is possible whenever the destination output port is accessible by the switch in question, and of course if the latter is bidirectional. Such premature thinking is therefore possible:<ul id="ul0026" list-style="dash" compact="compact"><li>on the first stage TS if the destination port belongs to the group of ports connected to the switch considered TSx,</li><li>on the second stage AS if the destination port belongs to the terminal unit in which the switch considered ASx is located,</li><li>on the third stage PSa if the destination port belongs to the group of 4 terminal units to which the switch in question TSax is connected.</li></ul></li><li>Said possibilities of premature reflection induce the following characteristic properties:<ul id="ul0027" list-style="dash" compact="compact"><li>a relative reduction in the load of the internal cells insofar as part of the cell traffic does not cross all the stages of the switching network,</li><li>an increase in the number of possible paths, namely the paths reflected prematurely,</li><li>the possibility of sub-equipping the switching network of the figure in number of stages and of carrying out successive extensions of stages without modification of wiring between stages, insofar as each bidirectional stage 1, 2 or 3 may temporarily constitute the last equipped floor and then carry out the compulsory refl ections of a mirror floor.</li></ul></li></ul>
It is obvious that the foregoing descriptions have been given only by way of nonlimiting example and that numerous variants can be imagined, without thereby departing from the scope of the invention. Numerical details, in particular, may change with each application case.
13 sheets
Sheet 1 Sheet 2 Sheet 3 Sheet 4 Sheet 5 Sheet 6 Sheet 7 Sheet 8 Sheet 9 Sheet 10 Sheet 11 Sheet 12 Sheet 13
Every citation, both waysCites: the store holds 3 of 4
| Document | Relation | Office |
|---|---|---|
| EP0229299A | Cites | European Patent Office (EPO) |
| EP0241152A | Cites | European Patent Office (EPO) |
| WO8603355A | Cites | World Intellectual Property Organization (WIPO) |
| IEEE TRANSACTIONS ON COMMUNICATIONS. vol. 36, no. 6, juin 1988, NEW YORK US pages 734 - 743; J.S. Turner: "Design of a Broadcast Packet Switching Network" | Non-patent | – |
| AT & T TECHNICAL JOURNAL. vol. 68, no. 2, mars 1989, NEW YORK US pages 36 - 50; J.J. Degan et al.: "FAST PACKET TECHNOLOGY FOR FUTURE SWITCHES" | Non-patent | – |
| IEEE Pacific Rim Conference on Communications, Computers and Signal Processing, June 1st- 2nd, 1989 Victoria,BC, Canada, pages 620-623,New York ,US; G.B.Mund et al.:" A 2x2 Switching Element for Broadband ISDN " | Non-patent | – |
19 members in 11 offices
Priority claims5
| Document | Office | Kind | Date |
|---|---|---|---|
| 9003246 | France | A | |
| 9003246 | France | A | |
| 9003246 | France | – | |
| 9003246 | – | – | – |
| FR19900003246 | – | – | – |
Members19
| Document | Office | Kind | |
|---|---|---|---|
| CA2051696A1 | Canada | A1 | |
| EP0446540A1 | European Patent Office (EPO) | A1 | |
| AU7283791A | Australia | A | |
| WO9114325A1 | World Intellectual Property Organization (WIPO) | A1 | |
| FR2659819A1 | France | A1 | |
| FR2659819B1 | France | B1 | |
| KR920702124A | Republic of Korea | A | |
| JPH04505843A | Japan | A | |
| US5237565A | United States of America | A | |
| AU641735B2 | Australia | B2 | |
| CA2051696C | Canada | C | |
| EP0446540B1This record | European Patent Office (EPO) | B1 | |
| AT136707T | Austria | T | |
| ATE136707T1 | Austria | T1 | |
| DE69026494D1 | Germany | D1 | |
| ES2088422T3 | Spain | T3 | |
| DE69026494T2 | Germany | T2 | |
| KR0146281B1 | Republic of Korea | B1 | |
| JP2996353B2 | Japan | B2 |
40 legal events, as 5 offices reported them to INPADOC
Over the term
Point at a mark for the eventEvents
| Event | Code | Office | |
|---|---|---|---|
| Lapsed in a contracting state [announced via postgrant information from national office to epo]LapsedPG25 | PG25 | EP | |
| Lapsed in a contracting state [announced via postgrant information from national office to epo]LapsedPG25 | PG25 | EP | |
| Lapsed in a contracting state [announced via postgrant information from national office to epo]LapsedPG25 | PG25 | EP | |
| Lapsed in a contracting state [announced via postgrant information from national office to epo]LapsedPG25 | PG25 | EP | |
| Announcement of lapse in spainLapsedFD2A | FD2A | ES | |
| Patent ceasedCeasedPL | PL | CH | |
| Discontinued because of reaching the maximum lifetime of a patentV4 | V4 | NL | |
| Annual fee paid to national office [announced via postgrant information from national office to epo]GrantedPGFP | PGFP | EP | |
| Annual fee paid to national office [announced via postgrant information from national office to epo]GrantedPGFP | PGFP | EP | |
| Annual fee paid to national office [announced via postgrant information from national office to epo]GrantedPGFP | PGFP | EP | |
| Annual fee paid to national office [announced via postgrant information from national office to epo]GrantedPGFP | PGFP | EP | |
| Annual fee paid to national office [announced via postgrant information from national office to epo]GrantedPGFP | PGFP | EP | |
| Annual fee paid to national office [announced via postgrant information from national office to epo]GrantedPGFP | PGFP | EP | |
| Annual fee paid to national office [announced via postgrant information from national office to epo]GrantedPGFP | PGFP | EP | |
| Annual fee paid to national office [announced via postgrant information from national office to epo]GrantedPGFP | PGFP | EP | |
| Se: european patent has lapsedLapsedEUG | EUG | EP | |
| Be: lapsedLapsedBERE | BERE | EP | |
| Lapsed in a contracting state [announced via postgrant information from national office to epo]LapsedPG25 | PG25 | EP | |
| Lapsed in a contracting state [announced via postgrant information from national office to epo]LapsedPG25 | PG25 | EP | |
| Annual fee paid to national office [announced via postgrant information from national office to epo]GrantedPGFP | PGFP | EP | |
| Annual fee paid to national office [announced via postgrant information from national office to epo]GrantedPGFP | PGFP | EP | |
| European patent in force as of 2002-01-01IF02 | IF02 | GB | |
| No opposition filedOpposition26N | 26N | EP | |
| No opposition filed within time limitOppositionORIGINAL CODE: 0009261PLBE | PLBE | EP | |
| Information on the status of an ep patent application or granted ep patentGrantedSTATUS: NO OPPOSITION FILED WITHIN TIME LIMITSTAA | STAA | EP | |
| Definitive protectionFG2A | FG2A | ES | |
| Patent application publishedBA2A | BA2A | ES | |
| It: translation for a ep patent filedITF | ITF | EP | |
| It: translation for a ep patent filedITF | ITF | EP | |
| Gb: translation of ep patent filed (gb section 77(6)(a)/1977)GBT | GBT | EP | |
| Corresponds to:REF | REF | EP | |
| New agentNV | NV | CH | |
| New agentNV | NV | CH | |
| Designated contracting statesAK | AK | EP | |
| Corresponds to:REF | REF | EP | |
| (expected) grantORIGINAL CODE: 0009210GRAA | GRAA | EP | |
| First examination report despatched17Q | 17Q | EP | |
| Request for examination filed17P | 17P | EP | |
| Designated contracting statesAK | AK | EP | |
| Public reference made under article 153(3) epc to a published international application that has entered the european phaseORIGINAL CODE: 0009012PUAI | PUAI | EP |
Numbers
- Publication
- 0446540
- Publication, DOCDB
- 0446540
- Publication, EPODOC
- EP0446540
- Application
- 90401393
- Application, DOCDB
- 90401393
- Application, EPODOC
- EP19900401393
Titles3
- German
- Selbstleitweglenkendes Mehrwege-Vermittlungsnetzwerk zum Vermittlen von Zellen mit asynchroner Zeitvielfachübermittlung
- English
- Self-routing multipath switching network for switching cells in the asynchronous transfer mode
- French
- Réseau de commutation à trajets multiples et à auto-acheminement pour la commutation de cellules à multiplexage temporel asynchrone
Classification
- CPC, 6
- H04L12/5601
- H04L49/106
- H04L49/203
- H04L49/256
- H04L2012/5672
- H04L2012/5681
- IPC, 5
- H04L12 54
- H04L12 70
- H04L12 931
- H04L12 933
- H04L12 947
Designated states1
- Contracting states, 1
- Sweden
