AU635584B2

AU635584B2 - Local area network

Info

Publication number: AU635584B2
Application number: AU67046/90A
Authority: AU
Inventors: Gerard Le Bihan
Original assignee: Alcatel NV
Current assignee: Alcatel Lucent NV
Priority date: 1989-12-07
Filing date: 1990-11-29
Publication date: 1993-03-25
Anticipated expiration: 2010-11-29
Also published as: AU6704690A; FR2655795A1; EP0431511A1; CA2031708A1

Description

6358

S@*

S

060@ c 'cc.

cc a .c~

S.

S S

S

c *S0 U *c S ccc...

S

ccc...

S

cc..

S S S. S

S.

cc

ORIIGINAL.

COMMONWEALTH OF AUSTRALIA PATENTS ACT 1952-1969 COMPLETE SPECIFICATION FOR THE INVETION ENTITLED "LOCAL AREA NETWORK' t The following statement is a full description of this invention, including the best method of performing it known to us:- This invention relates to a local area network for data transmission, interconnecting stations and allowing them to share their resources at very high bit rates up to 600 Megabits per second.

There are several known types of local area networks (LAN). The first type comprises a bus to which all stations are connected in parallel. Standard IEEE 802.3 defines the bus access protocol. This access protocol consists mainly, for each station, of checking the availability of the bus before transmitting data to the latter. A second type of LAN, called token ring, comprises a ring-shaped link which transmits via each of the stations, the .S-Q latter repeating the information transmitted on the ring. When the bus is 9*Os available, a token made up of a binary word circulates through this ring and can be intercepted by the station wishing to transmit data to the ring. This protocol is described by standard IEEE 802.5. A third type of LAN comprises a token bus, the stations being physically connected to a bus and logically connected to a ring. The corresponding protocol is described by standard IEEE 802.4.

e* 6eSS These known types of LAN have the disadvantage of being relatively slow, since their operation is based on a quite complex dialogue, the duration of which depends both on the hardware technology and the length of the cables linking the stations. The bit rate achievable is in the order of Megabits/second. A fourth type of LAN comprises an integrated services dig- 9 •"ital private automatic exchange which switches synchronous channels. The various services are supplied by separate service providers, in particular one service provider for data packet switching. This type of service provider has an architecture which limits the data channel bit rate to Megabits per second.

An object of the invention is to provide a LAN which can achieve much higher bit rates, of up to 600 Megabits/second. A further object of the invention is to provide a LAN having, as support, an asynchronous time-division multiplexed (TDM) connection network, each station fitted with devices necessary to implement a protocol offering services for access to any given support, as defined in standards IEE 802, and to implement a data transmission protocol in the form of cells having a standardised formal for asynchronous TDM mode.

According to the present invention, there is provided a LAN for data transmission, comprising at least a first asynchronous TDM connection network, to establish the first semi-permanent virtual circuits, a plurality of stations, each of which is fitted with means for information interchange in the form of cells, in accordance with a communications protocol suited for asynchronous TDM connection networks, each cell comprising a label identifying e Q7 one of the semi-permanent virtual circuits; said means comprising at least one input and one output respectively coupled to one input and one output of the first transfer network.

ao em.

mPreferably, the LAN includes a second asynchronous TDM connection network, to duplicate the links between stations, and each station is fitted with devices capable of deteting transmission errors and overcoming this problem by selecting to transmit the information via either the first or the second network.

In order that the invention may be readily carried into effect, embodiments thereof will now be described in relation to the drawings, in which: Figure 1 represents the synoptic diagram of a first example of implementation of the LAN in accordance with the invention, to link 16 stations.

Figure 2 represents the synoptic diagram of a second example of implementation of the LAN in accordance with the invention, to link 128 stations.

Figure 3 represents a synoptic diagram of a third example of implementation of the LAN in accordance with the invention, to link 240 stations.

Figures 4 to 6 illustrate a data transmission procedure, and in particular the allocation of labels to the cells carrying the data.

Figure 7 illustrates the implementation of a protocol to remedy the transmission errors.

The implementation example shown in Figure 1 comprises 16 stations (1 to 16) and two asynchronous TDM switching matrices (17 and 18), each comprising 16 ports made up of one input and one output. Each of the stations includes a protocol processing circuit (19) for information interchange. Circuit (19) is fitted firstly, with one output and one input respectively connected to the input and the output of one of the ports of matrix and secondly, with one output and one input respectively connected to the input and the output of a similar port on matrix (18).

Matrices (17) and (18) are of a type known in the art of TDM switch telecommunications system. Each protocol processing circuit (19) transmits and receives data in the form of cells, in accordance with a communications protocol suited for asynchronous TDM connection networks. Each cell comprises a 1" label used to establish a virtual circuit to route it towards the destination station. In addition, each cell comprises a control word and a total control field used for error detection. Such a protocol is described for instance in the paper by J. Coupaye, G. Gastaud, G. LeBihan: :Nouveau protocole pour la *S transmission de donnes en cellules MTA" Revue des Telecommunications, vol 62, number 3-4 1988, pages 338-342.

This protocol provides for: transmission error detection, one cell at a time; detection of loss or addition of cells; fragmentation of messages into cells before transmission; and de-interlacing of cells associated with several messages, at reception time. This protocol performs the functions of sub-layer No. 1 of layer No. 2 of the logic link, in the International Standardisation Organisation reference model. Each circuit (19) can be a typical protocol processing circuit made up for instance of a network of known programmable logic cells.

Matrices (17) and (18) establish semi-permanent virtual circuits, between stations 1 to 16, providing for two types of communications: bidirectional communications from one station to another single station; and broadcast communications from one station to all the other stations. The label structure used to establish these semi-permanent virtual circuits will be described later.

Matrices (17) and (18) operate in parallel. Protocol processing circuit (19) implements a protocol used to detect transmission errors and to overcome them, thanks to the duplication of virtual circuits thus performed. This protocol will be described later. The same label structure and the same protocols can be implemented for any number of stations, and with different network architectures. Figures 2 and 3 represent other examples of network architecture.

6.44 *Figure 2 represents the synoptic diagram of a second example of implementation of the LAN in accordance with the invention, for 128 stations (100 to 228). Each of the stations is fitted with a protocol processing circuit similar to circuit (19) previously described. Each station is fitted, firstly, with an input and an output respectively connected to an output and an input making up a port in the first asynchronous TDM switching network (247). In addition, each station is fitted, secondly, with an input and an output respectively connected to a port in the second asynchronous TDM switching network (248). These two networks provide data transmission safety by duplicating the virtual circuits established between the stations.

Networks (247) and (248) are strictly identical. Network (248) components performing the same functions as network (247) components are identified with the same item number to which is added the index Network (247) consists of a first stage including 16 matrices (229 to 244), each matrix being fitted with 16 ports; and a second stage made up of 8 matrices (245 to 252), each matrix being fitted with 16 ports and being identical to matrices (229) to (244). Ports are numbered 0 to 15. Ports No. 8 of the 16 first stage matrices are respectively connected to 16 ports of matrix (245) which is matrix No. 0 in the second stage. Ports No. 9 of the 16 first stage matrices are respectively connected to 16 ports of matrix (246) (not illustrated) which is matrix No. 1 in the second stage. Similarly, ports 10, 11, 12, 13, 14, and of the 16 first stage matrices are connected to the ports of second stage matrices No. 2 to No. 7.

The architecture of the connection networks (247 and 248) makes it possible to link 128 stations while authorising all stations to transmit data at the maximum admissible bit rate per matrix port, ie. 600 Megabits/sec. This means that, the fact that one or several stations are transmitting at this bit rate does not reduce in any way the possible bit rate of the other stations.

This is a major advantage over the known types of LAN.

Figure 3 represents the synoptic diagram of an example of implementation of a LAN in accordance with the invention, allowing to link 240 stations via two asynchronous TDM connection networks (558) and This implementation example differs from the previous one by the fact that each of the networks is simpler but does not allow the stations to simultaneously have the maximum admissible bit rate per matrix port.

The mean admissible bit rate per station is 30 Megabits/sec. approximately. If each of the stations transmits data at a bit rate lower than this value, usually, there is no loss of data. If at least one of the stations *5 transmits data at a bit rate higher than this value, there is a risk that at least one queue in one of the connection circuits will be saturated, and hence a risk of loss of data. Network (558') duplicates the virtual circuits established by network (558) in order to overcome possible loss of data. Nevertheless, this safety cannot be assured if some stations have a bit rate higher than 30 Megabit/sec.

Networks (558) and (558') are fitted with identical components. Network (558') components performing the same the same functions as network (558) components are identified with the same item number to which is added the index Network (558) consists of a first stage including 16 matrices (541 -o 556), each matrix being fitted with 16 ports; and a second stage made up of a single matrix 557, having 16 ports and being identical to each of matrices (541) to (556).

Stations (300) to (540) are each fitted firstly with one input and one output respectively connected to one output and one input making up a port of network (558). Secondly, each of these stations are fitted with an input and an output respectively connected to an output and an input making up a port of network The ports of matrices (541) to (557) are numbered 0 to Ports No. 0 to No. 14 of first stage matrices (541) to (556) make up 240 ports of network (558) which are connected to stations (300) to (540). Ports No. 15 of the 16 first stage matrices (541 to 556) are respectively connected to the 16 ports (No. 0 to No. 15) of matrix (557) making up the second stage.

'O0' It is therefore evident that this network architecture has 15 matrices less than network (247) described previously. The maximum number of virtual cir-

S*

cults which can be established is therefore greatly reduced. The total traffic in a first stage matrix (541 for instance) is concentrated on a single multiplex towards second stage matrix (557). This explains the possibility of locking and hence the limitation of the bit rate mentioned previously.

The invention is not limited to the implementation examples described above. It is within the capability of the man of the trade to implement other asynchronous TDM connection networks, by combining several stages of elementary matrices of a known type. In some applications, it is possible to eliminate the second connection networks (18, 248, 558') if occasional loss of data is not a serious fault.

"Figures 4 and 6 illustrate the data cell label allocation process, in an example of LAN implementation in accordance with the invention. The link of a first station transmitting data to a second station receiving this data consists of a transmit virtual circuit, going from the first station to the connection network, and of a receive virtual circuit going from the connection network to the second station. The transmit virtual circuit is identified by the label given to each data cell transmitted through this virtual circuit by the first station. The receive virtual circuit is identified by the label given to each data cell transmitted through this virtual circuit by the connection network. Typically, in a telecommunications switchboard including an asynchronous TDM connection network, the network extracts, then analyses the label of an incoming cell in the network, and deduces the routing of this cell, ie. towards which output the information is to be restored; and a new label to be allocated to this cell, at this output. A signalling network allows the destination terminals to establish the relationship between the cell labels and the identity of these terminals. In the LAN, in accordance with the invention, there is no signalling network, the connections are semipermanent and the stations are fitted with protocol processing circuits (19) knowing the meaning of the labels allocated to the cells.

Figure 4 illustrates a case, whereby a station (600) bearing the number o transmits data to a station (601) bearing the number via an asynchroo nous TDM connection network (603), such as one of those previously described in conjunction with figures 1 to 3. In this example, a third station (602) bears the number Those three stations are respectively connected to ports A, B and C of connection network (603). During transmission, each station uses a point-to-point virtual circuit to transmit to another station separately, and uses a broadcast virtual circuit to simultaneously transmit data to all other stations. During reception, each station uses a point-to-point virtual circuit to receive the data from each of the other stations, separately; and uses a broadcast virtual circuit to receive the data transmitted from one station towards all the other stations.

The protocol processing circuit (19) allocates a label to the cell to be transmitted and recongises the label of the cell to be received. These labels identify the semi-permanent circuits encountered by the cells during their routing.

Figure 4 illustrates data transmission from station No. i to station No.

j. The point-to-point transmit virtual circuit, starting from station No. i, then bears the destination station No. j. The point-to-point receive virtual circuit, arriving at destination station No. j bears the No. i j.

Figure 5 illustrates data transmission from station No. i to station No.

k. The point-to-point receive virtual circuit, arriving at destination station No. k then bears the No. i k.

Figure 6 illustrates data broadcasting from station No. i to all the other network stations. The broadcast transmit virtual circuit, starting from station No. i, bears the No. i. The broadcast receive virtual circuit, arriving at station No. j and station No. k also bears the No. i. Hence the stations can distinguish unambiguously between data cells individually address and data cells which are simultaneously addressed to all.

Figure 7 illustrates the safety protocol implemented to overcome transmission errors, generally due to loss of data by overflow of a queue in the o*66 connection network.

In accordance with the invention, in order to transmit data from station 31 (referred to as the calling station) to station 32 (referred to as the called station), the protocol comprises transmitting data from statio: S3 over a first link going from Si to 32; waiting for an acknowledgment transmitted by station S2 over a first and second link (c and d) going from 32 to S1; and retransmitting the same data over a second link going from 31 to 32, if calling station SI does not receive the acknowledgment sent by called station 32.

Each of the two stations linked receives in parallel over two links: c and d for SI; a and b for 32. Each of the two stations independently select one out of two virtual transmit links: a or b for Si; c or d for 32.

Here is a more detailed study of this protocol. The purpose of the safety protocol is to switch between a normal link and a standby link, and vice versa, without interrupting the transfer connection established between two stations. All transport protocol data units are used. The safety protocol consists of allocation to a transfer connection at least four unidirectional links made up of virtual circuits, two in each direction. One link and one link from the calling station (S1) to the called station S' I one link and one link from the called station (S2) to the calling station Usually, a single virtual circuit, for instance, is used to transmit from Si to 32. The link selected is the link having been used to send a connection request to S2. The link used in the direction S2 to SI is the link over which the connection confirmation is sent by S2. Reception is ensured simultaneously over links and by station S2. Reception is ensured simultaneously over links and by station 31.

Switching between a normal link and a standby link depends on the follow- 3-g criteria: During the connection phase: Calling station SI switches over a virtual circuit establishing the standby link when station S1 receives nei- S ther a connection confirmation nor a disconnection request, after having sent nl times a connection request over the virtual circuit of link which is active at the instant considered. Switching also occurs in the case where station S1 has received ml times the same connection confirmation, which proves that station S2 is not receiving the acknowledgment nor the data sent by Sl. Called station 32 switches over a standby virtual circuit, utilising link if station S2 has received nl times the same connection request over S one of the two links or which provides that the acknowledgment sent 0• by S2 over link did not reach station Sl; or in the case where station S2 receives neither acknowledgment nor data, after sending ml times the same connection confirmation over link or which is active at the instant considered.

During the data transfer phase: Calling station Sl switches over standby link if it has sent n2 times the data over link which is active at the instant considered, without receiving acknowleuegent. Called station S2 switches over standby link if it receives the same data m2 times, which proves that its acknowledgment did not reach calling station Sl. Station S1 also switches over standby link when it does not receive the idle test acknowledge, after having sent n2 times an idle test request. Station 32 also switches over standby link when it receives n2 times the same idle test request over the same link or this occurs if its acknowledge does not reach station Sl.

During the clearing phase: Station S1 switches over standby link (b) when it does not receive any disconnection confirmation, after having sent n3 times a disconnection request over link which is active at the instant considered. Station S2 switches over standby link if it receives m3 times the same disconnection request over link which indicates that its acknowledge did not reach S1.

During the re-initialisation phase: Station S1 switches over standby link if it does not receive a re-initialisation acknowledge, after having sent n2 times a re-initialisation command to station S2; or if it receives m2 times a re-initialisation acknowledge. Called station 32 switches over standby line if it receives m2 times the re-initialisation comand, or if it does not receive any acknowledge after having sent n2 times a reinitialisation acknowledge, which indicates that this acknowledge did not r* each station S1.

The values of nl, n2, n3, ml, m2 and m3 may be selected equal to 2 for instance.

S S

Claims

1. A local area network for data transrr in, comprising at least a first asynchronous time-division multiplexing switching matrix for setting up first semi-permanent virtual circuits; a plurality of stations, each station including means for information interchange in the form of cells, using a communications protocol adapted to asynchronous time-division multiplex switching matrices, each cell carrying a label designating one of the semi-permanent virtual circuits; these means including at least one input and one output coupled respectively to an input and an output of the first asynchronous time-division multiplexing switching matrix.

2. A network as claimed in claim 1, wherein in order to improve data transmission security, it includes at least a second asynchronous time-division multiplexing switching matrix setting up second semi-permanent virtual circuits, each station including a second input and a second output respectively connected to an output and an input of the second switching matrix; and in that the said means for information interchange include means for detecting transmission errors and for deciding to transmit cells of data either by the first or by the second switching matrix.

3. A network as caimed in claim 1, wherein the said means for information 20 interchange include means for assigning to each cell of data that is to be transmitted from a first station to a single second station, a fixed label g. designating the second station; and assigning to each cell of data to be transmitted from a first station to a plurality of second stations a fixed label only designating the first station. 25

4. A network as claimed in claim 1, wherein the said means for information t interchange include means for recognising, in a cell of data received, a label that designates a calling station in the case of a multi-station or broacast call; and recognising a label designating simultaneously a calling and a called station, in the case of a call to a single station.

5. A network as claimed in claim 2, wherein the said means for information interchange include means for transmitting over a first link selected from among a plurality of links that are set up in the direction of a called station; receiving in Sparallel over a plurality of links originating from the same called station; and 13 then sending over a different link set up in the direction of the called station if no information is received originating from the called stations (S2).

6. A local area network substantially as herein described with reference to Figures 1 7 of the accompanying drawings. DATED THIS TWENTY-NINTH DAY OF JANUARY 1993 ALCATEL N.V.