WO2001015383A1

WO2001015383A1 - A switch and a distributed network protection switching system incorporating said switch

Info

Publication number: WO2001015383A1
Application number: PCT/AU2000/000998
Authority: WO
Inventors: Nils Marchant
Original assignee: Nils Marchant
Priority date: 1999-08-25
Filing date: 2000-08-24
Publication date: 2001-03-01
Also published as: AUPQ244499A0

Abstract

Switch (10) is provided with three ports P1 - P3. Port P1 has an input (12) and output (14), port P2 has an input (16) and an output (18), and port P3 has an input (20) and an output (22). Switch (10) is arranged so that a first flow presented to the input of any one of ports P1 - P3 (eg flow 24i presented to input (12)) is delivered or internally routed to the output of an other of P1 - P3 (eg port (18) of P2) and a second flow presented to the input of that other of P1 - P3 (eg flow 26i at input (16) of port P2) is delivered to the output of said one of port P1 - P3 (eg output (14) of port P1). A detector is included in the switch (10) for detecting a predetermined characteristic of the flow presented to the inputs of each of ports P1 - P3. If the detector detects the existence of a predetermined characteristic of the flow presented to input (12) or input (16) (for example at input (12)) it automatically re-routes the input of the other port (eg port (16)) to the output of the remaining port (eg output (22) of port P3).

Description

Title

A SWITCH AND A DISTRIBUTED NETWORK PROTECTION SWITCHING SYSTEM INCORPORATING SAID SWITCH

Field of the Invention

This invention relates to a switch and a distributed network protection switching system incorporating said switch, particularly, although not exclusively, for the protection of flows through networks against the failure of channels or sites in the network.

Background of the Invention

Networks are employed to convey flows between end users. The end users may be persons, computers, other machines, biological organs or other entities which are able to transmit or receive flows. The events beyond the control of the network user cause networks to fail. Each failure within a network may disrupt one or more end users, in particular when the network is being relied upon for mission critical purposes. Network protection switching systems reduce the detrimental effects of failures upon end users. Some systems do so by switching flows away from failed parts of the network to operational parts, if any exist. The failure and the protection switching action both lead to a period of disruption to the end user. A quick protection switching response time will reduce the disruption experienced by the end users.

Throughout this specification the term "network" is used in a generic sense to describe a set of two or more sites and one or more links that connect those sites together in any topology. A network supports the end to end transfer of flows between sites across a concatenation of one or more links within that network. Each link is unidirectional, has one source end, and has one or multiple destination ends. Each link transfers a flow or flows from the source end to one or more destination ends. A flow transmitted from a site onto an operational link is transported to the destination site or sites. To form a bidirectional communication channel between two sites, links can be assembled as contra- flowing pairs.

It is important to note that the nature of the flow in one direction need not be the same as the flow in the opposite direction. Each site is able to transmit one or more flows onto one or more links, and to receive flows from one or more links. Each link at each site is either an incoming link or an outgoing link depending on the direction of flow carried by that link. The receipt of any flow by a site from an incoming link may become unreliable while that link has failed. The transmission of a flow from a site may become unreliable when the site has failed.

Throughout this specification the word "flow" is intended to denote anything that can flow between two sites and includes the transfer of a communication, signal, substance, entity, energy, pressure, absence of pressure, vacuum, waves, particles or a group of vehicles from one site to another site or sites. A single link can carry simultaneously one or more distinct and parallel flows. A single physical medium may carry distinct and opposing links or flows. Specific examples of flows include electrical, radio or optical signals, whether in digital or analogue form that convey information between sites within the network; hydraulic pressure conveyed along tubes, pipes or hoses; gas, liquid, slurry, solids, people, other beings or organic matter; pneumatically propelled canisters; a vacuum propagated along a tube, pipe or hose; or a transfer of energy.

One of the difficulties against which network engineers and protocol architects have been struggling is that of ensuring that information about a network failure is conveyed to a prescribed site in order to initiate protection switching. The problem is to carry such information through a faulty network. It is not always straight forward to determine which part or parts of a network can be relied upon to carry this information. The information about the failures should be carried on channels or links which have not failed. Sometimes even this cannot be achieved. There have been many proposals to protect networks against failures but these all have one or more of the following disadvantages:

• They require a separate network in addition to the protected network on which to transmit fault monitoring and control signals. ^β The devices which make and implement protection switching decisions are installed at sites which are geographically or topologically remote from the links or sites which they protect. In order to switch flows away from failed parts of the network, they rely upon the transmission through that network of signals which carry information about the failure. Systems which rely upon the transmission of such fault information signals have the following disadvantages:

• It is not always possible to find a path which has not failed for the fault signals. It is thus not always possible to guarantee that the fault information signals are able to reach their destinations.

• Network users may suffer unnecessarily long disruptions because switching decisions and operations are not performed near to the points of failure and because of the time required for: i) fault signals to be generated; ii) information about a fault to be conveyed across part of a network; iii) that information to be processed; iv) protection switching decisions to be made; v) control signals to be conveyed across part of a network in order to implement switching decisions; and vi) the switching decisions to be executed.

• The network may become flooded with a large number of fault messages which may overload the network and reduce the availability of the network to the end users.

• They rely upon central sites which themselves are vulnerable to failure.

• The time required to detect that a fault has occurred may be long. The decision that a link or site has failed is delayed until after a flow has become absent for at least a specified "time-out" period, resulting in a delay before protection switching may be initiated.

• The time required for protection switching to take effect is proportional to the number of sites in a network.

• They do not protect networks whose topology is more complex than a simple point to point link.

• They require inflexible and costly non-optimal network topologies to be installed.

• They require looped or ring topology networks.

• They do not protect against the failure of sites.

• They do not permit more than one redundant link to protect each protected link. • They do not support an arbitrary level of redundancy.

• They require redundant links to be installed where they are not required.

• In communications networks they do not operate independently of and transparently to open systems interconnection (OSI) layer services or protocols above the physical medium dependent (PMD) sub-layer of the physical layer (layer one). Separate protection switching devices must be constructed for each type of protected network.

• They degrade network performance while a fault condition exists

• They reduce network capacity while a fault condition exists.

• They require the unnecessary duplication or replication and simultaneous transmission of flows at times when no fault exists.

• They do not co-ordinate the protection switching at each end of a failed bi-directional link.

Summary of the Invention

It is an object of the present invention to provide a switch and a distributed network protection switching system that enables end to end connectivity across a network to be maintained in a highly reliable fashion by switching flows away from failed parts of the network.

According to a first aspect of this invention there is provided a switch comprising at least: first, second and third ports (P)-P₃) each having an input and an output; the switch arranged so that a first flow presented to the input of one of P,-P₃ is delivered to the output of an other of P,-P₃, and a second flow presented to the input of said other of P,-P₃ is delivered to the output of said one of said P,-P₃; detection means for detecting a predetermined characteristic of the flows presented at the input of each of P,-P₃; and, control means which, upon the detecting means detecting said predetermined characteristics in one of said first flow and said second flow, internally diverts the other of the first flow and second flow to be presented to the output of a remaining one ofP,-P₃. Preferably said switch further includes timer means for counting a time T for which the detecting means detects the existence of said predetermined characteristic of the flows and wherein said control means only diverts the other of the first and second flows to the output of said remaining one of P,-P₃ when the time T is equal to or exceeds a predetermined time T_wait.

Preferably said switch includes a dummy flow means for producing a dummy flow and, said control means delivers said dummy flow to the output of the port to which said one of said first and second flows would be delivered in the absence of that flow being detected as having said predetermined characteristic.

In one embodiment, said predetermined characteristic is the absence of said flow for said period T^,. In one embodiment, the period T_wait is 0.

In an alternate embodiment, said predetermined characteristic is a predetermined reduction in the rate of flow at said inputs. In an alternate embodiment when said flows relate to communication signals, said predetermined characteristic is a predetermined bit error rate, or signal to noise ratio.

In one embodiment, said dummy flow means is in the form of a generator for generating a flow of the same type as the flow presented to the inputs of said switch.

In an alternate embodiment, said dummy flow means includes means for sampling and subsequently replicating the flow presented to the inputs of the switch.

Preferably said detection means is further able to detect the absence of said predetermined characteristic after said control means has internally diverted said other of the first flow and second flow to the output of the remaining one of P,-P₃, whereup said control means rediverts said other of the first flow and second flow to be presented to the output of the other or one of P,-P₃, as ^{me case ma}y be. According to a further aspect of the present invention there is provided a method of using and operating a switch having first, second and third ports (P,- P₃), each port having an input and an output, said method comprising the steps of : coupling an incoming first flow of a first channel to the input of one of P,-P₃; internally routing said incoming first flow to the output of an other of P,-P₃; coupling an incoming second flow of a second channel to the input of said other of said P,-P₃; internally routing said incoming second flow to the output of said one of said P,-P₃; monitoring said inputs to detect a predetermined characteristic of the flow at said inputs; upon detecting said predetermined characteristic in one of said first flow and said second flow, internally re-routing the other of the first flow and the second flow to the output of a remaining one of the ports P,-P₃.

Preferably said method further includes a step of counting a time T for which said predetermined characteristic is detected and wherein said step internally of re-routing only occurs if said time T is equal to or exceeds a predetermined time T_wail. .

Preferably said method further includes a step of generating a dummy flow and internally routing said dummy flow to the output of the port to which said one of said first and second flow would be delivered in the absence of that flow being detected as having said predetermined characteristic.

According to a further aspect of the present invention there is provided a network having a distributed network protection switching system the network comprising at least: first and second sites (X„ X₂) for transmitting and receiving one or more flows; a first channel to allow bidirectional transfer of flows between said sites; at least one further channel to provide an alternate route for bidirectional transfer of flows between said sites, each of said first channel and said at least one further channel having a unidirectional incoming link and a unidirectional outgoing link; a first switch (S,) and a second switch (S₂), each of S„ and S₂ being in accordance with the first aspect of this invention; 51 coupled to the first site (X,) so that a flow out of X, is presented to the input of any one of P,-P₃ of S, and a flow into X, is delivered from the output of said one of P,-P₃ of S,;

5₂ coupled to X₂ so that a flow out of X₂ is presented to the input of any one of P,-P₃ of S₂ and a flow into X₂ is delivered from the output of said one of P,-P₃ of S₂; the outgoing link of channel C„ viewed from X,, connected between the output of an other of P,-P₃ of switch S, and the input of an other of P,-P₃ of switch S₂; the incoming link of channel C„ viewed from site X„ connected between the input of said other of P,-P₃ of switch S, and the output of said other of P,-P₃ of switch S₂; the outgoing link of channel C₂, viewed from site X„ connected between the output of the remaining one of P,-P₃ of switch S, and the input of the remaining one of P,-P₃ of switch S₂ and, the incoming link of channel C₂ viewed from site X, being connected between the input of the remaining one of ports P,-P₃ of switch S, and the output of the remaining one of P,-P₃ of switch S₂; whereby, in use, upon said detection means of one of S, and S₂ detecting a predetermined characteristic of a flow presented at its input from channel C„ said one of S, and S₂ internally diverts the flow directed to the output of the port containing that input to the output of the remaining port thereby causing the detection means of the other one of S, and S₂ to detect the absence of a flow at the input of the other one of switches S, and S₂ from channel C, so that the flow delivered to the output of the other port of switch S₂ is directed to the output of the remaining port of switch S₂ thereby switching the channel of communication between the first and second sites X,,X₂ from channel C, to channel C₂.

Brief Description of the Drawings

Embodiments of the present invention will now be described in detail with reference to the accompanying drawings in which:

Figure 1 is a block diagram of a switch in accordance with an embodiment of the present invention; Figures 2A, 2B & 2C depict three internal switching configurations of the switch shown in Figure 1 ; Figure 3 is a block diagram of the switch depicted in Figure 1 ;

Figures 4-14 depict the switching configurations employed in 11 different states of one state machine used to describe the operation of the switch shown in Figure 1 ; Figure 15 illustrates an embodiment of one network incorporating the switch and distributed network protection switching system in accordance with the present invention; Figure 16 illustrates a distributed network protection switching system for protecting a network against a site failure; Figure 17 illustrates a distributed network protection switching system for protecting a network against a point to point channel failure; Figure 18 illustrates a distributed network protection switching system for protecting a network against a link failure between switches in the network; Figure 19 illustrates on possible set of timing relationships in the operation of the switch; Figure 20 depicts a bypass device that can be incorporated in or associated with the switch; and Figure 21 is a block diagram of a further embodiment of the switch.

Detailed Description of the Preferred Embodiments

Referring to the accompanying drawings, switch 10 in accordance with an embodiment of this invention is provided with three ports P,-P₃. Each port has an input I and an output O. The input of port 1 IP,) and the output of port 1 (OP,) are designated by reference numbers 12 and 14 respectively. Similarly IP₂, OP₂, IP₃, OP₃, are designated as reference numerals 16, 18, 20 and 22 respectively. The switch 10 has three incoming links 24i, 26i and 28i for presenting flows to the inputs 12, 16, and 20 respectively and three outgoing links 24o, 26o, and 28o for carrying flows delivered from the outputs 14, 18 and 22 respectively. The pairs of links at each port can be considered to form a bidirectional channels 24, 26 and 28.

The switch 10 is arranged so that a first flow presented to the input of any one of ports P,- P₃ (for example a flow on link 24i to input 12) is delivered to the output of an other of P,- P₃ (for example output 18 of port P₂) and a second flow presented to the input I of that other of P,-P₃ (ie in this case the flow on link 26i to the input 16 of port P₂) is delivered to the output O of said one of ports P,-P₃ (ie output 14 of port P,). Thus, the channel 24 is connected to the channel 26 via internal paths in switch 10 connecting input 12 to output 18, and input 16 to output 14.

Detection means (not shown in Figure 1) is included in the switch 10 for detecting a predetermined characteristic of the flows presented to the inputs of each of the ports P|-P₃. Associated with the detection means is a control means (not shown in Figure 1) that, upon the detecting means detecting the predetermined characteristic in one of the flows presented to the inputs 12 or 16 internally diverts the flow from the other one of these inputs to the output 22 of the remaining port P₃.

Depending on the state of the switch 10, flows through the switch 10 can be passed through different combinations of pairs of ports P,-P₃. Figure 2A illustrates the switch 10 in a configuration where flows pass through ports P, and P₂ of the switch 10. In Figure 2B, the flows are directed through ports P, and P₃; and in Figure 2C the flows are directed through ports P₂ and P₃. The configuration taken by the switch 10 is dependent upon the characteristics detected at each of the inputs of all three ports as well as the current state of the switch 10.

Referring to Figure 3, the switch 10 is provided with flow detectors 30, 32 and 34 connected with the inputs 12, 16 and 20 respectively for detecting a predetermined characteristic of the flows presented to those inputs by the incoming links 24i,26i and 28i. As explained in greater detail hereinafter, the predetermined characteristic may be the simple presence or absence of a flow, or a predetermined bit error rate (for a digital transmission) or a signal to noise ratio (for an analogue transmission). However for the purpose of the following discussion, the detectors 30, 32 and 34 are structured to detect the present or absence of the flow, including a dummy flow, at each of the incoming links 24i,

26i and 28i. The presence of a flow is interpreted as "link healthy". The absence of a flow for greater than a time T_delect (T_{d ect} is described below) is interpreted as "link not healthy" or "link failed". Each of the detectors 30, 32 and 34 is coupled by a corresponding line to a controller 36. The flow on incoming link 24i is passed to internal switch 38 for directing the flow along one of alternate paths 40,42. The flow on incoming path 26i can be switched via internal switch 44 along alternate paths 46,48. The flow on incoming line 28i is provided to switch 50 for switching between alternate paths 52 and 54. The switch 10 includes internal output switch 56 leading to output 14; output switch 58 leading to output 18; and output switch 60 leading to output 22. Switch 10 further includes a dummy signal means 62 for generating a dummy flow or sampling and replicating a flow on one of the incoming links 24i, 26i, 28i. The output of the dummy generator is presented to switch 64. The switch 64 is linked to switch 60 by path 66, linked to switch 56 by path 68, and linked to switch 58 by path 70. Each of the switches 38, 44, 50, 56, 58, 60 and 64 has an associated control block 72. Each control block 72 receives control signals from the controller 36 via respective signal lines 74.

Also included is a power supply block 76 for providing a regulated power supply to the internal components of switch 10.

The purpose of the dummy flow produced by the dummy generator 62 is to enable one or more switches 10 elsewhere in the network to detect the presence of that dummy flow and to interpret the detection of that flow as "incoming link healthy". A dummy flow from one switch 10 is able to stimulate switching decisions at another switch 10. The requirements of the dummy flow and the dummy flow source which generated are not stringent. For example, in an optical embodiment of the switch 10, a continuous stream of light of suitable wavelength would suffice as a dummy flow. Where optical receiver power specifications or safety requirements are to be met (for example to avoid eye damage) a low duty cycle rectangular wave form would be employed to reduce the transmitted optical power. In an electrical embodiment, a simple square wave or rectangular wave would suffice. Where a periodic wave form is used the longest off period should be less than a predetermined time T_detect.

Optionally, instead of generating a dummy flow, in another embodiment of the invention the dummy flow means would replicate a selected incoming flow so that a copy of that incoming flow could be transmitted onto one or more selected outgoing links.

The controller 36 monitors the presence or absence of flows detected at each of the incoming links 24i, 26i and 28i via corresponding flow detectors 30, 32 and 34. In this configuration it can be seen that the flow on incoming link 24i can be delivered to the output 18 via switch 38, path 40 and switch 58; or output 22 via switch 38, path 42, and switch 60. The flow at input 16 carried on incoming line 26i can be delivered to the output 14 via switch 44, path 46 and switch 56; or output 22 via switch 44, path 48 and switch 60. A flow incoming on line 28i at input 20 can be delivered to the output 14 via switch 50, path 52 and switch 56; or to the output 18 via switch 50, path 54 and switch 58. The dummy flow provided by dummy generator 62 can be delivered to output 14 via switch 64, path 68 and switch 56; to output 18 via switch 64, path 70 and switch 58 or to output 22 via switch 64, path 66 and switch 60.

By supplying the appropriate combination of control signals to these internal switches, the controller 36 is able to cause a flow presented at the input of one port to be directed to the output to either one of the remaining ports. The controller 36 is also able to cause the dummy flow to be switched to any one of the three output ports, 14, 18 and 22. Additionally, the controller 36 is able to sense whether adequate power supply is available from the power supply box 76. In the absence of adequate power, or if no flow is detected on both normal incoming links 24i and 26i, the control block 36 activates two nominal normal paths to allow a flow presented at input 12 to flow to output 18 and a flow presented at input 16 to flow to output 14.

It is possible to implement the controller 36, dummy flow generator 62 and switches 38, 44, 50, 56, 58, 60 and 64 in a field programmable gate array (FPGA) such as the ALTERA EPF8282V which performs a switching algorithm based on a state machine described hereinafter. The input and output at each port can be realised by the use of a transceiver such as the Crystal Technologies CS61304A T1E1 line interface. This would provide signal bit rates at 1.544Mbps and 2.048Mbps. This component extracts bit clock and line data from an incoming flow, and generates a "loss of signal" (LOS) output signal when the absence of an incoming flow is detected. The data, clock and LOS outputs are supplied to the FPGA. A local timing source can be provided by a crystal such as the 32.768Mhz FOX F5C HS oscillator. The FPGA generates a dummy flow derived from the local clock and in accordance with the state machine (to be described below). Incoming and dummy data and clock signals are supplied from the FPGA to the line interfaces which transmit them to the outgoing links. Local and remote management interfaces enable the switch 10 to provide status information and to be controlled.

The operation of the switch 10 may be specified by a state machine. Figures 4-14 illustrate the 11 possible states in which the switch 10 would be found.

The 11 states are: OFF (O); NORMAL (N); IDLE (I); WAIT A (WA); WAIT B (WB); SKEW A (SA); SKEW B (SB); LOCK A (LA); LOCK B (LB); HUNT A (HA); and HUNT B (HB). Each is described below.

OFF (O) (Figure 4)

No power is supplied to the switch 10. If a by-pass device is installed it operates to bypass the switch 10. Each of inputs 12 and 16 is through enabled if possible to outputs 18 and 14 respectively. The broken lines represent that. Any flows which might be present at links 24i and 26i are thus transmitted to links 26o and 24o respectively. The switch 10 remains in the OFF state until power is supplied to it. It does not perform any active functions. Should power be applied the switch 10 switches to one of the other ten states whose descriptions follow below. The switching decision is not affected by the presence or the absence or nature of any flow at 28i, 24i or 26i. No flow is transmitted from 28o. No dummy flow is transmitted.

NORMAL (N) (Figure 5)

A flow is detected at inputs 12 and 16 for incoming links 24i and 26i. The switch 10 is through enabled. Each flow from incoming links 24i and 26i is transmitted onto the normal outgoing links 26o and 24o respectively via outputs 18 and 14. The switching decision is not affected by the presence or absence or nature of any flow on link 28i. No flow is transmitted from output 22 onto link 28o. No dummy flow is transmitted. IDLE (I) (Figure 6)

No flow is detected at either inputs 12 or 16 for links 24i or 26i. The switch 10 is through enabled. No flows are transmitted from outputs 14 or 18 onto links 24o or 26o. The switching decision is not affected by the presence or absence or nature of any flow presented to input 20 on link 28i. No flow is transmitted from output 22 onto link 28o. No dummy flow is transmitted.

WAIT A (WA) (Figure 7)

No flow is detected at input 16 for link 26i. A flow is detected at input 12 on link 24i. Link 24i is through enabled so that any flow at link 24i is transmitted from link 26o. The dummy flow 62 is active and transmitted from 24o. A timer TB is timing the delay period T_WAIT- (T_WAIT ^is described below.) The switching decision is not affected by the presence or absence or nature of any flow at link 28i. No flow is transmitted from link 28o.

WAIT B (WB) (Figure 8)

No flow is detected at input 12 for link 24i. A flow is detected at input 16 on link 26i. The flow on link 26i is through enabled to be delivered to output 14 and thus be transmitted on link 24o. The dummy flow 62 is active and delivered to output 18 to flow on link 26o. A timer TA is timing the delay period T_WAIT. (T_WAIT is described below.) The switch 10 is not affected by the presence or absence or nature of any flow on link 28i. No flow is transmitted from 28o.

SKEW A (SA) (Figure 9)

A flow on link 24i is detected at input 12. No flow is detected at input 16 for link 26i. The link 24i flow is delivered to output 22 to be transmitted on the alternative outgoing link 28o. The dummy flow 62 is active and delivered to output 14 to be transmitted on link 24o. No flow is transmitted on link 26o.

SKEW B (SB) (Figure 10) A flow on link 26i is detected at input 16. No flow is detected at input 12 for link 24i.

The link 26i flow is delivered to output 22 to be transmitted on alternative outgoing link 28o. The dummy flow 62 is active and transmitted from link 26o. No flow is transmitted

LOCK A (LA) (Figure 11)

A flow is detected at each of inputs 12 and 20 for incoming links 24i and 28i. No flow is detected at input 16 for link 26i. The incoming flow from link 24i is transmitted onto link 28o. The incoming flow from link 28i is transmitted onto link 24o. No flow is transmitted from link 26o. No dummy flow is transmitted.

LOCK B (LB) (Figure 12) A flow is detected at each of the inputs 16 and 20 of incoming links 26i and 28i. No flow is detected at input 12 for link 24i. The incoming flow from link 26i is transmitted onto link 28o. The incoming flow from link 28i is transmitted onto link 26o. No flow is transmitted from link 24o . No dummy flow is transmitted.

HUNT A (HA) (Figure 13)

No flow is detected at input 16 for link 26i. A flow is detected on link 24i. The flow received on link 24i is through enabled and transmitted from link 26o. The dummy flow 62 is active and transmitted from link 28o. No flow is detected at link 28i. No flow is transmitted from link 24o.

HUNT B (HB) (Figure 14)

No flow is detected at input 12 for link 24i. A flow is detected on link 26i. The flow received on link 26i is through enabled and transmitted from link 24o. The dummy flow 62 is active and transmitted from link 28o. No flow is detected at link 28i. No flow is transmitted from link 26o.

Transitions between the above described states are caused by the following events: El Power On The switch 10 commences to receive an adequate power supply. E2 A Found A flow is detected at link 24i.

E3 A Lost A flow at link 24i ceases to be detected.

E4 B Found A flow is detected at link 26i. E5 B Lost A flow at link 26i ceases to be detected.

E6 Alt Found A flow is detected at link 28i.

E7 Alt Lost A flow at link 28i ceases to be detected.

E9 TA Timeout The T_WArr timer T expires. El l TB Timeout The T_WArr timer T expires.

E12 Power Off The switch 10 ceases to receive an adequate power supply.

The switch 10 returns to the OFF state after this event, irrespective of any prior state.

The algorithm implemented at each switch 10, and the set of states, events and transitions between states which are employed here to describe the invention are not unique. The invention may be performed with alternative state machines and algorithms. Some state machines may employ a different number of state transitions than described here. For example the invention may also be performed using the state machine described here and with the following additional state transitions: IDLE to NORMAL; WAIT A to LOCK A; and WAIT B to LOCK B.

The switch 10 contains two delay timer functions TA and TB. TA is activated when the event "A Lost" occurs; TB is activated when the event "B Lost" occurs. Both TA and TB are configured to measure the delay time T_WArr. Expiry of the timer TA after delay T_WArr causes the switch 10 to switch from the WAIT B state to the SKEW B state. Expiry of the timer TB causes the switch 10 to switch from the WAIT A state to the SKEW A state. While the switch 10 is in the WAIT A state, if the flow at link 24i is also subsequently lost the switch 10 remains in the WAIT A state until either: the timer TB expires; or the flow at link 26i is newly detected.

Optionally, the delay time T_WArr may be set to zero in some switches 10. In such cases the transition through the WAIT states will be instantaneous.

Optionally, the TA delay time T_WArr may be set to a different value from the TB delay time T_WAIT. One switch 10 state transition table is shown in table one below. While the switch is in any of the states shown at the left of the table, the occurrence of any of the events shown at the top of the table causes the switch 10 to change to the new state shown within the table. The Power On event only occurs while in the OFF state and is the only event possible in the OFF state.

Events

Present

El E2 E3 E4 E5 E6 E7 E9 El l E12

State Power A A B B Alt Alt Lost TA TB Power i On Found Lost Found Lost Found Timeout Timeout Off

OFF I

I HA HB I I OFF

HA I N LA OFF

HB N I LB OFF

N WB WA N N OFF

WA WA WA N WA WA SA OFF

WB N WB WB WB WB SB OFF

SA I N LA OFF

SB N I LB OFF

LA I N SA OFF

LB N I SB OFF

Table 1: Network Repair Module State Transitions

Figure 21 illustrates a further embodiment of the switch 10' in which like reference numbers are used to denote like features. In short, the switch 10' includes various features in addition to those of the switch 10 depicted in Figures 1 and 3. Most notably, the switch 10' is provided with a fourth port P₄. Port P₄ has an input 140 and an output 142 to facilitate coupling with an ingoing link 144i and an outgoing link 144o. The input 140 and output 142 communicate via paths 146 and 148 respectively with the controller 36. The main purpose of port P₄ is to allow external communication with and control of the switch 10'. For example control can be exerted on the switch 10' via a signal input and link 144i instructing the controller 36 to force the switch 10' to change from one state to another.

The switch 10' also includes a status signal generator 150 which is shown in Figure 3 as a portion of a controller 36. However the status signal generator may also be incorporated in the detectors 30, 32 and 34 with a switch 10 or indeed be distributed between the controller 36 and the detectors 30, 32 and 34. The status signal generator 150 monitors the state of the switch and its operational status. The status signal generated by the generator 150 can be communicated to an external controller or control panel either via channel 148, output port 142 and link 144o or via the output of any one of the other ports P_ι.-P₃. When the status signal is communicated via port P₄ , the outgoing link 144o will carry only the status signal. However the controller 36 or indeed other parts of the switch 10' can be programmed, hardwired, or otherwise arranged so as to add or embed the status signal onto a flow which is outgoing from one of the outputs of ports P,-P₃. For example, if the switch 10' is operating in a SONET network the signal generator 150 can operate to change the state of one or more bits in the SONET line, section or path overhead fields to provide status information. On the other hand, if the switch 10' is operating so in a water reticulation or distribution system, signal generator 150 may act to add a dye to the water to provide information regarding the status of the switch and the flows it receives.

The status signal generator 150 can also act as or include a report flow generator to communicate with external entities the status of the switch 10'. The external entities may be other switches 10 or other devices. The report flow generator would generate a report flow when a predetermined characteristic of a flow is detected in any one or more of the inputs of ports P,-P₃. The nature of the report flow is such that it would be possible for an external entity to determine from the report flow the complete and exact status of the switch 10' including, for each of ports P,-P₃, whether the predetermined characteristic is present or absent. As with the status signal, the report flow would be delivered under the control of the controller 36 to any one or more of the outputs of ports P,-P₄.

The interaction between switches 10 and the operation of the state machine at each switch 10 is now described in detail by way of example. Two cases are examined: link failure; and site failure.

Figure 15 shows nine switches 1 O_A- 10, that are arranged to protect a bidirectional channel C, comprising a pair of links marked 78 in one direction and 80 in the opposite direction. The links 78 and 80 through switches 10_A-10_D form the normal channel which would be used when there are no faults. There are two redundant paths or channels: the first redundant channel C₂ comprises links 82,84 through switches 10_E-10_G; and the second redundant channel C₃ comprises links 86 and 88 through switches 10_H and 10,. The switches 10_A and 10_B would be at one site and the Switches 10_c and 10_D would be at a separate site.

The sequence of states in which each switch 10 finds itself during protection switching after a link failure is shown in Table 2 below. The description and the table illustrate only one of many possible sequences of state transitions which may occur as a result of the failure. The description and the Table do not illustrate all state changes which may occur.

Switches 10_A-10_D each commence in the NORMAL state (N) and carry flows along links 78,80. No flow is carried on the redundant channels C₂,C₃ and Switches 10_E-10, commence in the IDLE state (I).

Let a failure occur in channel C, between switches 10_B and 10_c. It does not matter whether the failure occurs only on a single link 78 or 80, or on both links 78 and 80, or if there is a time delay between a failure on each; the present invention will operate in all cases. In this example we shall assume that the failure occurs either on link 80 only, or on both links 78 and 80. The failure causes switch 10_B to detect the absence of a flow at its input 16, where the flow from switch 10_c via link 80 normally would be received. The absence of that flow causes switch 10_B to enter the WAIT A state (WA) and to present a dummy flow to its output 14 onto link 80 towards switch 10_A. The dummy flow is detected by switch 10_A and further downstream on link 80. If for some reason the link is repaired and the incoming flow on link 80 at input 16 of switch 10_B is again detected then switch 10_B returns to the NORMAL state. However, if the flow remains absent for at least a time T_WArr (which would cause the timer TB to expire and the TB Timeout event to occur) then switch 10_B enters the SKEW A state (SA), where it ceases to deliver any flow to output 18 and thus link 78 ceases to transmit any flow towards switch 10_c. Instead, the incoming flow at input 12 of switch 10_B is switched onto the alternative outgoing link 82 via its output 22 and travels towards switch 10_E.

Switch 10A ιo_B 10c 10_D 1°E ιo_F 10_o 10„ 10,

Initial State N N N N I I

After failure N WA N N I I

After T_WAΓΓ at 10_B N SA N N I I

Flow link 82 reaches 10_E N SA N N HA I

Flow link 82 reaches 10_F N SA N N HA HA

Flow link 82 reaches 10_o N SA N N HA HA HA

Loss detected at 10_c N SA WB N HA HA HA

After T_WArr at 10_c N SA SB N HA HA HA

Flow link 84 reaches 10_G N SA SB N HA HA N

Flow link 84 reaches 10_F N SA SB N HA N N

Flow link 84 reaches 10_E N SA SB N N N N

Flow link 84 reaches 10_B N LA SB N N N N 1

Flow link 82 reaches 10_c N LA LB N N N N 1

Final state N LA LB N N N N I

Table 2: Summary of Significant State Transitions in Response to Link Failure.

Note 1 : The sequence of transitions shown above is only one of many valid possible sequences. Note 2: No time scale is implied by Table 2.

Switch 10_E in the IDLE state, detects at its input 12 via link 82 the flow from output 22 of switch 10_B. That event causes switch 10_E to enter the HUNT A state (HA) in which it: i) delivers a dummy flow to its P₃ output 22; and ii) transmits the incoming flow it receives at its input 12 onto link 82 via its P₂ output 18. Switches 10_F and 10_G subsequently change from the IDLE state to the HUNT A state as a result of detecting the flow propagated to them along link 82 from switch 10_B.

After the disappearance of the flow on link 78 and thus at input 12 of P, of switch 10_c, switch 10_c enters the WAIT B state (WB) and then after a time delay T_WAIT enters the SKEW B state (SB) if it has not yet detected any flow at input 20 of P, via link 82. A dummy flow is transmitted by switch 10_c downstream along link 78 to switch 10_D and beyond. In the SKEW B state the incoming flow on link 80 at input 16 of P₂ of switch 10_c is switched to output 22 of P₃(10_c). After travelling along link 84 that flow is detected at input 16 P₂ (10_G). Switch 10_G then either: if no flow is detected at input 12, P, (10_G), changes from the IDLE state to the HUNT B state (HB); or, (as shown in Table 2) if it has already detected a flow at input 12, P, (10_G), changes from the HUNT A state to the NORMAL state. Eventually switch 10_G detects a flow at both input 12,P, and input 16,P₂ and enters the NORMAL state. The flow from output 14, P, (10_G) propagates further downstream along link 84 through switches 10_F and 10_E in that order, eventually causing these switches to switch into the NORMAL state.

The flow from switch 10_c via link 84 and switches 10_G, 10_F, 10_E is detected at input 20, P₃ of switch 10_B. That event causes switch 10_B to change from the SKEW A state to the LOCK A state (LA). The flow propagated along the link 82 from switch 10_G causes switch 10_c to change from the SKEW B state to the LOCK B state (LB) when it detects the flow at input 20,P₃ . A complete bidirectional channel C₂ via switches 10_A, 10_B, 10_E, 10_p, 10_G, 10_c and 10_D now operates around the failed normal channel C,. A subsequent failure of the redundant links 82 or 84 in channel C₂ causes a similar sequence of events and state changes at switches 10_A, 10_H, 10, and 10_D which result in diversion of flows along channel C₃ via switches 10_A, 10_H, 10, and 10_D. An arbitrary level of redundancy may be achieved by recursively nesting protection channels and switches.

Figure 16 shows nine switches 10_A-10, that are arranged to protect a network against the failure of a site 90. Site 90 contains a set of one or more nodes 92 and two switches 10_B, 10_c. Table 3 shows the sequence of states after the failure or removal of site 90. Switches 10_A-10_D commence in the NORMAL state (N), and switches 10_E- 10, commence in the IDLE state (I). Site 90 is constructed in such a manner that any failure at site 90 causes all the flows from site 90 to cease being transmitted. If site 90 fails no flows are transmitted on any of the four outgoing links 11, 13, 15 and 17 which emanate from site 90. Switch 10_A detects the loss of flow at its input 16 and switch 10_D detects the loss of flow at its input 12. The switch 10_A changes from the NORMAL state to the WAIT A state (WA) and switch 10_D changes from the NORMAL state to the WAIT B state (WB). Dummy flows are transmitted separately by switch 10_A onto link 19 and by switch 10_D onto link 21.

After a time delay T_WArr at switch 10_A, this switch enters the SKEW A (SA) state. After a time delay T_WArr at switch 10_D, this switch enters the SKEW B (SB) state. The flow on incoming link 23 to switch 10_A is transmitted onto link 25 coupled to output 22 of switch 10_A; and the flow on incoming link 27 to switch 10_D is transmitted onto link 29 coupled to output 22 of switch 10_D. Switches 10_E, 10_F, 10_G switch in that sequence from the IDLE state to the HUNT A state (HA) as the flow is propagated along the link 25 from switch 10_A. Similarly, as the flow from output 22 of switch 10_D is propagated along link 29, switches 10„ 10_H change from the IDLE state to the HUNT B state (HB) in that sequence. It is a combination of both the topology shown and the switching algorithms that ensures that switches 10_G and 10_H make contact and receive flows from each other.

When switch 10_G detects the flow at its input 20 from link 31, it changes from the HUNT A state to the LOCK A state (LA). Switch 10_G detects flows at input 12 from link 25 and at input 20 from link 31. When switch 10_H detects the flow on link 33 from switch 10_G at input 20 it changes from the HUNT B state to the LOCK B state (LB). The flow received by switch 10_G at its input 20 is switched onto its output 14 and propagated on link 35 via switches 10_F and 10_E to switch 10_A. That flow causes switches 10_F and 10_E each to change from the HUNT A state to the NORMAL state in that sequence. When the flow is detected at input 20 of switch 10_A, that switch changes from the SKEW A state to the LOCK A state. Similarly, the flow received by switch 10_H at its input 20 is switched onto its output 18 and propagated by link 17 to switch 10_D via switch 10,. That causes switch 10, to change from the HUNT B state to the NORMAL state. When the flow is detected at input 20 of switch 10_D, that switch changes from the SKEW B state to the LOCK B state.

The switches have now constructed a bidirectional channel around site 90 via switches 10_A, 10_E, 10_F, 10_G, 10_H, 10, and 10_D. The flow on incoming link 23 to input 12 of switch 10_A is carried via links 25, 33 and 17 to switch 10_D and is transmitted by switch 10_D onto link 21. The flow on incoming link 27 presented to input 16 of switch 10_D is carried via links 29, 31 and 35 to switch 10_A and is delivered by switch 10_A onto its output 14 to flow on outgoing link 19. The failed site 90 is thus by-passed.

Switch 10A ιo„ 10c ιo_D ιo_E 10_F 10_α ιo„ 10,

Initial State N N N N I I I I

After site failure WA N N WB I I I I

After T_WAJT delays SA N N SB I 1 I I

Flow reach 10_E and 10, SA N N SB HA 1 I HB

Flow reach 10_F and 10_H SA N N SB HA HA HB HB

Flow reaches 10_o SA N N SB HA HA HA HB HB

Flow from 10_H reaches 10_G SA N N SB HA HA LA HB HB

Flow from 10_G reaches 10_H SA N N SB HA HA LA LB HB

Flow from 10_H reaches 10_F SA N N SB HA N LA LB HB

Flow from 10_H reaches 10_E SA N N SB N N LA LB HB

Flow from 10_H reaches 10_A LA N N SB N N LA LB HB

Flow from 10_o reaches 10, LA N N SB N N LA LB N

Flow from 10_G reaches 10_D LA N N LB N N LA LB N

Final state LA N N LB N N LA LB N

Table 3: Summary of Significant State Transitions in Response to Site Failure.

Note 1 : The sequence of transitions shown above is only one of many valid possible sequences. Note 2: No time scale is implied by Table 3.

Site 90 is arranged in such a manner that a power failure at switch 10_B or 10_c or the node causes the power supply to all switches at site 90 to be switched OFF.

Note that the topology shown above is suitable to protect against the failure of a site, the failure of one or more links to the site, or combinations of failures of links and the site.

Switches 10 are able to protect simple point to point link topologies. The time delay T_WArr may be set optionally to zero at the innermost two switches which are immediately adjacent to a protected link. Two benefits are provided by setting the time delay T_WArr to zero:

• very high speed protection switching performance of the switches 10 may be achieved; and • the design and manufacture of switch 10 may be simplified.

Figure 17 shows one topology which would be used to protect a point to point channel C, comprising a pair of links 37 between a site 94 and a site 96. Two redundant channels C₂ and C₃ are shown. When channel C, represents a subscriber access connection to a public telecommunications network the site 94 would represent the subscriber's premises and the phantom line 39 represents the subscriber interface which would be crossed by the subscriber access links and redundant links. The equipment 98 would represent customer premise equipment (CPE) or other end user equipment. The site 96 would represent the public network. (Alternately when channel C, represents a trunk the sites 94 and 96 would represent separate exchange buildings or central offices.) Redundant channel C₂ is connected to the protected links at switches 10_B and 10_D. Redundant channel C₃ connected to the protected links at switches 10_A and 10_E. Further switches 10_c and 10_F are shown which provide additional internal protection to the network.

The innermost switches 10_B and 10_D which connect to the first redundant channel C₂ are the first switches to operate after a failure of any link 37 of channel C,. It is possible to dispense with the TA and TB timers in this way because the switches are not expected to receive any transmission "gaps" from the failed link channel. The state machines described above still apply, however switches 10_B and 10_D would pass through the WAIT

A and WAIT B states instantaneously after detecting any loss of flow on channel C,. In effect those switches would make a state transition directly from the NORMAL state to a LOCK or SKEW state. The setting of T_WArr to zero is optional: T_WAIT may be zero; and T_WAIT ^{mav De} non-zero. Both options are available in embodiment of the invention.

The following provides a detailed description of the delays which are encountered during protection switching. The failure of a link causes a transmission "gap" to be propagated downstream along the link from the point of failure. Gaps with a duration of less than a time T_WAIT do not activate protection switching in the switches 10. Only a first downstream switch 10 performs protection switching in response to that loss of flow, and only after a time greater than T_WAIT. Where T_WArr is zero protection switching occurs immediately after the loss of flow condition has been detected. Therefore, when a switch 10 in the NORMAL state detects the loss of flow on either inputs 12 or 16:

1. the switch 10 switches from the NORMAL state into either the WAIT A state or the WAIT B state and as soon as possible thereafter commences transmitting the dummy flow downstream on the failed link so that the duration of the transmission gap sensed downstream is minimised;

2. if the delay time T_WArr is greater than zero, the switch 10 waits one delay time T_WArr to allow any transmission gaps to pass by before switching to one of the LOCK A, LOCK B, SKEW A or SKEW B states; 3a. if any flow is detected again (ie. the flow has been restored for some reason or a dummy flow is detected) the switch 10 reverts immediately to the NORMAL state; 3b. if the flow is still absent at the end of the time T_WArr, the switch 10 switches to one of the LOCK A, LOCK B, SKEW A or SKEW B states.

The delay T_WArr is calculated from the time of the most recent loss of flow. When the flow is intermittent the T_WArr timer is restarted at each instant that the loss of flow is newly detected. When the detected transmission gaps are shorter than T_WArr the switch 10 returns to the NORMAL state at the end of each gap. Only when the flow has been lost for a continuous period greater than T_WArr the switch 10 switches to a SKEW or LOCK state. The delay T_WArr ensures that any switch 10 which may exist downstream from the point of failure does not disrupt unnecessarily the contra-flowing flow. Only the switch 10 adjacent to and immediately downstream from a point of failure switches into the LOCK A, LOCK B, SKEW A or SKEW B state.

Table 4: Timing Relationships

Note: No time scale is implied by Table 4.

Figure 18 shows two sites 100 and 102 connected by a network and switches 10_A and 10_B. Table 4 illustrates the delays which occur after the failure of link 41 and the relationship with state transitions at switches 10_A and 10_B. Time in the table passes from the top downwards. "X" represents an indeterminate transitional value which may occur during switching in real devices. Initially both switches 10_A and 10_B are in the NORMAL state and site 100 transmits to site 102 via the link 41. The network is arranged so that the transmission delay between switch 10_B and the receiving site 102 is negligible.

Link 41 fails at location 53. The loss of flow is detected at input 12 of switch 10_B. Switch 10_B takes a finite switching time T_swrrcH before it is able to activate and to transmit its dummy flow in state WAIT B towards site 102. If no incoming flow from switch 10_A via link 41 is detected at switch 10_B at the end of time T_WArr, switch 10_B switches to the SKEW B state. After the propagation delay incurred in traversing link 43 the flow at input 16 of switch 10_A is lost. Switch 10_A switches from the NORMAL state to the WAIT A state. Let switch 10_A detect at its input 20 the incoming flow from switch 10_B while still in the WAIT A state. Switch 10_A switches to the LOCK A state after the delay T_WA1T. The flow on link 49 presented to input 12 of switch 10_A is then transmitted to switch 10_B via the protection link 51. When detected at input 20 of switch 10_B after the propagation delay incurred in traversing link 51, switch 10_B switches from the SKEW B state to the LOCK B state, completing the repair action. The required flow from site 100 is again received at site 102.

Figure 19 illustrates the relationship between the time delays. At the top is shown a digital link flow whose value alternates between a high value 1 and a low value 0 as time passes to the right. The illustration is not to scale. The illustration may represent a digital optical flow which turns ON and OFF, or an electrical flow which alternates between a high voltage and a low voltage. The time 104 is the longest expected zero sequence in a normal flow. To successfully detect the loss of flow condition the switch 10 must fail to detect a flow for at least the time T_DETECr 106, which consists of the longest expected zero sequence 104 together with an error tolerance 108. The longest expected zero sequence 104 is the longest expected period for which received flow may be absent from a healthy flow including a dummy flow. Transmission gaps with a duration of less than a time T_DETECT 106 are not interpreted as "linked failed". The absence of a flow for greater than T_DETECT 106 is interpreted as "link failed". Therefore T_DETECT is dependent upon the transmission bit rate and the line coding scheme used. For example, in a digital optical embodiment in which logical "zeros" are represented by the absence of light, T_DETECT would be greater than the duration of the greatest expected number of consecutive optical "zeros" divided by the line rate measured in bits per unit time at the lowest transmission bit rate for which the module is designed.

The figure shows a healthy flow. If on the other hand the flow remained at zero for at least the time T_DETECT 106 then the flow would be deemed lost and one or more switch 10 state changes would be initiated. At the instant marked 110 at the end of T_DETECT 106 if the flow is still absent (not shown in the figure) a switch 10 switches from the NORMAL state to a WAIT state.

Any realisation of the switch 10 requires a finite time to perform switching operations and to activate and transmit the dummy flow if necessary. T_swrrcH 112 is the maximum time required to implement any changes of state at any switch 10 in a network. T_swrrcH 112 is equal to or greater than the time required to activate, switch and transmit the dummy flow and to perform all other switching functions. The longest possible transmission gap which is propagated downstream is less than or equal to the sum of T_DETEcr 106 and T_swrrcH 112. Downstream switches 10 remain in a WAIT state for at least a time T_WArr in order to allow any transmission gaps from upstream to flow by. In any realisation of a switch 10 with

T_WAIT ^not equal to zero, T_WArr 114 is therefore required to be equal to or greater than the sum of: i) the switching time T_swrrcH 112; ii) T_DETECT 106; and iii) the further delay 116, which provides a sufficiently large margin so that T_WArr at any switch 10 in the network is always greater than the longest possible transmission gap.

For example: if the longest expected zero sequence 104 on a lOKm, 155.52 Mbps optical point to point link is four optical zeros then T_DETECr 106 would be greater than the time 104 and would be given by:

TDETECT > 4 / ( 155.52 x 10° ) seconds = 25.7 nanoseconds

With a 74 nanosecond error tolerance 108 the decision to switch to the WAIT A or

WAIT B state from the NORMAL state may be made within 100 nanoseconds of the detection of loss of flow. With an additional maximum switching time T_SWITCH 112 also of

100 nanoseconds the maximum duration of any transmission gap would be 200 nanoseconds. Setting the TA and TB timer delays T_WArr 114 to 500 nanoseconds each provides a 300 nanosecond margin 116. Switching to a SKEW or LOCK state would occur within 500 nanoseconds of the detection of loss of flow at each switch. If the flow travelled at 2 _x 10⁸ m/sec then the round trip flow propagation delay would be

2 x 10^ / 2 _x 10^ seconds = 100 μseconds

Therefore in the worst case which occurs when one link fails near a transmitter protection switching would be completed within 101 μseconds. In the best case which occurs when each link fails near a receiver protection switching would occur within 500 nanoseconds. In the best case the protected flow in one direction would be switched to and transmitted along the redundant link almost immediately after the link failure. The receiving site downstream would receive the flow from the redundant link/channel almost immediately after detecting the loss of flow (assuming that the transmission delays of the protected link and the redundant link were comparable). By making the redundant link shorter than the protected link, the receiver would receive the diverted flow from the redundant link before the receiver lost the flow from the normal link. Setting the TA and TB timer delays T_WAIT to zero furnishes very high speed protection switching. For the example above, protection switching would be completed in less than 200 nanoseconds at each switch 10. Better performance would be achieved by reducing the generous margins given in this example. When the propagation delay in the protected link is identical to the propagation delay in the redundant link it is possible to switch paths without losing bit or frame synchronisation at the receiving sites.

After the failed links and sites have been repaired flows would be restored to the links which were originally used. Restoration of flows is achieved by forcing one or more switches 10 into the NORMAL state. A forced state transition at one switch 10 may automatically initiate state transitions at other remote switches 10. The forced switch 10 together with other switches 10 co-operatively achieve restoration. It is not necessary to force state transitions at all switches 10. For example the restoration of a flow to one of the pair of links 37 from one of the redundant links of channel C₂ in Figure 17 after the failure and subsequent repair of channel C, would be achieved by forcing only one local switch 10_D at for example the exchange end into the NORMAL state. The loss of flow from the redundant links and the detection again of a flow on the normal link at the remote switch 10_B together are sufficient to cause switch 10_B also to switch to the NORMAL state, so completing restoration.

Switches 10 and switching systems in accordance with this invention can thus be controlled from remote locations. This feature is of benefit to network operators who would be spared the expenses of additional network management networks, and travelling of technicians to switches 10 at remote sites.

Networks would be protected against the failure of switches 10 either: by installing switches together with separate active or passive by-pass devices; or by incorporating a passive by-pass device into switches. Figure 20 shows a separate by-pass device 118. Operation of the by-pass device would be controlled either: directly by the switch 10_A via a control line 120 as shown; or by some other device. Failure of the switch 10_A or its power supply would cause the by-pass device 118 to be switched so that the bidirectional through paths 122 and 124 are enabled and so that the failed switch 10_A is by-passed. When in the by-passed state any flow or flows received at input 126 are transferred to output 128 and any flows received at input 130 are transferred to output 132. When not in the by-passed state the flows at inputs 126 and 130 are transferred to the switch 10_A. The failure of any power supply which may be required by the by-pass device 118 would also cause the switch 10A to be by-passed.

Now that embodiments of this invention have been described in detail it will be apparent to those skilled in the relevant arts that numerous modifications and variations may be made without departing from the basic inventive concepts. For example, embodiments of the invention may be realised using a microprocessor to control the switches 10 instead of a field programmable gate array. Indeed, embodiments of the invention may be realised entirely in software that may be hosted in a computer, communications equipment or other equipment that may be installed at sites in a network. The software would implement the invention by altering routing tables, configuration tables or other mapping tables which specify associations between incoming and outgoing links and any of packet source addresses; packet destination addresses; telephone numbers; sockets; ports; incoming links; outgoing links; node addresses; connections; virtual channels; virtual paths; SDH tributary units (TU-ns); SONET virtual tributaries (VT-ns); user group multiplex labels; channel identifiers; routes; and other types of channels. Channels, paths, connections and routes would thus be remapped to links that connect to healthy parts of the network in response to the detection of a loss of flow condition. The loss of flow condition may be detected either by software or by signals from line interfaces and other devices within the host.

It is also important to note that embodiments of the invention may be structured in a manner so that say P, and P₃ would support links with different characteristics from the links at the other port P₂. For example one embodiment of the switch 10 for telecommunications networks would include high power interfaces at P, and P₃ that support long range flows, while the remaining port P₂ will include a low power interface for short range flows. This arrangement would enable the switch 10 to be installed in an exchange building or central office between existing transmission equipment and the transmission line. The low power port P₂ would connect to nearby equipment over shorter range links. The high powered ports P„ P₃ will connect to the longer range protected and protection lengths. Further, in embodiments of the switch 10 line interfaces can be used at the ports in which the bandwidth transmitted onto one or more outgoing links is not equal to the band width received from an incoming link. Switches 10 may thus be constructed for installation in networks that support asymmetric digital subscribed lines (ADSLs).

Also, while embodiments of the invention have been described incorporating a dummy flow generator such a generator is not essential for the protection of point to point links and other network topologies.

Where a synchronising timing structure is embedded within a flow, embodiments of the invention may include a time extraction mechanism that recovers synchronising timing from the flow. The timings though recovered may be employed to synchronise the transmission of one or more outgoing flows from a switch. The use of timing recovery and flow generation would enable the maximum number of switches 10 that may be concatenated along a set of links to be greater than embodiments that do not perform regeneration. A wide range of timing recovery components is available both as discrete components or integrated into receiver components.

Embodiments of the switch may include a device or system that monitors and optionally performs switching as a result of fault status or alarm signals furnished by and received from other sites in the network. For example, a switch 10 may be constructed that performs switching functions in response to the receipt of fault status, alarm or other signals.

Embodiments of the switch 10 can also be arranged or operated to detect the restoration of a normally used link after having failed, and automatically re-route a flow back onto that normal link. The requirement for manual intervention to divert a flow from a protection link to a protected link would thus be eliminated. The detection of link restoration may be achieved by transmitting a signal onto a failed link which would be received after restoration of that link.

One range of embodiments of the switch 10 may be constructed to protect the flow of the electrical energy through transmission and distribution networks that span or are contained within one or more continents or nations, a smaller region, a vehicle or machine, a circuit board, and a silicon or other microchip. Each switch 10 would generate a flow of electrical protection signals and transmit onto the electrical power transmission conductors, in both directions. Each physical link in the transmission network would facilitate both a flow of electrical energy in either or both directions; and, a flow of protection signals in either or both directions. The switches 10 would divert the protected power flow onto healthy links in response to loss of receipt of protection signals.

Finally, as mentioned above, the switches 10 can be constructed to protect networks other than electrical/electro-magnetic communication systems for example to protect networks of vacuum tubes that transport pneumatically propelled canisters; to protect blood vessels and other carriageways in human and other bodies to retain full connectivity between say the heart or other pump and other parts of the body. In this embodiment, the detection means would detect characteristics of blood flow such as blood pressure. Loss of blood pressure along one blood vessel would activate bypass switching to a standby redundant path blood vessel. A completely distributed network of blood carrying tubes or conduits connected via switches 10 could be constructed to create an extremely robust being.

All such modifications and variations together with others that would be obvious to a person of ordinary skill in the art are deemed to be within the scope of the present invention the nature of which is to be determined from the above description and the appended claims.

Claims

The claims defining the invention are as follows:

1. A switch comprising at least: first, second and third ports (P,-P₃) each having an input and an output; the switch arranged so that a first flow presented to the input of one of P,-P₃ is delivered to the output of an other of P,-P₃, and a second flow presented to the input of said other of P,-P₃ is delivered to the output of said one of said P,-P₃; detection means for detecting a predetermined characteristic of the flows presented at the input of each of P,-P₃; and, control means which, upon the detecting means detecting said predetermined characteristics in one of said first flow and said second flow, internally diverts the other of the first flow and second flow to be presented to the output of a remaining one ofP,-P₃.

2. The switch according to claim 1 further including timer means for counting a time T for which the detecting means detects the existence of said predetermined characteristic of the flows and wherein said control means only diverts the other of the first and second flows to the output of said remaining one of P,-P₃ when the time T is equal to or exceeds a predetermined time T_wait.

3. The switch according to claim 2 further including a dummy flow means for producing a dummy flow and, said control means delivers said dummy flow to the output of the port to which said one of said first and second flows would be delivered in the absence of that flow being detected as having said predetermined characteristic.

4. The switch according to claim 2 or 3 wherein , said predetermined characteristic is the absence of said flow for said period T_waιt..

5. The switch according to claims 2 - 4 wherein said period T_waιl is 0.

6. The switch according to any one of claims 2-5 wherein said period T_WAIT is different for each of P,-P₃.

7. The switch according to any one of claims 1 to 6 wherein said predetermined characteristic is a predetermined reduction in the rate of flow at said inputs.

8. The switch according to any one of claims 1 to 6 wherein when said flows relate to communication signals, said predetermined characteristic is a predetermined bit error rate, or signal to noise ratio.

9. The switch according to any one of claims 3 to 8 wherein said dummy flow means is in the form of a generator for generating a flow of the same type as the flow presented to the inputs of said switch.

10. The switch according to any one of claims 3 to 8 wherein said dummy flow means includes means for sampling and subsequently replicating the flow presented to the inputs of the switch.

11. The switch according to any one of claims 1 to 10 wherein said detection means is further able to detect the absence of said predetermined characteristic after said control means has internally diverted said other of the first flow and second flow to the output of the remaining one of P,-P₃, whereup said control means rediverts said other of the first flow and second flow to be presented to the output of the other or one of P,-P₃, as the case may be.

12. The switch according to any one of claims 1 to 1 1 further including a fourth port P₄ having an input and an output in communication with said control means for allowing external control of said control means including to control said control means to force a change in state of said switch.

13. The switch according to any one of claims 1 to 12 further including signal generating means for generating a status signal containing information relating to the status of said switch including any faults detected by said detecting means and wherein said status signal is delivered to an output of one of said ports P,-P₄.

14. The switch according to claim 13 wherein said control means is configured to act as said signal generating means.

15. The switch according to claims 13 or 14 wherein said control means is configured to add or embed said status signal to the flow delivered to the output of said other or said remaining one of said ports P,-P₃, or the output of port P₄.

16. A distributed network protection switching system for a network having at least first and second sites (X,, X₂) a first channel C, to allow bidirectional transfer of flows between said sites; and at least one further channel C₂ to provide an alternate route for bidirectional transfer of flows between said sites, each channel having a unidirectional incoming link and an unidirectional outgoing link; the system including at least:

a first switch (S,) and a second switch (S₂), each of S„ and S₂ being in accordance with any one of claims 1-15, SI coupled to the first site (X,) so that a flow out of X, is presented to the input of any one of P,-P₃ of S, and a flow into X, is delivered from the output of said one of P,-P₃ of S,; S₂ coupled to X₂ so that a flow out of X₂ is presented to the input of any one of P,-P₃ of S₂ and a flow into X₂ is delivered from the output of said one of P,-P₃ of S₂; an outgoing link of channel C,, viewed from X,, connected between the output of an other of P,-P₃ of switch S, and the input of an other of P,-P₃ of switch S₂; an incoming link of channel C,, viewed from site X„ connected between the input of said other of P,-P₃ of switch S, and the output of said other of P,-P₃ of switch S₂; an outgoing link of channel C₂, viewed from site X,, connected between the output of the remaining one of P,-P₃ of switch S, and the input of the remaining one of P,-P₃ of switch S₂ and, an incoming link of channel C₂ viewed from site X, being connected between the input of the remaining one of ports P,-P₃ of switch S, and the output of the remaining one of P,-P₃ of switch S₂; whereby, in use, upon said detection means of one of S, and S₂ detecting a predetermined characteristic of a flow presented at its input from channel C, internally diverts the flow directed to the output of the port containing that input to the output of the remaining port thereby causing the detection means of the other one of S, and S₂ to detect the absence of a flow at the input of the other one of switches S, and S₂ from channel C, so that the flow delivered to the output of the other port of switch S₂ is diverted to the output of the remaining port of switch S₂ thereby switching the channel of communication between the first and second sites X,,X₂ from channel C, to channel

C₂.

17. A network with distributed switching protection the network including at least: first and second sites (X,, X₂) for transmitting and receiving a flow; a first channel to allow bidirectional transfer of flows between said sites; at least one further channel to provide an alternate route for bidirectional transfer of flows between said sites, each of said first channel and said at least one further channel having a unidirectional incoming link and a unidirectional outgoing link; a first switch (S,) and a second switch (S₂), each of S,, and S₂ being in accordance with any one of claims 1-15;

51 coupled to the first site (X,) so that a flow out of X, is presented to the input of any one of P,-P₃ of S, and a flow into X, is delivered from the output of said one of P,-P₃ of S,;

5₂ coupled to X₂ so that a flow out of X₂ is presented to the input of any one of P,-P₃ of S₂ and a flow into X₂ is delivered from the output of said one of P,-P₃ of S₂; the outgoing link of channel C,, viewed from X,, connected between the output of an other of P,-P₃ of switch S, and the input of an other of P,-P₃ of switch S₂; the incoming link of channel C,, viewed from site X,, connected between the input of said other of P,-P₃ of switch S, and the output of said other of P,-P₃ of switch S₂; the outgoing link of channel C₂, viewed from site X,, connected between the output of the remaining one of P,-P₃ of switch S, and the input of the remaining one of P,-P₃ of switch S₂ and, the incoming link of channel C₂ viewed from site X, being connected between the input of the remaining one of ports P,-P₃ of switch S, and the output of the remaining one of P,-P₃ of switch S₂; whereby, in use, upon said detection means of one of S, and S₂ detecting a predetermined characteristic of a flow presented at its input from channel C,, said one of S, and S₂ internally diverts the flow directed to the output of the port containing that input to the output of the remaining port thereby causing the detection means of the other one of S, and S₂ to detect the absence of a flow at the input of the other one of switches S, and S₂ from channel C, so that the flow delivered to the output of the other port of switch S₂ is diverted to the output of the remaining port of switch S₂ thereby switching the channel of communication between the first and second sites

X„X₂ from channel C, to channel C₂.

18. A method of using and operating a switch having first, second and third ports (P,- P₃), each port having an input and an output, said method comprising the steps of : coupling an incoming first flow of a first channel to the input of one of P,-P₃; internally routing said incoming first flow to the output of an other of P,-P₃; coupling an incoming second flow of a second channel to the input of said other of said P,-P₃; internally routing said incoming second flow to the output of said one of said P,-P₃; monitoring said inputs to detect a predetermined characteristic of the flow at said inputs; upon detecting said predetermined characteristic in one of said first flow and said second flow, internally re-routing the other of the first flow and the second flow to the output of a remaining one of the ports P,-P₃.

19. The method according to claim 18 further including a step of counting a time T for which said predetermined characteristic is detected and wherein said step internally of re-routing only occurs if said time T is equal to or exceeds a predetermined time T_wait. .

20. A method according to claim 19 further including a step of generating a dummy flow and internally routing said dummy flow to the output of the port to which said one of said first and second flow would be delivered in the absence of that flow being detected as having said predetermined characteristic.

21. The method according to any one of claims 18-20 wherein said monitoring step includes monitoring said inputs to detect a predetermined reduction in the rate of flows at said inputs.

22. The method according to any one of claims 18-20 wherein, when said flows are in relation to communication signals, said monitoring step includes monitoring said inputs to detect a predetermined bit error rate, or signal to noise ratio.

23. The method according to any one of claims 18-20 wherein said monitoring step includes monitoring said inputs to detect an absence of the flow at said inputs.

24. A method according to claim 20 wherein said step of generating a dummy flow includes the steps of taking one or more samples of said flow presented at the inputs of said switch and constructing said dummy flow from said one or more samples as a replica of the flow presented to the inputs of said switch.