Redundancy matrix-based system

Abstract

The invention provides a system (100) comprising input or output channels connected to nominal and redundant equipment, and at least one redundancy matrix (10) for routing signals from a channel of equipment to redundant (20') or nominal (20) equipment, represented by a connection grid having rows corresponding to the system channels and columns corresponding to the equipment (20, 20') of the system, and connection points comprising: - the intersection points of a nominal path located on a nominal diagonal of the connection grid, and - the intersection points of a redundant path of the connection grid located between each row corresponding to a channel and a column corresponding to a redundant element of index , being surjective, the redundancy matrix being implemented as a circuit comprising switches arranged according to the connection grid. Figure for the abstract: Fig. 1

Description

Title of the invention: Redundancy matrix-based system Technical field

[0001] The invention relates generally to signal switching and in particular to a system based on a redundancy matrix.

[0002] In certain fields, it is known to use redundancy matrices to reroute signals from equipment to redundant equipment in the event of a failure on one equipment.

[0003] For example, in the space industry, reconfiguration devices using microwave switch networks, constituting redundancy matrices, are implemented to reconfigure the equipment of a satellite in the event of a satellite failure. Such devices make it possible to ensure the reliability of the payload during the satellite mission.

[0004] A redundancy matrix consists of a network of switches.

[0005] Conventionally, the “redundancy matrix” function is fulfilled by equipment called “redundancy rings”, consisting of multiple T-switches or R-switches interconnected according to a specific circuit.

[0006] With the constant increase in the number of channels and equipment installed on board VHTS (Very High Throughput Satellites), the complexity, volume, surface area and cost of redundancy matrices are increasing critically to the point of significantly encroaching on the surfaces, budget and mass allocated to the useful on-board electronics.

[0007] For example, for the Payloads of VHTS (Very High Throughput Satellites) type satellites, the number of channels and equipment increases significantly compared to conventional Payloads, which causes the complexity of the redundancy rings, their surface area, mass and costs to explode. The surface area available for useful on-board electronics is then significantly reduced.

[0008] The need to further integrate redundancy matrices has supported the development of more integrated switch technologies from MEMS (Micro Electro-Mechanical Systems), electronic switches, and micro-^PZ>T (miniaturized electromechanical switches).

[0009] These technologies have in common that they are based on an elementary switching element of the SPST (Single Pole Single Throw) type, which allows, depending on their respective integration capacity, the development of more complex switches of the nPmT (n Pole m Throw), R-switch, T-switch, C-switch type.

[0010] Known redundancy matrices generally rely on discrete electromechanical switches (coaxial T-switches or waveguide R-switches) mounted directly on the structure and interconnected by a radio frequency harness and controlled at the satellite level by individual LLC control harnesses (PLDIU cards).

[0011] However, it is necessary to improve the level of integration of such devices conventionally made from discrete electromechanical switches, individually controlled by the platform.

[0012] To improve the integration of the redundancy matrices, it may be envisaged to use specific switches such as, for example, phase change material switches (“Phase Change Materials” or “PCM”), electronic switches or miniaturized electromechanical switches.

[0013] However, these components have SP DT (Single Pole Double Throw) connectivity and not R / T-switches type, which makes them electrically incompatible in terms of number of accesses and switching states with conventional RF electromechanical switches.

[0014] There are very few redundancy matrices compatible with SPDT switches, and when they are compatible with SPDT switches, they cause a major problem of asymmetry between paths (length of lines and number of switches crossed) which results in prohibitive differential losses between the different paths.

[0015] Existing redundancy matrices are indeed subject to strict constraints of attenuation disparity between the paths. In particular, when R / T-switches are manufactured from SPDT switches to be able to refer to a conventional architecture, any attenuation disparity present in the reference architecture is amplified. This phenomenon of differential attenuation amplification is all the more problematic since conventional architectures are themselves poorly balanced in terms of the number of R / T-switches from one path to another.

[0016] It is also known to use redundancy matrices of the “dedicated nominal matrix” type, compatible with SPDT (or more generally SPnT) switches. In a “dedicated nominal matrix”, each channel has exclusive access to one of the nominal devices. To ensure the redundancy function, each channel also has access to all of the redundant devices.

[0017] However, such a matrix also exhibits a natural asymmetry between the different paths. The difference in attenuation between the best path and the worst path is greater the higher the number of channels.

[0018] There is thus a need for a device and a method for reconfiguring satellite equipment using redundancy matrices allowing an improved level of integration and which is compatible with SPnT switches, General definition of the invention

[0019] The invention improves the situation by proposing a system comprising a set of equipment. The equipment comprises one or more nominal equipment and one or more redundant equipment, a redundant equipment being associated with one or more nominal equipment, the system comprising n channels connected to said equipment, the channels connected to the equipment being input or output channels of the equipment, the system comprising at least one redundancy matrix of dimension nxm configured to route the signals of a channel of a given equipment to a redundant equipment associated with said given equipment or to another nominal equipment, in response to the detection of a rerouting triggering event relating to an equipment. Advantageously, each redundancy matrix is ​​represented by a connection grid having a matrix representation comprising rows and columns, each row corresponding to a channel C.0 < z < n of the system and each column corresponding to a piece of equipment in the system. The connection grid comprises a set of connection points corresponding to an intersection point between a channel and a piece of equipment, the connection points of the connection grid comprising at least: . - the set of intersection points of a nominal path comprising the intersection points of the connection grid located on a nominal diagonal of the connection grid, between each row corresponding to a channel of given index i and a column corresponding to a nominal device of the same index i, and - the set of intersection points of a redundant path comprising the intersection points of the connection grid located at the intersection between each row corresponding to a channel Cj of given index i and a column corresponding to a redundant element of index f (i) such that the function f is a surjective function defined from the set [0; n - 1]| on |]] for any index i.

[0020] The redundancy matrix is ​​implemented as a circuit comprising a plurality of switches arranged according to the connection grid.

[0021] In one embodiment, each redundancy matrix is ​​a sliding redundancy matrix and the connection points of the connection grid further comprise: - the set of intersection points of a sliding path comprising the intersection points of the connection grid located on an upper diagonal of the connection grid, between each row corresponding to a channel Q of given index i and a nominal device of index i + (e - 1) modulo n with e designating a sliding capacity of the matrix ete strictly greater than two, the sliding capacity being the number of points of the matrix located between a given point of the nominal diagonal on a given column of the connection grid and a point of the diagonal upper located on the same column of the connection grid.

[0022] Each redundancy matrix may be a sliding redundancy matrix, the sliding path further comprising the intersection points of the connection grid located on one or more intermediate diagonals of the connection grid, between each line corresponding to a channel Q of given index i and a nominal device of index Ei+k mocjujo n, with ( hk ) belonging to S and S c [ 0; n - 1 ]] X [ 1 ; e - 2 ] .

[0023] Advantageously, if 11 is strictly greater than zero and m is strictly greater than n + 1, the sliding capacity e satisfies the following condition:

[0024] 2 se s lJ

[0025] A perfect redundancy matrix has a sliding capacity e satisfying the condition:

[0026] e > m -n.

[0027] In particular, a sliding capacity redundancy matrix ~ n is perfect if and only if all points of the intermediate diagonals of the sliding path are connection points.

[0028] A perfect redundancy matrix has a perfection rate equal to one, the perfection rate representing the number of rerouting trigger events covered by the redundancy matrix.

[0029] In embodiments, the system may include an input redundancy matrix associated with the input channels of the equipment and an output redundancy matrix associated with the output channels of the equipment.

[0030] The switches can be of the SPnT type.

[0031] In one embodiment, the circuit implementing the redundancy matrix may be based on SPnT switch trees.

[0032] Advantageously, the circuit implementing the redundancy matrix may comprise a tree corresponding to each row and a tree corresponding to each column of the connection grid, the root of each SPnT switch tree of the circuit corresponding to a row of the connection grid associated with a channel or to a column of the connection grid associated with a device.

[0033] In one embodiment, a tree of the circuit implementing the redundancy matrix connected to a row corresponding to a channel has as many branches as there are connection points on the row in the connection grid and a tree of the circuit implementing the redundancy matrix connected to a column corresponding to a device has as many branches as there are connection points on the column in the connection grid.

[0034] In one embodiment, the trees corresponding to the grid lines of connection are connected to the trees corresponding to the columns of the connection grid.

[0035] The system may be configured to manage a number of failures on the equipment, the rerouting condition being the occurrence of a failure on a piece of equipment.

[0036] In one embodiment, the number of equipment failures supported by the system may be an increasing function of the sliding capacity of the matrix, when the intermediate diagonals are all full.

[0037] The embodiments of the invention provide a new architecture of redundancy matrices using SPnT switches. Due to its regular structure, such an architecture does not have the differential loss problems of conventional architectures. On the contrary, it makes individual balancing of the channels possible.

[0038] The redundancy matrix architecture according to the embodiments of the invention is based on a step-by-step sliding technique to exploit all of the redundant equipment installed on board without creating long paths, or paths containing a large number of switches.

[0039] Among other advantages, it has the ability to be simply cut into basic bricks and can be modulated in complexity to meet the exact reliability requirement at the payload level. Brief Description of Figures

[0040] Other characteristics, details and advantages of the invention will emerge from reading the description given with reference to the appended drawings given by way of example and which represent, respectively:

[0041] [Fig.l] [Fig.l] schematically represents a system comprising a set of nominal and redundant equipment, in which the redundancy matrix according to the embodiments of the invention can be used.

[0042] [Fig.2] [Fig.2] represents a system comprising a pair of redundancy matrices, according to an embodiment of the invention.

[0043] [Fig.3] [Fig.3] represents an example of a redundancy matrix based on a 1 for 4 redundancy scheme, with one redundant device for every 4 nominal devices.

[0044] [Fig.4] [Fig.4] shows another example of a redundancy matrix having a 1 to 4 redundancy scheme used in a system comprising 16 channels and 16 devices.

[0045] [Fig.5] [Fig.5] represents a redundancy matrix having similar connectivity to the matrix of [Fig.4], but with a grouping of nominal equipment and redundant equipment.

[0046] [Fig.6] [Fig.6] illustrates an example of a size-sliding redundancy matrix 16x20 composed of 4 4x5 matrices, suitable for managing a second specific failure, in the same group of nominal equipment.

[0047] [Fig.7] [Fig.7] represents a redundancy matrix of dimension 16x20 with sliding comprising a nominal diagonal and a neighboring full diagonal, according to an embodiment of the invention adapted to the management of any combination of two failures among the equipment.

[0048] [Fig.8] [Fig.8] represents an example of a redundancy matrix with sliding comprising a nominal diagonal and three neighboring full diagonals, according to an embodiment of the invention adapted to the management of any combination of 4 failures among the equipment.

[0049] [Fig.9] [Fig.9] illustrates a division of a 16x20 sliding redundancy matrix into four elementary rectangular bricks, according to an exemplary embodiment.

[0050] [Fig.l0][Fig. 10] represents a sliding redundancy matrix of dimension 16x20, configured to manage 3 specific failures.

[0051] [Fig. 1 l] [Fig. 11] represents a sliding redundancy matrix with a sparse diagonal.

[0052] [Fig.l2][Fig. 12] illustrates the three types of connection node paths in an example of a perfect 16x20 redundancy matrix.

[0053] [Fig.l3][Fig.13] illustrates the sliding capacity of a redundancy matrix.

[0054] [Fig.l4][Fig. 14] represents an example of a perfect sliding redundancy matrix, of dimension 16x20, having an irregular distribution of the channel and redundant equipment associations.

[0055] [Fig.l5][Fig. 15] illustrates an example of a sliding redundancy matrix (sparse matrix), of dimension 16x20, with an intermediate filling.

[0056] [Fig.l6][Fig. 16] illustrates an example of a sliding redundancy matrix (sparse matrix), of dimension 16x20, with minimum padding.

[0057] [Fig.l7][Fig. 17] shows a subset of the system comprising an input redundancy matrix 10-A and an output redundancy matrix 10-B.

[0058] [Fig.l8][Fig. 18] shows examples of k-branch SPDT switch trees that may be used in an implementation of the sliding redundancy matrix, according to embodiments of the invention.

[0059] [Fig. 19] [Fig. 19] shows an example implementation of the perfect 16x20 sliding redundancy matrix.

[0060] [Fig.20]Figure 20 shows an example implementation of the perfect 16x20 sliding redundancy matrix shown in Figure 20, in the form of a circuit comprising SP DT switch trees, Detailed description of the request

[0061] [Fig.l] schematically represents a system 100 comprising a set of equipment 2.

[0062] Equipment means a device or entity applying a certain operation to a signal arriving at its input to produce a signal at its output.

[0063] The equipment 2 may comprise one or more nominal equipment 20 and one or more redundant equipment 20'.

[0064] A redundant device 20' can be associated with one or more nominal devices 20.

[0065] The system 100 comprises at least one redundancy matrix 10 (hereinafter also called 'switching matrix') configured to route the signals from a given equipment 20 to another operational redundant equipment 20', associated with the given equipment 20, or from a nominal equipment 20 to another nominal equipment 20, in response to the detection of a rerouting trigger event relating to an equipment.

[0066] In embodiments of the invention, the rerouting triggering event (also referred to as a 'switching fabric reconfiguration event') may be a change in a set of excluded equipment. As used herein, an 'excluded equipment' refers to equipment excluded from the list of equipment to which the channels may be connected (the change may be, for example, an exclusion of a new equipment).

[0067] In an example of application of the invention to fault management, the set of excluded equipment includes equipment that has failed following the occurrence of one or more faults. In such an embodiment, the triggering event is then the detection of a fault on a piece of equipment.

[0068] For a given subset of excluded equipment, the role of a redundancy matrix is ​​to attempt to connect each channel to a non-excluded equipment. As used herein, a redundant equipment 20' designates another operational equipment, similar to the associating nominal equipment, capable of performing the same function as the operational equipment.

[0069] A piece of equipment 2 is said to be “operational” if it is not faulty (not subject to a breakdown) and activated (not switched off).

[0070] In an example of application of the invention, the system 100 may be, for example and without limitation, a satellite. In such an application of the invention, the equipment 2 then corresponds to the equipment of the payload of the satellite and is arranged on a radiofrequency chain. The redundancy matrices 10 are used to route radiofrequency (RF) signals from the channels of a nominal equipment 20 to a redundant equipment 20' of the system 100 or of an equipment nominal 20 to another nominal 20 equipment.

[0071] In such an example of application of the invention to the space industry, the impossibility of ensuring the maintenance of the satellites in flight (system 100) requires making the Payloads more reliable by adding redundant equipment 20' all along the RF chain, which remains switched off and therefore does not age throughout the duration of the mission, until the occurrence of a failure of a given equipment 20, where then one of the redundant equipment 20' is put into service to replace a faulty nominal equipment 20. The redundancy matrices 10 are then used to remove the faulty equipment 20 from the Payload routing and insert the redundant equipment 20' in its place.

[0072] For example, the redundancy matrices can be used to ensure the redundancy of equipment 20 of the frequency converter type.

[0073] Such redundancy matrices can also be used in optical applications.

[0074] The invention applies more generally to any system 100 in which redundancy matrices are used to reroute signals of channels 3 from a nominal equipment 20 to a redundant equipment 20' or from a nominal equipment 20 to another nominal equipment 20.

[0075] For example, the redundancy matrix 10 according to the embodiments of the invention can be used in a test system 100 configured to test equipment 20 of the DUT (Device Under Test) type, the channels 3 then corresponding to the different measurement channels (corresponding to the channels).

[0076] The system 100 can comprise as many channels 3 as there are useful signals.

[0077] Channels 3 include input channels connected to system equipment or the output channels connected to the system equipment, the redundancy matrix 10 then being configured to reroute the signals of the input channels and / or the output channels of the nominal equipment 20.

[0078] An equipment access (also called 'equipment port') corresponds to the interface between the redundancy matrix 10 and the input or output of an equipment 2.

[0079] A channel designates an arrival path of a signal to be processed by equipment, or an departure path of a signal having been processed by equipment.

[0080] A channel access (or 'channel port') designates an interface between the redundancy matrix and a channel.

[0081] [Fig.2] represents a system 100 comprising a pair of redundancy matrices, according to an embodiment of the invention.

[0082] The pair of redundancy matrix comprises a first redundancy matrix 10A (input matrix) and a second symmetrical matrix 10B arranged at the output (output matrix) framing the nominal equipment 20. In such a mode of rea lization, the first redundancy matrix 10A arranged upstream of the equipment 20 is configured to distribute the signals of the input channels 3A to redundant equipment 20' while the second redundancy matrix 10B performs a function symmetrical to that of the first redundancy matrix 10A by recovering the signals from the nominal equipment 20' and rerouting them to the output channels 3B for transparent rerouting management and continuity of service.

[0083] In the case of an input matrix 10A located upstream of the equipment, the inputs of the matrix (rows) correspond to the channel port, while the outputs of the matrix (column) correspond to the equipment ports.

[0084] In the case of an output matrix 10B located downstream of the equipment, the inputs of the matrix (rows) correspond to the equipment ports, while the outputs of the matrix (columns) correspond to the channel ports.

[0085] To facilitate understanding of the invention, the remainder of the description will be made mainly with reference to an input matrix whose rows correspond to the channel ports and whose columns to the equipment ports. However, those skilled in the art will easily understand that the embodiments described below apply similarly to the output matrices.

[0086] Each redundancy matrix 10 is associated with a connection grid representing the connectivity of the matrix, the redundancy matrix being implemented in the form of a circuit from the connection grid. The connection grid represents the connection points (or connection nodes) between a given channel Q and a nominal equipment E. (20), and the connection points between a given channel C{ and a redundant equipment R^ (20').

[0087] Advantageously, a circuit implementing the redundancy matrix comprises a pair of SPDT switches for each connection point of the connection grid.

[0088] In the remainder of the description, reference 10 will be used to designate the redundancy matrix, its connection grid or the circuit implementing the connection grid.

[0089] The system 100 may be based on a chosen redundancy scheme. A “P for K” redundancy scheme defines the number P of redundant equipment 20' for K nominal equipment 20.

[0090] A piece of equipment may consist of one or more components.

[0091] In an application of the invention to satellite systems, a piece of equipment may, for example, group together several amplifiers or frequency converters.

[0092] Alternatively, a piece of equipment may be a single component, such as an individual amplifier or frequency converter.

[0093] A device 20 can thus have one or more input channels and one or several exit routes.

[0094] The connection grid of the redundancy matrix 10 has a matrix representation comprising rows L, and columns T

[0095] The index i corresponds to the index of the i-th row and the index j corresponds to the index of the J-th row.

[0096] Each row of the redundancy matrix 10 represents an input of the redundancy matrix 10, and each column represents an output of the redundancy matrix 10.

[0097] Thus, in the embodiments of the invention where the redundancy matrix 10 is used to connect a channel 3 with a device 2, each row of the connection grid corresponds to a channel of the system 3 and each column of the connection grid corresponds to a device 2 of the system 100.

[0098] Each line L, thus corresponds to a channel C, while each column T. corresponds to a piece of equipment 20 which can be a nominal piece of equipment E. or a redundant piece of equipment E. associated with one or more nominal pieces of equipment.

[0099] In the remainder of the description, parameter 11 will be used to designate the total number of channel accesses 3 (number of lines) of a redundancy matrix. The channel accesses of a redundancy matrix 10 will be noted C- 0 ■ < n.

[0100] The parameter m will be used to designate the total number of equipment accesses 2 (number of columns) of the redundancy matrix. Among these equipment accesses, n equipments (corresponding to 11 columns of the connection grid) are called "nominal" and will be noted E; 0 < ? < n. The other equipments 20' are called "redundant" and will be noted Rjj < m-»-

[0101] The connection grid associated with the redundancy matrix 10 therefore has a dimension nxm (il pgnes C( m coionnes).

[0102] Each intersection between a line L-, and a column T. represents a possible connection between the channel corresponding to the line L, and the equipment 20 (E. or corresponding to the column TE.

[0103] Only some of the intersections of the connection grid correspond to connection points between a channel 3 and a device 2.

[0104] The connection grid is configured so that only one connection is possible simultaneously, i.e. a row C can only connect to a single column T and conversely a column T can only be connected to a single row U.

[0105] A group of nominal equipment 20 designates all of the nominal equipment 20 which are associated with the same redundant equipment 20'.

[0106] According to the embodiments of the invention, the connection grid of the redundancy matrix 10 is configured so that a channel Cj can have possible connection points with more than one nominal equipment 20 to manage the occurrence of a number of failures (more generally a number of occurrences of an event triggering a rerouting) successive on a group of equipment associated with the same redundant equipment. Furthermore, the connection grid associated with the redundancy matrix 10 only comprises one connection between a given channel Cj and a redundant equipment (20'). In the event of a failure on an equipment E., the channel Cj associated with the faulty equipment E. will connect to a predetermined redundant equipment Rf^) from the possible connection points of the channel Cj with the equipment Ek. ​​A channel Cj will connect in priority to a predetermined redundant equipment Rf^) if it is available.

[0107] To facilitate understanding of the invention, simplified connection grid configurations associated with the redundancy matrix 10 will be described in relation to FIGS. 3 to 7.

[0108] [Fig.3] represents an example of a redundancy matrix 10 based on a 1 for 4 redundancy scheme, i.e. with redundant equipment 20' for 4 nominal equipment 20.

[0109] Figure 3 shows an example of a redundancy matrix 10 of size 4*5 on the left simplified, represented in the form of a connection grid, and on the right, in an SPDT implementation. The redundancy matrix of Figure 3 is associated with 4 nominal devices Ej (20) and 1 redundant device (20') associated with the 4 devices. The redundancy matrix 10 comprises 4 channel accesses Cj, each channel Cj being connected to a device E, of the same index i on the nominal diagonal D of the matrix.

[0110] With such a redundancy matrix 10 (represented by the connection grid of figure 3), each channel Cj can be connected to its nominal equipment Ej (20) or to the associated redundant equipment Rq (20'), if the equipment E, fails.

[0111] Figure 4 shows another example of a redundancy matrix 10 having a 1 for 4 redundancy scheme used in a system 100 comprising 16 channels 3 and 16 devices 20. The system 100 uses 4 redundant devices 20' and 4 redundancy matrices 10 juxtaposed to satisfy the 1 for 4 redundancy scheme, with the simplest possible implementation. In Figure 3, a redundant device Rj is placed after the 4 nominal devices Ej (20) associated with it. The redundancy matrix is ​​of size 16*20 (20 columns corresponding to the 16 nominal devices E. and the 4 redundant devices R ) and juxtaposes 4 A 4x5 matrices.

[0112] Figure 5 represents a redundancy matrix 10 corresponding to that of Figure 3 with modifications to have equivalent connectivity, but with a grouping of the nominal equipment E. (20) in the first n columns and the redundant equipment Rj (20') in the last m 'n columns. The diagonal of this matrix corresponds to the nominal diagonal. The nominal diagonal comprises the set of intersection points of the redundancy matrix 10 between each row corresponding to a channel C, of ​​given index i and a column corresponding to a nominal equipment £ / having the same index i.

[0113] It should be noted that the permutation of rows or columns does not alter the connectivity described by the connection grid.

[0114] A connection between a channel C, and a piece of equipment E. will be noted below (Q ** E. ). Similarly, a connection between a channel Cf and a redundant piece of equipment Rj will be noted below (QR .y

[0115] Thus, in the matrix configuration of Figure 5, the first n columns of the redundancy matrix 10 correspond to the nominal equipment E. (20), while the following m "n columns successively comprise the redundant equipment R, (20'), with n = 16 and m - n - 4. The nominal equipment E. (20) J k of the same group of nominal equipment (i.e. associated with the same redundant equipment 20') are themselves arranged one after the other and the order of the redundant equipment corresponds to the order of the groups of nominal equipment associated in the first n columns.

[0116] Such a redundancy matrix 10 makes it possible to route any channel C / (3) to the associated redundant equipment Rk (20) at the first failure, the index k of the corresponding redundant equipment being given by k — [ 1' / K ] (in the example of figure 5, K = 4), in a “P for K” redundancy scheme.

[0117] It should be noted that such switching at the first failure is possible without requiring disconnection of other equipment 20, which satisfies the constraints linked to the redundancy rings according to which the channels other than that corresponding to the path that failed at the first failure must not be disconnected.

[0118] Assuming that a second failure then occurs on a device E. in a group of devices G other than the group Gk associated with the redundant device Rk, the channel C. can be connected to the redundant device Rq with q=\j!K\ *k and thus reroute the signals of the channel C. to the redundant device Ri.

[0119] However, with such a redundant matrix configuration, the C channel cannot not be rerouted if a device E. fails belonging to the same device group Gk as that associated with because such a device is associated with the same redundant device.

[0120] The redundancy matrix 10 can be configured to allow the rerouting of two channels associated with two devices of the same group of devices to a redundant device for one of the channels and to a neighboring device determined according to the sliding principle (the neighboring device will itself be routed to its neighboring device and so on until the redundant device of the next group is reachable). Such a redundancy matrix, hereinafter called a “sliding redundancy matrix” is configured to “slide” all the channels C, associated with another device E of the same group Gk with an index higher than that of the first device Ej to fail, to the neighboring device E}., +} and to fetch the redundant device of the neighboring device group Gk+ ] in the event of failure of the device E., of the same group, by adding connection points. Thus, if a first device E.fails in a group Gk associated with a redundant device Rk, the channels Cas associated with the other devices E., of the same group, of index j ' > j, will connect to the neighboring device E .,+ A of the following index j ' + 1 in the event of failure of the device E., so as to be rerouted to the redundant device of the neighboring device group Gk +1 thanks to the addition of auxiliary connection points between Cet E.,+ j. .

[0121] [Fig.6] illustrates a 16x20 sliding redundancy matrix 10 composed of 4 4x5 matrices. It shows the addition of connection points in the redundancy matrix 10 to ensure continuity of service in the event of a second specific failure.

[0122] In this example, it is considered that a first failure occurs in an initial state such that i = 4 is the index of the first failure (failure on the equipment E4 associated with the channel C4) and then subsequently that a second failure of index J = 6 occurs (failure on the equipment E6 associated with the channel C6). The two failed equipments E4 and E(, belong to the same group G} of 4 equipments associated with the same redundant equipment R^. Thus, all the equipments of the group Gq (i.e. £4, E^ E6 and A?) are initially configured to be connected to the redundant equipment At in the event of failure of one of them. However, this embodiment with sliding redundancy matrix 10 makes it possible to modify the active connection between a channel and an equipment in the case where a failure occurs successively on more than one equipment of the same group. Thus, during the first failure on £4, the active connection (C4 ** E4) becomes the point (C4 Aj. Then during the second failure, by adding the connection points (C6 E^ and (C7 E^ by shifting the equipment E^ and Ej with an index greater than i — 6 in the group G] to the following equipment £7 and E#, and by activating them, the channel connects to E7 (instead of E6\ the channel G7 to E^ (instead of £7) and finally the channel C8 to R2. The added connection points thus make it possible to carry out a step-by-step shift of the channels towards the next redundant equipment R2.

[0123] The redundancy matrix 10 can thus be improved to allow the processing of two failures occurring successively on two pieces of equipment within the same group of equipment by generalizing the sliding process by recurrence.

[0124] Thus, in one embodiment, the redundancy matrix 10 has a diagonal close to the nominal diagonal D, full (that is to say that all the points of the diagonal close to the nominal neighbor correspond to additional connection points which can be activated between a channel C, and a piece of equipment Ei + j close to a piece of equipment connected E, to the same channel on the nominal diagonal).

[0125] The redundancy matrix then includes connections (Ci E^ on the nominal diagonal D and auxiliary connections (C, E(i + mod N) on the neighboring diagonal for all values ​​of i between 0 and N - 1.

[0126] Figure 7 represents a redundancy matrix with sliding comprising a nominal diagonal D and a full neighboring diagonal, according to an embodiment of the invention. The redundancy matrix 10 is of dimension 16x20 and is composed of four interconnected sub-matrices of dimension 4x5.

[0127] It can be observed that in the neighboring diagonal Di, a connection point is added between the last channel of largest index C15 and the first equipment of smallest index E^ (Cl5 Eo) in order to allow a sliding of the last group G4 of four equipments (£\2, £13' ^f4- £15), associated with the last redundant equipment of largest index Rj, towards the first group Gj, associated with the first redundant equipment Rq, thus closing the sliding loop.

[0128] In another embodiment, the redundancy matrix 10 can be further improved to allow the processing of more than two failures occurring successively on three pieces of equipment within the same group of equipment by again applying a generalization by recurrence of the sliding method.

[0129] More precisely, in one embodiment, the redundancy matrix 10 can have i neighboring diagonals Dt of the nominal diagonal, full, with E = e — 1 (i designates the number of neighboring diagonals, without the nominal diagonal).

[0130] For the / - th neighboring diagonal all the points of the neighboring diagonal Dj correspond to an additional connection between a channel C, and a neighboring device Ei + ! of a connected device Ej to the same channel G, on the nominal diagonal D.

[0131] The redundancy matrix then includes connections (G) ** E^ on the diagonal nominal and auxiliary connections (QE,. + N) on each neighboring diagonal D} for all values ​​of i between 0 and n - 1.

[0132] Figure 8 represents a redundancy matrix with sliding comprising a nominal diagonal D and three neighboring full diagonals D2 and D3, according to an embodiment of the invention. The redundancy matrix 10 is of dimension 16x20 and is composed of four interconnected sub-matrices of dimension 4x5. Four redundant devices are used Rq, R^, R2, and R3. The redundancy matrix of [Fig.8] makes it possible to reroute the channels to redundant devices even when 4 failures occur successively on the devices of the same group.

[0133] The redundancy matrix 10 thus produced is perfect, which means that for any combination of failures of K equipment (K less than or equal to the number of redundant equipment), there is a configuration of the redundancy matrix guaranteeing continuity of service of the system by rerouting all the channels associated with the faulty equipment to a functional equipment (redundant equipment).

[0134] A perfect NXM matrix by definition supports any combination of M - N failures among the M devices (without distinction between nominal and redundant devices). For example, the 16x20 matrix of [Fig.8] supports any combination of 4 failures among the 20 devices and a perfect 27x30 matrix supports any combination of 3 failures.

[0135] Unlike existing solutions, the use of such a redundancy matrix by the system 100 makes it possible to minimize the asymmetry of the number of switches crossed between paths, regardless of the number of failures. It also makes it possible to distribute the accesses of redundant equipment Rk in the redundancy matrix in such a way that the matrix has a division by elementary bricks, allowing a modular mechanical architecture.

[0136] Indeed, the redundancy matrix according to the embodiments of the invention is used to produce a hardware implementation (circuit) in the form of an arrangement of miniaturized switches mounted on printed circuit boards. The surface area of ​​a printed circuit may be insufficient to accommodate the entire circuit (and in particular all the switches). The circuit may then be produced using several separate boards, themselves potentially mounted in separate mechanical structures using the division into elementary bricks.

[0137] An elementary brick corresponds to a subset of the complete circuit implementing the redundancy matrix.

[0138] [Fig.9] illustrates such a division of a 16x20 sliding redundancy matrix into four elementary rectangular bricks Bb B2, B3 and B4, according to an example of realization lization. A rectangular brick is formed around the contiguous connections associated with each equipment group. A group of equipment can be associated with two elementary bricks.

[0139] Advantageously, the elementary bricks require a limited number of interconnections (limited number of RF cables) and use an interconnection topology in the form of a simple loop (B 1 —> B2 —> B3 —> B4 —> Bl), which allows the use of short cables.

[0140] More precisely, the number of interconnections between a brick Bk and the next one Bk+1 is equal to the number of sliding diagonals (e - 1).

[0141] The division into elementary bricks also allows optimization of the non-recurring development cost. Rather than developing a new product for each dimension of redundancy matrix, a set of constructed elementary bricks can be simply assembled for subsequent uses.

[0142] It should be noted that a printed circuit board may in some cases be wide enough to accommodate several bricks. For example, one PCB ('Printed Circuit Board') may integrate bricks B1 and B2 and another PCB may integrate bricks B3 and B4. However, in such cases, it is advantageous to minimize the number of interconnections between bricks B2 and B3 and between B4 and B1.

[0143] The connection grids describe a connectivity, not a mechanical arrangement (the permutation of rows or columns does not affect this connectivity). The redundant equipment 20' can be represented at the end of the connection grid (last columns in the figures) to highlight the diagonals. In [Fig.9], the redundant equipment 20' is distributed among the nominal equipment 20 to highlight the elementary bricks.

[0144] To further optimize the structure of the redundancy matrix 10, in embodiments, a redundancy matrix 10 may be underpopulated, by removing connection points, thereby simplifying the hardware implementation of the SPnT^ switches and thus reducing the complexity of the redundancy matrix.

[0145] In the remainder of the description, reference will be made to the perfection rate of a redundancy matrix.

[0146] In general applications of the invention, the reconfiguration event of the redundancy matrix is ​​the change of the excluded set (in an application of the invention to fault management, the exclusion of a piece of equipment is triggered by a fault of this piece of equipment).

[0147] A matrix is ​​perfect if for any subset of excluded devices it is capable of connecting each channel to a non-excluded device (provided that there are at least as many non-excluded devices as channels).

[0148] As used herein, the perfection rate PLk of a 10 to n redundancy matrix

[0149]

[0150]

[0151]

[0152]

[0153]

[0154] inputs and m outputs is defined by: _ 1 {combinations of k exclusions covered} | k “ | {combinations of k exclusions} | Let: _ ] { combinations of k covered exclusions} | (1) fLk “ 7^ km For example, in an application of the invention to fault management, the perfection rate of a redundancy matrix 10 designates the number of faults covered by the redundancy matrix, divided by the total number of possible faults. The matrix is ​​said to be “perfect” when this number is equal to 1. Thus, in this example of application of the invention to fault management, for a redundancy matrix with n inputs and m outputs, the perfection rate PLk at the kth fault is defined as: | {combinations of k covered purlins} | _ | {combinations of k covered purlins} | k “ U combinations of k purlins { | An underpopulation of the redundancy matrix 10 could in certain cases degrade the perfection rate of the matrix but with a very progressive reduction in the reliability of the subsystem constituted by a set of equipment, including nominal equipment 20 and redundant equipment 20', framed by two redundancy matrices 10 according to embodiments of the invention. However, an underpopulated 10 redundancy matrix allows for hardware simplification and therefore a reduction in cost and therefore a good compromise between perfection rate and complexity of the hardware structure of the redundancy matrix. Figure 10 represents a sliding redundancy matrix of dimension 16x20, according to another example, configured to manage 3 specific failures. This redundancy matrix 10 makes it possible to illustrate the connection points which can be removed to implement the redundancy matrix embodiment with sliding and underpopulation. The redundancy matrix of Figure 10 is obtained from the matrix structure of Figure 6 (management of two first failures on E4 and E^Y by adding in addition connection points according to the sliding method to ensure continuity of service in the event of a 3rd failure on E5; - channel C4 is rerouted for the first failure on E4; - for the 2nd failure on E6, the C6 channel is rerouted to E7, C7 is rerouted to £g, then C8 is rerouted to R2, thanks to the addition and activation of the connections (C6 £7), ( C7 E^ and (C8 R2) to manage a 2nd failure; thus R2 compensates for the loss of equipment E^; - for the 3rd failure on the channel C5 is rerouted to E7, C7 is rerouted to £9, C9 is rerouted to E} b C]} is rerouted to £13 and finally C13 is rerouted to R3 thanks to the addition and activation of the connections (C'5 E7\ (C7 E9\ (C9 E}A\ ( C] f ** E13) and (C13 ** R3) to handle a third failure.

[0155] The nodes (C6 E7\ (C7 Es) and (Cg Z?2) linked to the 2nd breakdown will be replaced by the nodes (C6 ** E^) and (C8 7?2) to allow the node (C5 Ef) to be established.

[0156] It should be noted that these additions of connections reveal a third neighboring diagonal D7 higher than the second neighboring diagonal D\.

[0157] In the embodiment of redundancy matrix 10 with sliding and underpopulation, the connection points corresponding to the connections added to the second failure on E^ or the points (C6 ** E^ e,t(C7^ E^) can be advantageously deleted, while retaining the capacity of the redundancy matrix to manage the 2nd and 3rd failure using the connection points added for the 3rd failure.

[0158] [Fig.l 1] illustrates such a simplification of the redundancy matrix (of dimension 16x20 composed of 4 interconnected 4x5 sub-matrices).

[0159] Figure 11 highlights the importance of the high diagonal D3,_ Such a diagonal has a preponderant weight on the perfection rate.

[0160] In the embodiments of the invention, the depopulation of the redundancy matrix thus only concerns the intermediate diagonals, which can thus be at least partially eliminated, while the highest diagonal is preserved.

[0161] Figure 11 shows the deletion of points (C6 E7) and (C7 E^\ the points of connections added by sliding to manage a 3rd failure (here on E5) allowing to manage the 2nd failure (here on E^. To manage the 3rd failure: - equipment E^ is indirectly replaced by R2, during a 2nd failure on E6 thanks to the addition and activation of the connection (C6 ** Es) and the connection (C8 R2) added to manage a 3rd failure (channel Cb connects to E* and channel C8 to R2); - equipment E5 is indirectly replaced by R3, for the 3rd failure on E5 thanks to the addition and activation of the connections (C5 E7\ (C7 ** E9\ (C9 ^£^),( Ci 1 ** 3) and (Cj3 R^; thus, during the 3rd failure, channel C5 connects to E7 , channel C7 on £9, channel Cg on Ej 1, channel Cj 1 on £13, which leads to the connection of channel C]3 which is connected to R^.

[0162] Thus in embodiments, the redundancy matrix 10 may have hollow diagonals such that for an equal number of diagonals, the “highest” diagonals (of higher rank l) make it possible to cover a greater number of failures than the neighboring diagonals of lower rank (of higher rank ?).

[0163] The redundancy matrices 10 in which neighboring diagonals of intermediate rank l are removed entirely or partially (1 l < L where E is the total number of neighboring diagonals) will hereinafter be called “sparse diagonal sliding redundancy matrix” or “sub-sparse sliding redundancy matrix”. populated” (i.e. redundancy matrices in which neighboring diagonals of intermediate rank / are not completely filled with nodes).

[0164] Such sparse diagonal sliding redundancy matrices are not perfect. Indeed, they have a perfection rate strictly less than 1. However, the deletion of intermediate diagonals, rather than the "high" neighboring diagonal, corresponding to the DL diagonal of highest rank L, makes it possible to obtain matrices with a higher perfection rate.

[0165] The embodiments thus provide a sliding redundancy matrix 10 defining connection points on the one hand between the channels (rows of the matrix) and the nominal equipment and on the other hand between channels and redundant equipment for the rerouting of a channel 3 to a redundant equipment 20' in the event of one or more failures of the nominal equipment 20.

[0166] The “Sliding Redundancy Matrix” thus makes it possible to define the connectivity of the system 100, i.e. the set of connections permitted between the channel accesses 3 and the equipment accesses 20 / 20'. This connectivity described by the positioning of connection nodes in the connection grid representing the matrix and can be critical since it determines the compromise between the complexity of the system and the reliability of the system.

[0167] The complexity of the redundancy matrix, in terms of the number of SPnT switches corresponding to a connection, is directly proportional to the number of connection nodes in the connection grid. By optimizing the number of connections in the connection grid representing the redundancy matrix 10, the number of SPnT switches used to manage equipment failures is optimized.

[0168] Furthermore, the connectivity represented by the connection grid associated with the redundancy matrix 10 determines the set of equipment failure combinations that the redundancy matrix 10 is capable of supporting. An insufficient number of connection nodes or a non-optimal placement of the nodes could significantly degrade the reliability at the payload level.

[0169] The redundancy matrix can belong to two categories of sliding redundancy matrices: - perfect matrices, and - non-perfect matrices.

[0170] A perfect matrix is ​​configured to support all combinations of faults and has well-defined connectivity.

[0171] Non-perfect matrices are obtained from perfect matrices by adapting two parameters: - A first parameter corresponding to a complexity modulation parameter defined to modulate the complexity of the matrix (number of connection nodes), - A second parameter corresponding to a parameter relating to the perfection rate defined to adapt the perfection rate of the matrix (percentage of failure cases covered) as needed.

[0172] To describe a sliding redundancy matrix, different categories of paths that make up the matrix are distinguished: - nominal paths, - redundant paths, and - the sliding paths.

[0173] A nominal path of the redundancy matrix is ​​used in the absence of faults on the equipment to connect each channel 3 to its nominal equipment 20. On the connection grid representing the redundancy matrix, a nominal path comprises the set of connection points (Q ** £)), forming a first full diagonal 0, corresponding to the nominal diagonal (a nominal path comprises each connection between a channel C; of given index i and the equipment of the same index).

[0174] A redundant path allows each channel 3 to access only one of the redundant devices 20' (there is therefore only one redundant node per line). A redundant node is a connection node between a channel 3 and a redundant device 20'. On the connection grid representing the redundancy matrix 10, a redundant path comprises the set of connection points (C, ** R.) between a channel C\ and a device R. By construction, a redundant path is located on the right part of the redundancy matrix.

[0175] However, the position of the columns of the redundant elements 20' (order of the redundant elements) in the connection grid does not affect the connectivity of the underlying matrix (as shown in [Fig.8]), i.e., the order of the rows and columns does not impact the connectivity of the matrix.

[0176] When failures occur, redundant paths are preferred since they do not cause transient interruptions of neighboring channels during reconfiguration.

[0177] A sliding path makes it possible to cover a wider set of fault combinations. The sliding paths allow the operation of redundant equipment 20' located far from the faulty equipment thanks to a sliding phenomenon from near to far. On the connection grid corresponding to the redundancy matrix 10, they form the additional neighboring diagonals which can be more or less full and which are positioned above the nominal diagonal D. On the connection grid representing the redundancy matrix 10, a sliding path sliding includes the set of connection points (Q ** E.) between a channel C( and a piece of equipment , with E. belonging to a neighboring diagonal Db

[0178] A full neighboring diagonal D[ includes the set of connection points (G i + / mod n)-

[0179] [Fig. 12] illustrates the 3 types of connection node paths in an example of a perfect 16x20 10 redundancy matrix.

[0180] The redundancy matrix 10 has several degrees of freedom for positioning connection nodes for a given matrix dimension (for a given number of channels 3 and equipment 20), such connection nodes serving to reroute the channels 3 to redundant equipment 20' or to other nominal equipment 20 (sliding paths), in the event that one or more failures occur on the equipment 20 of the system 100.

[0181] The definition of a sliding redundancy matrix 10 is associated with a sliding capacity, noted e, representing the thickness of the diagonal formed by the nominal paths and the sliding paths, as illustrated by [Fig.13].

[0182] The total number of diagonals (nominal and neighboring) is thus equal to the parameter e. Thus, the number of neighboring diagonals is equal to E = e — 1.

[0183] According to the embodiments of the invention, an nxm matrix is ​​perfect if its sliding capacity e must satisfy the condition:

[0184] e > mn(2)

[0185] In particular, an nxm matrix of sliding capacity e ~m~ n is perfect if and only if all its sliding path diagonals are full.

[0186] For sliding paths to exist in the redundancy matrix 10 and for the matrix to be of sliding matrix type, the capacity e is greater than or equal to 2.

[0187] For the matrix to be decomposable into chainable elementary bricks, the sliding capacity e is less than or equal to [n / (m - «)].

[0188] The embodiments of the invention thus provide a redundancy matrix of dimension nxm having a sliding capacity verifying the condition:

[0189] 2 <e < (3)

[0190] In the example of figure 13, this condition is satisfied, the capacity e being equal to 4 (^ = 20 - 16 allows the matrix to be perfect).

[0191] The upper bound of inequality (3) makes it possible to avoid too high a number of diagonals which would make possible slides beyond the next group (towards Gk + 2 for example).

[0192] The number L designating the number of neighboring diagonals, without the nominal diagonal (with L = e - 1), then verifies:

[0193] 1 < L < [ ] - 1 (4)

[0194] A sliding matrix is ​​further defined by the allocation of redundant equipment 20'.

[0195] Although it is advantageous to optimize the division of the matrix into elementary bricks to distribute the Q channels in blocks and in an equitable manner among the redundant equipment (for example 4 channels per redundant equipment for a matrix of dimension 16x20), a sliding matrix can nevertheless be perfect despite an irregular distribution of the channels among the redundant equipment 20' (the number of channels per redundant equipment can vary from one redundant equipment to another). Thus, it is sufficient that each redundant equipment 20' is connectable to at least one channel 3, that is to say that none of the columns associated with the redundant equipment 20' is empty (i.e. that each column associated with a redundant equipment 20' includes at least one connection with a channel 3).

[0196] Figure 14 represents an example of a perfect sliding redundancy matrix, of dimension 16x20, despite an irregular distribution of the channel associations 3 and redundant equipment 20': is associated with the first 7 channels (Co, C6), while each redundant equipment R] to Rj is associated with 3 channels (Z?i to C7-C9, Z^2 to C io-C12 and R3 to C!3-C 15).

[0197] Another adjustable characteristic of the connection grid associated with a sliding redundancy matrix 10 is the degree of filling of the general diagonal delimited by the nominal diagonal D and the neighboring diagonal of highest rank L. In order to optimize the complexity of the hardware implementation of a redundancy matrix, nodes located between the diagonal of the nominal paths (nominal diagonal D) and the diagonal of the highest-rank sliding paths (high diagonal) can be deleted or partially depopulated (only a few nodes can be removed from the intermediate diagonals Dj with / between 1 and L- 1).

[0198] Figure 15 illustrates an intermediate case of filling while Figure 16 illustrates a case of minimal filling to highlight the trade-off between the reliability of the redundancy matrix and the complexity of the redundancy matrix (expressed here for an implementation based on SPDTy switches

[0199] A case of maximum filling of redundancy matrix corresponds to figure 13, with a complexity in number of SPDT switches used to realize the hardware implementation of the matrix of 124 (124 SPDT switches) and a perfection rate at the 4th failure of PL4 = 100%, which corresponds to a perfect matrix.

[0200] Figure 15 corresponds to an intermediate filling of sliding redundancy matrix (sparse matrix), of dimension 16x20, with a complexity in number of SPDT switches used to carry out the hardware implementation of the matrix of 76 (76 SPDT switches) and a perfection rate at the 4th failure PL4 = 96.62%.

[0201] Figure 16 thus corresponds to a minimum filling of sliding redundancy matrix (sparse matrix), of dimension 16x20, with a complexity in number of SPDT switches used to carry out the hardware implementation of the matrix of 60 (60 SPDT switches) and its perfection rate at the 4th failure is PLA = 90.67%.

[0202] The diagonal of the highest slip paths DL (of highest rank) has an important role for the performance of the matrix. In preferred embodiments, the highest diagonal DL (of highest rank) of the redundancy matrix is ​​advantageously full. Thus, even in a minimum filling matrix configuration, the connection frame associated with the redundancy matrix can advantageously comprise at least the nominal diagonal D of the nominal paths and the diagonal DL of the high slip paths.

[0203] Embodiments of the invention provide an optimized connection device corresponding to the hardware implementation of one or more sliding redundancy matrices 10 represented by a connection frame, the connection device comprising one or more SPnT switches according to the connection frame.

[0204] The redundancy matrix 10 according to the embodiments of the invention is thus advantageously compatible with SPnT switches without, however, presenting the prohibitive defects of the prior art.

[0205] In particular, the redundancy matrix structure according to the embodiments of the invention has a structure which can be regular to greatly simplify the balancing of the paths.

[0206] The implementation of the connection grid can be carried out with a circuit in trees of SPnT switches, which makes it possible to adjust the length, and therefore the attenuation, of each path individually.

[0207] Examples of SPnt switches include, but are not limited to, STDP, SP3T, SPAT, or SPnT switches.

[0208] The number of SPDT switches used to implement a K-node N x M connection grid is equal to TK — N — M.

[0209] It should be noted that the hardware implementation of the redundancy matrix is ​​not unique. There are indeed several degrees of freedom on the construction of the switch trees.

[0210] The redundant equipment 20' can be distributed among the channels to allow a division into elementary bricks of the complete system 100, which is particularly advantageous: - in situations where, for mechanical constraints, for example, the dimension of the printed circuit used for the implementation of the redundancy matrix is ​​too limited to integrate the entire system, the division then optimizing the implementation with a limited number of interconnections between bricks; - to develop generic, reusable bricks, covering a wide range of needs when assembled together, which optimizes development costs.

[0211] The redundancy matrix is ​​furthermore adaptive. Indeed, the perfection rate of the matrix can be modulated by adjusting the thickness of the diagonal of the matrix and the degree of filling of the redundancy matrix. The complexity of the hardware implementation of the matrix can thus be optimized to best meet the reliability constraints (for example in the case of an application of the invention to the space industry, the reliability of the payload of the system 100). Such granularity does not exist with the redundancy matrices of the prior art.

[0212] The classes of sliding redundancy matrices according to the embodiments of the invention are thus defined by the parameters n!mi e (integers) comprising: - The parameter n which designates the total number of channel accesses of a redundancy matrix belonging to the class. - The parameter m which designates the total number of equipment accesses of a redundancy matrix belonging to the class. - The parameter e which denotes the sliding capacity of a redundancy matrix of the class. Such a quantity represents, on the connection grid of the matrix, the thickness of the diagonal formed by the nominal paths and the sliding paths.

[0213] According to the embodiments of the invention, the integers n, m, and e satisfy the following conditions: n > 0, m>n + l, 2 < e < [ ml n 1 ($)

[0214] Thus, to belong to the class of sliding redundancy matrices, the thickness e of the block of diagonals, including the nominal diagonal, corresponding to the graphical representation of the total number of diagonals, must satisfy condition (5).

[0215] The notation [ u}J- n 1 denotes the integer part of

[0216] A sliding redundancy matrix {n / m / f] is defined as any redundancy matrix with n channel accesses and m equipment accesses whose connection grid includes: - a set of nominal paths, which form the first diagonal called the “nominal diagonal” D, the nominal paths being defined by:

[0217] Vie [0; II - 1 ] , Cl connectable to El- a set of redundant paths, distributed over the m " 11 redundant equipment 20' so that each channel C, is connectable to a single redundant equipment 20' and that each redundant is connectable to at least one channel and defined by:

[0218] 3 / : [ 0; / / -l ] I 0;mnl ]] surjective / Life [[ 0;nl ] , Q connectable to

[0219] The connection grid may further comprise a set of sliding paths, which form at least one full “high diagonal” DL and optionally partially filled “intermediate diagonals” D) defined by:

[0220] Life I 0: » - 1 ]] , C. connectable to E. . ,, . ,

[0221] Sie>2.9 S c Π0; w -1 ]] x [[ 1; e-2 ]] / V (i, k) e S, Ct connectable to Ei + kn

[0222] In one embodiment, the function / may be surjective and increasing.

[0223] In one embodiment, the redundancy matrix 10 may be configured to so as to achieve the optimum in terms of population of the matrix so as to obtain a target reliability for the a subset of the system 100 consisting of nominal equipment 20 framed by an input redundancy matrix 10-A and an output redundancy matrix 10-B, and associated redundant equipment 20'.

[0224] Figure 17 represents such a subset of the system 100 comprising nominal equipment (Eo - El5) and redundant equipment 20 (^o ' ^3) (subsystem) as well as an input redundancy matrix 10-A and an output redundancy matrix 10-B.

[0225] The inventors have established that there is no universal optimal depopulation level of a redundancy matrix since the reliability of the subsystem is not only a function of the perfection rates of the matrix but also of the reliability of the equipment connected to the redundancy matrix (nominal and redundant).

[0226] Determining the optimum underpopulation of a redundancy matrix 10 consists of partially or completely removing intermediate diagonals from a redundancy matrix while retaining the nominal diagonal D (“nominal diagonal”) and the highest-ranking diagonal DL (“high diagonal”). The 'partial' removal of an intermediate diagonal corresponds to the removal of only certain nodes from the intermediate diagonal.

[0227] At least some or all of the intermediate diagonals Dk (of rank l < L) can be removed from the redundancy matrix 10.

[0228] From the connection grid of each redundancy matrix 10 (which thus comprises at least the nominal diagonal D and the high diagonal DL, which can be completed by one or more intermediate diagonals ^ / ), the hardware implementation of the corresponding redundancy matrix 10 can be produced from the connection grid in the form of a circuit (as used here, the hardware “implementation” of the redundancy matrix 10 thus refers to the circuit corresponding to the connection grid).

[0229] In one embodiment, the circuit used to implement the redundancy matrix 10 may be based on SPnT switch trees,

[0230] It should be noted that this switch tree-based implementation is not specific to sliding redundancy matrices and can be applied more generally to a redundancy matrix associated with any type of connection frame.

[0231] The SPnT switch trees of the circuit implementing a redundancy matrix 10 can be generated by taking each access of the redundancy matrix 10 as the root of one of the trees (a "redundancy matrix access" corresponds to a channel (row) access of the matrix or an equipment access of the matrix (column)). Thus, the circuit implementing the redundancy matrix 10 can comprise a tree per row and per column of the connection grid. A SPnT switch tree can thus be constructed at each channel or equipment access of the matrix, a channel / equipment access being connected to the root of the tree associated with it.

[0232] A circuit tree, implementing the redundancy matrix 10, connected to a channel access can have as many branches as there are connection nodes on the corresponding line in the connection grid.

[0233] A tree of the circuit 10 whose root is connected to an equipment access can have as many branches as there are connection nodes on the corresponding column in the grid.

[0234] The shape of a shaft can be varied. The circuit 10 can, for example, use balanced shafts or unbalanced shafts.

[0235] Each node in the connection grid on row L; and column T corresponds to a branch on a root switch tree connected to C. and a branch J on a switch tree with root connected to E. or R j.

[0236] These two branches are connected to each other to allow the connection CE or R r

[0237] Adding a node in the connection grid on a row L and a column T. involves adding a branch on the root switch tree Q and on the root switch tree E. (or R

[0238] Examples of switch trees are shown in [Fig. 18], in the case of SPDT type switches.

[0239] Figure 18 shows examples of k (( k - 1) SPDT) branch SPDT switch trees 50 that can be used in an implementation of the sliding redundancy matrix 10.

[0240] To form circuit 10, the paths described by the connection grid are then implemented by connecting the channel access trees to the access trees equipment, branch to branch. The shape of the tree branches to be connected together can vary.

[0241] [Fig. 19] represents a perfect 16x20 sliding redundancy matrix 10. The complete matrix represented in [Fig. 19] can be divided into 4 identical elementary bricks Bl, B2, B3 and B4.

[0242] [Fig. 19] highlights the interconnections between elementary bricks and shows that it is advantageous to attach redundant equipment to the brick of its group to avoid having to create new interconnections between bricks.

[0243] Figure 20 shows an exemplary implementation of the perfect 16x20 sliding redundancy matrix 10 shown in Figure 19, in the form of a circuit comprising SPDT switch trees,

[0244] The circuit 500 of Figure 20 corresponds only to the brick B1 and therefore only to a part of the final circuit implementing the redundancy matrix 10. The implementation of the other bricks of Figure 19 is similar by adapting the channel and equipment accesses. As shown in Figure 20, the circuit 500 comprises a tree connected to each channel access C, with as many branches as there are nodes on the corresponding row in the connection grid and a tree connected to an equipment access E. or with also as many branches as there are connection nodes on the corresponding column in the grid.

[0245] It should be noted that a sub-matrix of a redundancy matrix does not only have channel and equipment accesses but can also have incoming interconnection accesses Hi (in rows) and outgoing interconnection accesses IOj (in columns).

[0246] Some switch trees may be found straddling two sub-matrices (between 2 PCBs for example), that is to say that the root and part of the tree will be on the first board and the rest will be on the other, as can be seen in Figure 20. The incoming interconnection ports II i (“Inter In”) are not the roots of the trees (they are located on sub-matrix B4). Similarly, the last trees do not stop at the outgoing interconnection ports (“Inter out”) but continue on sub-matrix B2. This cutting in the middle of the trees makes it possible to minimize the number of interconnections between the sub-matrices.

[0247] Switch commands may depend on the implementation and nature of the switches used.

[0248] In an exemplary embodiment, when a fault condition is detected, a set of nodes can be identified in the connection grid which allows a new valid routing, i.e. which will associate an operational device (non-faulty device) with each channel.

[0249] Different algorithms may be used to enable the identification of a valid routing, if any, for a given fault condition (e.g., algorithms based on graph theory, maximum matching and / or bipartite graphs).

[0250] Once the nodes to be used are identified, the switches are controlled, depending on the switch's own technology, topology and position in the circuit.

[0251] In SPDT switch tree-based embodiments, a path can then be established between the root of each tree and the branch corresponding to the node to be used in the connection frame.

[0252] Advantageously, the embodiments of the invention make it possible to obtain a hardware implementation of a redundancy matrix having optimized connectivity, i.e. an optimization of all the channel-equipment connections that they can establish.

[0253] In a particular example of application, the embodiments of the invention can be used to replace heavy, bulky and expensive redundancy rings of the prior art, by the redundancy matrices according to the embodiments of the invention which can be based for example on miniaturized electromechanical SPDT switches, in the field of the space industry.

[0254] The invention can be applied more generally to any industrial application using systems or subsystems using combinational switching matrices, i.e. realizing the combinations of N inputs on M outputs, such as for example redundancy matrices, switching matrices in test systems, selectivity matrices, etc.

[0255] Furthermore, the invention is not limited to the embodiments described above as a non-limiting example. It encompasses all the variant embodiments that may be envisaged by those skilled in the art. In particular, those skilled in the art will understand that the invention applies to any type of application using redundancy matrices to reroute signals from a piece of equipment to a redundant or nominal piece of equipment.

[0256] Furthermore, although the invention has been made with reference to SPnT type switches, those skilled in the art will understand that the invention is not limited to such types of switches. For example, the hardware implementation of the redundancy matrix can use a so-called "cross bar" implementation and C-switch or R / T switches, the redundancy matrix 10 then being replaced by a total selectivity matrix which allows the connection of any input to any output, with a full connection grid. For our sliding redundancy matrix, the cross-bar network would be heavily depopulated to reveal only the nodes described by the connection grid.

Claims

Claims

1. System (100) comprising a set of equipment (20, 20'), said equipment comprising one or more nominal equipment (20) and one or more redundant equipment (20'), a redundant equipment (20') being associated with one or more nominal equipment (20), the system comprising n channels connected to said equipment, the channels connected to the equipment being input or output channels of the equipment, the system comprising at least one redundancy matrix (10) of dimension nxm configured to route the signals of a channel of a given equipment to a redundant equipment (20') associated with said given equipment or to another nominal equipment (20), in response to the detection of a rerouting trigger event relating to an equipment, characterized in that each redundancy matrix is represented by a connection grid having a matrix representation comprising rows and columns,each row corresponding to a channel 0 < n of the system and each column corresponding to a piece of equipment (20, 20') of the system, and in that the connection grid comprises a set of connection points corresponding to a point of intersection between a channel and a piece of equipment, said connection points of the connection grid comprising at least:, - the set of intersection points of a nominal path comprising the intersection points of the connection grid located on a nominal diagonal of the connection grid, between each row corresponding to a channel of given index / and a column corresponding to a nominal device of the same index i, and - the set of intersection points of a redundant path comprising the intersection points of the connection grid located at the intersection between each row corresponding to a channel Q of given index i and a column corresponding to a redundant element of index / (i) such that the function / is a surjective function defined from the set [ (); n - 1 ]] on [ (); m - n - 1 J for any index i, the redundancy matrix being implemented in the form of a circuit comprising a plurality of switches arranged according to the connection grid.

2. System according to claim 1, characterized in that each redundancy matrix is a sliding redundancy matrix and in that said connection points of the connection grid further comprise: - the set of intersection points of a sliding path comprising the intersection points of the connection grid located on an upper diagonal of the connection grid, between each line corresponding to a channel C( of given index i and a nominal device of index i + (e - 1) modulo n with e designating a sliding capacity of the matrix and e strictly greater than two, the sliding capacity being the number of points of the matrix located between a given point of the nominal diagonal on a given column of the connection grid and a point of the upper diagonal located on the same column of the connection grid.

3. System according to claim 1, characterized in that each redundancy matrix is a sliding redundancy matrix and in that the sliding path further comprises the intersection points of the connection grid located on one or more intermediate diagonals of the connection grid, between each line corresponding to a channel Cj of given index i and a nominal device of index + k mo(fuio n, with ( i, k ) belonging to S and SCI 0; ?î - 1 ] X [[ 1 ; e - 2 J .

4. System according to one of the preceding claims 2 and 3, characterized in that if n is strictly greater than zero and m is strictly greater than n + 1 the sliding capacity satisfies the following condition: 2 se < [ m _ n ]

5. System according to claim 4, characterized in that a perfect redundancy matrix has a sliding capacity satisfying the condition: em - H.

6. System according to claim 5, characterized in that a sliding capacity redundancy matrix = is perfect if and only if all the points of the intermediate diagonals of the sliding path are connection points.

7. System according to one of claims 5 and 6, characterized in that a perfect redundancy matrix has a perfection rate equal to one, the perfection rate representing the number of rerouting trigger events covered by the redundancy matrix.

8. System according to one of the preceding claims, characterized in that it comprises an input redundancy matrix associated with the input channels of the equipment and an output redundancy matrix associated with the equipment output channels.

9. System according to one of the preceding claims, characterized in that the switches are of the SPnT type.

10. System according to claim 9, characterized in that the circuit implementing the redundancy matrix is based on SPnT switch trees.

11. System according to claim 10, characterized in that the circuit implementing the redundancy matrix (10) comprises a tree corresponding to each row and a tree corresponding to each column of the connection grid, the root of each tree of SPnT switches of the circuit corresponding to a row of the connection grid associated with a channel or to a column of the connection grid associated with a device.

12. System according to claim 11, characterized in that a tree of said circuit implementing the redundancy matrix connected to a line corresponding to a channel has as many branches as there are connection points on said line in the connection grid and a tree of said circuit implementing the redundancy matrix connected to a column corresponding to a device has as many branches as there are connection points on said column in the connection grid.

13. System according to one of claims 11 and 12, characterized in that the trees corresponding to the rows of the connection grid are connected to the trees corresponding to the columns of the connection grid.

14. System according to one of the preceding claims, characterized in that it is configured to manage a number of failures on said equipment, said rerouting condition being the occurrence of a failure on a piece of equipment.

15. System according to one of the preceding claims, characterized in that the number of equipment failures supported by the system is an increasing function of the sliding capacity of the matrix, when the intermediate diagonals are all full.

Images