WO2005091142A1

WO2005091142A1 - Method and device for determining a capacity of a communication link of a network and network system

Info

Publication number: WO2005091142A1
Application number: PCT/SG2004/000061
Authority: WO
Inventors: Xiaofei Cheng; Tee Hiang Cheng; Chao Lu
Original assignee: Agency For Science, Technology And Research
Priority date: 2004-03-19
Filing date: 2004-03-19
Publication date: 2005-09-29

Abstract

A computer-based link capacity is determined for a communication network comprising primary paths and secondary paths which are each associated to a plurality of primary paths, wherein the primary paths and the secondary path each comprises at least one link. A plurality of sets comprising a plurality of links is then selected, wherein each link of the selected sets of links is included in a primary path, which is associated to a secondary path comprising the link, the capacity of which is to be determined. Further, for each set of the plurality of selected sets, a candidate capacity of the link is determined, such that, if the link comprises the determined candidate capacity, secondary paths comprising the link can act as backup paths for all their associated primary paths if all the links in the respective set fail and the maximum candidate capacity of the determined candidate capacities of the link according to each selected set is selected as the capacity.

Description

Method and device for determining a capacity of a communication link of a network and network system

Background of the Invention

The invention relates to a method and a device for determining a capacity of a communication link of a network and a network system.

A survivable communication network is a communication network that has the ability to restore connectivity in the event of communication network element failures. Two key approaches to designing a survivable communication network are protection and restoration. Protection is a scheme that reserves a backup path (BP) at the same time the active path (AP) is established. Restoration is a scheme that will find a suitable backup path dynamically from the pool of available resources when a fault occurs.

Many communication network applications are mission critical and hence require protection. [1], [2], [3] and [4] disclose studies on protection and restoration schemes for handling a single-link failure. Though a single-link failure occurs more often, a multiple- link failure is not uncommon especially for a large communication network. A common protection approach is 1:1 path protection. When a new connection request arrives, a link- disjoint pair of an active path and a backup path is chosen for the connection. The active path is established and the bandwidth in the backup path is reserved between the source node and destination node. When one or more link failures affect the active path of a connection, the connection will be switched to the backup path. l:n protection is a more general case in which protection bandwidth does not need to be dedicated to one connection but can be shared among multiple link-disjoint connections such that bandwidth can be utilized more efficiently.

Sharing of protection bandwidth among a few connections for a single link failure is studied in [1] or [2]. Consider an example of a network comprising five communication links which are denoted by a, b, c, d and e, respectively. Let there be two further connections in the network wherein a first connection has a bandwidth requirement of w_\ bandwidth units and a second connection has a bandwidth requirement of vv₂ bandwidth units, and wherein the active path of the first connection traverses link a and the active path of the second connection traverses link b and both the backup paths of the first connection and the second connection traverse link e. Since the two active paths will not be affected at the same time by any single-link failure, the backup bandwidth of the two connections on link e can be shared. By sharing the protection bandwidth, only the maximum of W and w₂ (max { w^vt^}) units instead of ι i+» ₂ units of bandwidth need to be reserved on the link e.

Summary of the Invention

The invention solves the problem of determining a link capacity of a communication link of a communication network comprising a plurality of communication links and comprising primary paths and secondary paths, wherein the primary paths and secondary paths each comprise communication links and the secondary paths are each associated to a primary path or to a plurality of primary paths.

The problem is solved by a computer-based method for determining a link capacity of a communication link of a communication network, preferably to protect against multiple failures of communication links, a computer-based method for controlling a communication network, a device for determining a link capacity of a communication link of a communication network and a communication network system with the features according to the independent claims.

According to the invention, there is provided a computer-based method for determining a link capacity of a communication link of a communication network comprising a plurality of communication links and comprising a plurality of predetermined primary paths and one or preferably a plurality of predetermined secondary paths which each is associated to one or preferably a plurality of primary paths, wherein the primary paths and the secondary paths each comprise at least one link, the method comprising the steps of: Selecting a plurality of sets comprising a plurality of links, wherein each link of the selected plurality of sets of links is included in a primary path, which is associated to a secondary path comprising the link, the capacity of which link is to be determined; determining, for each set of the plurality of selected sets, a candidate capacity of the link, such that, if the link comprises the determined candidate capacity, secondary paths comprising the link can act as backup paths for all their associated primary paths if all links in the respective set fail; selecting the maximum candidate capacity of the determined candidate capacities of the link according to each of the at least one selected sets as the capacity.

In a preferred embodiment, when there is a request for a new connection in a communication network, the following steps are taken:

- Selecting a primary path and a secondary path for the new connection. A link-disjointed primary path and secondary path for the new connection are selected based on a routing algorithm (e.g. an algorithm to determine a shortest path, etc.), wherein each path comprises a plurality of links, the capacity of these links are to be determined;

- determining/allocating bandwidth for each link of the working (primary) and the protection (secondary) path. For links of the primary path, a total connection bandwidth is allocated to it to transform the new connection. For each link of the secondary path, a plurality of fault link sets is formed. For each of the fault link sets, a candidate backup bandwidth is calculated; the maximum candidate backup bandwidth of all fault link sets is allocated to the link.

The commumcation network is assumed to comprise a set of nodes and a set of edges. The edges are also called links. A connection between a first node of the network and a second node of the network is a tuple of links that forms a path from the first node to the second node. A capacity is an amount of bandwidth. A primary path is a connection in the network, which, in a preferred embodiment, is used as an active path for a connection. A secondary path is also a connection in the network, which, in a preferred embodiment, is used as a backup path for a connection.

Further, according to the invention there is provided a device adapted to perform the method described above.

Further, according to the invention there is provided a computer based method for controlling a communication network comprising a plurality of communication links and comprising one or preferably a plurality of predetermined primary paths and one or preferably a plurality of predetermined secondary paths, preferably one for each connection, which is preferably each associated to at least one primary path, wherein the primary paths and the secondary paths each comprise at least one link, wherein: In case that the setup of a new connection is requested, a primary path and a secondary path for the connection, each comprising at least one link, are determined and for each link of the secondary path of the new connection, a capacity is determined by selecting a plurality of sets comprising a plurality of links, wherein each link of the selected plurality of sets of links is included in a primary path, which is associated to a secondary path comprising the link; determining, for each set of the plurality of selected sets, a candidate capacity of the link, such that, if the link comprises the determined candidate capacity, secondary paths comprising the link can act as backup paths for all their associated primary paths if all links in the respective set fail; selecting the maximum candidate capacity of the determined candidate capacities of the link according to each of the at least one selected sets as the capacity.

Further, according to the invention there is provided a computer based method for controlling a commumcation network comprising a plurality of communication links and comprising at least one predetermined primary path and at least one predetermined secondary path which is associated to one or preferably at least one primary path, wherein the at least one primary path and the at least one secondary path each comprise at least one link, wherein: In case that the termination of a connection is requested, for each link of the secondary path of the connection, a capacity is determined by selecting a plurality of sets comprising a plurality of links, wherein each link of the selected plurality of sets of links is included in a primary path, which is associated to a secondary path comprising the link; determining, for each set of the plurality of selected sets, a candidate capacity of the link, such that, if the link comprises the determined candidate capacity, secondary paths comprising the link can act as backup paths for all their associated primary paths if all links in the respective set fail; selecting the maximum candidate capacity of the determined candidate capacities of the link according to each of the at least one selected sets as the capacity.

Further, according to the invention there is provided a communication network system comprising: A communication network comprising a plurality of communication links and preferably comprising at least one predetermined primary path and at least one predetermined secondary path which is associated to at least one primary path, wherein the at least one primary path and the at least one secondary path each comprise at least one link; a routing device for determining a primary path and a secondary path for a requested new connection, each comprising at least one link; a capacity determining device for determining the capacity of each link of the secondary path of a requested new connection or a connection requested to be terminated, respectively, by selecting a plurality of sets comprising a plurality of links, wherein each link of the selected plurality of sets of links is included in a primary path, which is associated to a secondary path comprising the link; determining, for each set of the plurality of selected sets, a candidate capacity of the link, such that, if the link comprises the determined candidate capacity, secondary paths comprising the link can act as backup paths for all their associated primary paths if all links in the respective set fail; selecting the maximum candidate capacity of the determined candidate capacities of the link according to each of the at least one selected sets as the capacity.

A basic idea of the invention may be seen in selecting sets of links, wherein each link of the selected sets is included in a primary path which is associated to a secondary path comprising the link whose capacity is to be determined, and then to calculate for each such selected set a candidate capacity of the link such that, if the link comprises this candidate capacity, secondary paths comprising the link, preferably all secondary paths comprising the link, can protect their associated primary paths against a simultaneous failure of all the links in the respective set and to select the maximum candidate capacity of the candidate capacities according to each set as the capacity to be determined.

One advantage of the method, the device and the network system according to the invention is that there is provided a protection against multiple link failures, not just against single link failures.

A further advantage is that the amount of links, a simultaneous failure of which should be protected against, can be chosen freely. Thus, the invention can be used for a wide range of types of networks and levels of protection.

Additionally, since the invention provides sharing of backup bandwidth, the provided protection is very efficient concerning the bandwidth requirement.

In a preferred embodiment, the bandwidth corresponding to the determined capacity is reserved on the link. According to one aspect of the invention, in the case that there is a request for a new connection, a primary path and a secondary path for the connection are determined and the capacity of each link of the determined secondary path for the new connection is determined and preferably, a corresponding bandwidth is reserved.

In a preferred embodiment, in the case that for one set selected for one of the links of the determined secondary path for the new connection a candidate capacity cannot be determined such that, if the link comprises this candidate capacity, secondary paths comprising the link can protect their associated primary paths against a simultaneous failure of all the links in the respective set, the candidate capacity is determined to be infinite and the new connection is determined to be impossible to be set up.

Thus, only connections are set up which are protected against multiple link failures and no bandwidth is unnecessarily reserved for connections which cannot be established or cannot be properly protected against link failures.

According to another aspect of the invention in the case that there is a request for a connection to be terminated, the capacity of each link of the secondary path of the coimection to be terminated is determined and preferably a corresponding bandwidth is reserved.

Thus, after the termination of a connection, only as much bandwidth is reserved on the links as is still required. The bandwidth which is no longer required is released.

Clearly, the invention provides a scheme to share backup bandwidth among multiple connections in order to guard against multiple link failures in a multi-rate circuit-switched communication network or a statistically multiplexed communication network where an equivalent bandwidth could be defined for each connection. Such communication networks include the automatically switched optical communication network (ASON) currently being standardized by ITU-T Study Group 15 and Generalized Multiprotocol Label Switching (GMPLS) controlled optical communication networks that support fibre-switch capable, lambda switch- capable and TDM-capable nodes. Brief Description of the Drawings

Illustrative embodiments of the invention are explained below with reference to the drawings, in which

Figure 1 shows an embodiment of a communication network system according to the invention.

Figure 2 shows a simple example of a communication network.

Figures 3(a) to 3(h) illustrate the necessary backup bandwidths in an embodiment where there should be a protection against double link failures in the communication network shown in Figure 2.

Figure 4 shows a flow diagram according to an embodiment of the method according to the invention wherein a new connection is set up.

Figure 5 shows a flow diagram according to an embodiment of the method according to the invention wherein a connection is terminated.

Figure 6 shows the commumcation network of figure 2, wherein the active path and the backup path of connection 6 are highlighted.

Figures 7(a) and 7(b) show two communication networks which are simulated for a performance analysis.

Figures 8(a) to 8(d) show the simulation results of the simulation performed on the communication networks of Figures 7(a) and 7(b).

Detailed Description of the Invention

Figure 1 shows an embodiment of a communication network system according to the invention. A communication network 10 comprises a plurality of links and nodes (not shown), some of which are a first access node 11, a second access node 12, a third access node 13, a fourth access node 14 and a fifth access node 15. A first computer 20 is connected to the communication network 10 via the first access node 11. A second computer 21 is connected to the communication network 10 via the fourth access node 14. If the first computer 20 needs a connection to the second computer 21, the first computer 20 sends a connection request which is received by a commumcation network controlling device 40. The connection request includes, for example, information about the bandwidth needed for the connection, a specification of the ingress node of the requested connection, which is in this case the first access node 11 and a specification of the egress node of the requested connection, which is in this case the fourth access node 14.

The communication network controlling device 40 comprises a memory 41, a routing device 42 and a bandwidth determining device 43. In the memory 41, information about the communication network 10 is stored. For instance, information about active paths and backup paths of already established connections and information about bandwidth reservations are stored in the memory 41. If a connection request is received, the routing device 42 determines an active path and a backup path, which connect the ingress node and the egress node specified by the connection request. In this embodiment, the active path and the backup path are determined in such a way that they are link disjoint. The determination may comprise Dijkstra's shortest path algorithm or another suitable algorithm.

The bandwidth determining device 43 determines the amount of bandwidth which has to be reserved on the links of the communication network. In particular, according to the invention, the bandwidth determining device determines the bandwidth which has to be reserved on the links which are part of backup paths. For determining the amount of bandwidths to be reserved, the bandwidth determining device 43 uses the information stored in the memory 41. An embodiment of the method for determining the amount of bandwidths to be reserved according to the invention is described below.

In one embodiment, the communication network controlling device 40 determines, based on the result of its bandwidth determining device 42, whether the requested connection can be established, and if not, sends a message to the first computer 21, including the information that the connection request is denied. Alternatively, in one embodiment, the communication network controlling device 40 might use its routing device 42 to generate a new pair of active path and backup path different from the first for the connection, which is again provided to the bandwidth determining device for testing if it can be set up.

If the connection request is not denied, and if all necessary determinations are done, the communication network controlling device 40 sends messages to the communication network 10, in particular to the respective nodes, including the necessary commands for making the determined bandwidth reservations on the links of the backup path along the nodes, the bandwidth reservations on the links of the active path according to the bandwidth specified in the connection request and for establishing the requested connection.

In this embodiment, the connection between the communication network controlling device is separated from the access nodes. The command messages may in another embodiment be transmitted directly via the requested ingress node, that is, in this case, the first access node 11 and along the active path and backup path of the requested connection.

The communication network 10 receives the command message from the communication network controlling device 40 and, if no problems arise, sends a message back to the communication network controlling device 40 confirming that the connection is established and all bandwidth reservations are made. For example, in one embodiment, the egress node sends an ACK message back to the ingress node and the ingress node sends a confirmation signal to the communication network controlling device 40. The communication network controlling device updates its memory 41 according to the new connection, that is, it stores for example the active path and the backup path of the connection and the amount of bandwidth which has been reserved on the links of the communication network. Further, the communication network controlling device 40 informs the first computer 20 about the successful establishment of the requested connection. Again, this can happen via the ingress node, the first access node 11, itself.

Then, the first computer 20 begins with the transmission of data.

A similar process is carried out when the first computer 20 has finished its business with the second computer 21 and no longer require the connection to the second computer 21. Then, the first computer 20 sends a connection termination request message which is received by the communication network controlling device 40. The communication network controlling device 40 determines the active path and the backup path of the connection to be terminated by accessing the memory 41, in which the necessary information is stored. The bandwidth determining device 43 determines the amounts of bandwidth which still have to be reserved on the links of the communication network 10 after the termination of the connection. Analogous to above, the communication network controlling device 40 sends a command message to the communication network 10, which, if the termination is successful, in turn sends a confirmation message to the communication network controlling device 40. The communication network controlling device 40 then updates its memory 41 according to the terminated connection.

Figure 2 shows a simple example of a communication network, which is used to illustrate a particular problem which is solved by the invention and to exemplify the bandwidth efficiency for the two cases in which sharing of protection bandwidth among multiple connections to guard against multiple link failures are allowed and prohibited. The active paths of six connections, numbered as 1, 2, 3, 4, 5 and 6, traverse links a, b, c and d, and all their backup paths (not shown) traverse links e. It is assumed that the bandwidth requirement of each arbitrary connection is w, (wt=i units).

Figures 3(a) to 3(h) illustrate the necessary backup bandwidths for an embodiment where there should be a protection against double link failures. The numbers 1 to 6 correspond to the six connections shown in figure 2.

With reference to figure 3(a), to protect against the failures of link a and link b simultaneously, the total protection bandwidth on link e should be ΪV_I+W2+W₃. Similarly, with reference to figures 3(b), (c), (d), (e), (f), to protect against two simultaneous link failures, the protection bandwidth on link e should be w₁+w₂+.v₄+.V₅, wι+W₂+ws+wβ, W₂+W₃+W₅+W₆, W₄+W₅+W₆, respectively. Since the cases illustrated in figure 3(a) to 3(f) are assumed not to occur at the same time for two simultaneous link failures, and since in this embodiment there is only a protection against simultaneous failures of two links, the backup bandwidth reserved on link e to guard against two simultaneous link failures can be shared. Hence, only W = max{w₁+w₂+.v +.V₅, W₁+>V₂+VV₅+H'₆, W₂+H>₃+W₅+W₆, H>₄+W₅+W₆} units of backup bandwidth need to be reserved. Figure 3(g) shows the backup bandwidth required for this example, assuming that W - W2+W3+W5+W6. If no sharing of protection bandwidth is allowed and no other information except the residual bandwidth of link e is known when a connection needs to be set up, a total of vt>₁+w₂+w₃+H> +W₅+. ₆ units of protection bandwidth needs to be reserved, as depicted in figure 3(h).

Figure 4 shows a flow diagram according to an embodiment of the method according to the invention. This embodiment of the method corresponds to the embodiment of the communication network system according to the invention described above with reference to figure 1.

In step SI, the process is activated by the arrival of a new connection request, say k, at the communication network controlling device 40.

In step S2, the routing device 42 selects the active path and backup path for the connection based on some routing algorithms. It is assumed that the active path traverses a total ofp links; namely, a_\, c₂,.., α,-, ... , active path and the backup path traverses a total of q links; namely, b\, b₂,.., ty, ... , b_q.

In step S3, the iterative process starts with the first link of the backup path, i.e., b_\.

In the following steps, the following notations are used. ' : Set of connections whose active path traverse link e. A e : Total bandwidth of all connections in A' e •

_g' : Set of connections whose backup path traverse link e.

B : Backup bandwidth on link e for the set of connections in β' . Since sharing of protection bandwidth is allowed, B_e ≤ w_k .

R_e : Residue bandwidth of link e; i.e., R_e = C -A_e -B_e where C is the capacity of link e. ⁰ : Set of connections whose active path traverse link a and backup path traverse link b. " : Total bandwidth for all the connections in S'„ . This is the total bandwidth of all connections whose active path traverse link a and backup path traverse link b. I ^e _{a b c,}..._n '• Set of connections whose active path traverse links a, b, c, ... , n and whose backup path traverse link e. This set can be induced from S and V^e _a = S^K _a . _b X- ^Total bandwidth for all the connections in set I __ACt. _Λ ; That is total bandwidth of all connections whose active path traverse links a, b, c, ... , n and whose backup path traverse link e. U - _{b c,}.._Λ '• Set °f connections whose active path transverse any link a, b, c, ..., n and whose backup path traverse link e, i.e., U^κ _a = S^κ _a .

Uϊ_{j> c,}.._ji '• Total backup bandwidth needed on link e for all the connections in set

C_{a b c,}.._j> • ' Additional backup bandwidth needed on link e in order to use it as part of a backup path for a new connection to protect against simultaneous failures of links α, b, c... n.

C_e : Additional backup bandwidth needed on link e when a new connection is established.

With the above definitions, the following equations, which are used in the following steps, hold:

s: = ^_t (3)

/ ,,,__.- s _ nS n...nS^,e„ (4)

K»,,., = ∑w_k (7)

To provide further formulas used in the following steps, let E = {E, 11 ≤ -^" ≤ E} be the set that contains all the links in the communication network. Here, E denotes the total number of the links in the communication network. Let the number of simultaneous link failures considered be n. When a new connection with a bandwidth requirement of w is requested, the backup bandwidth has to be reserved on each link of the backup path of the new connection. It is assumed that the backup path traverses link e (eG ). The reserved bandwidth on link e should protect against n link failures. Let a set of links whose simultaneous failures are protected by e be Lf ={Eι,L₂, ...L„ }, E/C L. When the set of links in Lf fail simultaneously, all the connections whose active path traverses any link in Lf and backup path traverses link e should be protected by the reserved bandwidth on link e. As such, the total bandwidth needed to protect these connections on link e is:

A proof for this equation by induction over n is given in the following:

For n=l and n=2, U^e = S and U^e ^ = S + S^ -/^ , respectively. Substituting n with 1 and 2 in equation (8) yields the same expressions. Hence, equation (8) is correct for n=l and n=2.

For n=k>2, suppose Equation (8) is true: u-_M

^ ,. ₄nS =(S^nSι_i U(S!₂nS!_t u...U(S!_inS' )=/^,^_4tιu/' _4tιu...u/'!₄,_iJ

From equation (8% one obtains:

) '

If one substitutes n in Equation (8) with k+1, one also obtains the above equation. Hence, the induction is complete, and equation (8) is proven.

As an example for equation (8), consider the case 1 in figure 3(a) with Lf =(a,b). When link a and b fail simultaneously, the total bandwidth of protected connections on link e should be: 2 U'_Jt = C S_L ^e _ι -I_a ^e _Jb) = (w, + w₂) + (w₂ + w₃) - w₂ = ^ + w₂ + w₃

If the bandwidth already reserved on link e is denoted by B_e, to protect against the n simultaneous failures of links L_\∑₂,...L_n, the additional backup bandwidth needed on link e for setting up a new connection k is:

AP is the link set of the primary path of the new connection k. Note that C^ ^ =∞ means that it is not feasible to set up the new connection with a backup path that traverses link e.

In step S4, for each link, d, in L (the set that contains all the links in the communication network), S , which is the set of connections whose active path traverses link d and backup path traverses link bj is computed. This can clearly be done by using the equations (1) and (2). S J is calculated for usage in the following step.

In step S5, L is formed by including every link L, , whereby S **r is not an empty set. Let

L consists of m links. S ^ can be obtained from S ,V<2EE computed in step S4. The purpose of forming this set is to remove those irrelevant links from the universal set to reduce the number of times steps 6 to 10 need to be iterated later on. L is the set of all links which are part of an active path of a connection, which has a backup path including b_j. From this interpretation, it is clear that L includes all links whose failure is relevant for b_j. If a link fails which is not in L , then an active path including this link cannot be used, but since the backup path of this active path does not include bj, by definition of L , this failure has no impact on bj.

In step S6, a set of n links is selected from the m links in L in each iteration and there are ,L'₂ ,..E'_n } is used to represent the

set of n links selected in each iteration. In step S7, Cj j , _χ, , which is the backup bandwidth needed on link b_j to protect against simultaneous failures of links L ,L ,..17_n , is computed. This is done with the equations (1) to (9).

In step S8, it is tested if C Λ _τ, _r, = ∞. If this is the case, it implies that link bj cannot provide protection for the new connection.

Accordingly, in step S9, if CA _τ , , , = ∞ in step S8, the setting up of the new connection ^•*' 1>* 2'- ^•** n k cannot be completed and the process is terminated and returns to step SI to wait for the arrival of another coimection request.

In step S10, if all the possible sets of L have gone through steps S6 to S10, the process proceeds to step Sll. Else, another set of L is selected and is put through steps S6 to S10.

In step Sll, the additional bandwidth C_b needed on bj to cater for protection of the new connection is computed and added to B_b . , which is the total backup bandwidth needed on link b_j to protect all the connections whose backup path traverses b_j. C_b is calculated with the equation

C_e = maxCf. ,. ,. (10)

wherein e = b_j and the maximum is generated over all subsets L of L with n elements and wherein the elements of L ate denoted by L[,L₂' ,...,L_n' .

In step S12, it is tested if this iteration was for the last link, b_q, in the backup path of the new connection. If not, then the iteration continues, else the loop is finished.

In step S13, index / is incremented such that the next link in the backup path of the new connection will be used in the next iteration. In step S14, after the process has been repeated for all the links in the backup path of the new connection, the new connection is set up and the sets A' ,Vi and the sets B'_b, ,V/ are updated and stored in the memory 41 with the newly set up connection before the process terminates and returns to step 1 to wait for another new connection set up request.

Accordingly, the communication network controlling device 40 informs all the nodes in the active path and backup path to set up the new connection.

Figure 5 shows another flow diagram according to an embodiment of the method according to the invention. In contrast to figure 4, this flow diagram illustrates the process which according to this embodiment is carried out if a connection is terminated. This embodiment of the method corresponds to the embodiment of the communication network system according to the invention described above with reference to figure 1.

When a connection is terminated, the reserved bandwidth of all links that constitute the active path and backup path of the connection and some other information need to be updated. The flow as depicted in figure 5 is described below:

In step Tl, the process is activated by the arrival of a new connection termination request, say k, in the communication network controlling device 40.

In step T2, the active path and backup path of coimection k are obtained from the memory 41 of the communication network controlling device 40. It is assumed that its active path traverses a total oip links, namely,

--₍-, ... ,activepath and the backup path traverses a total of q links; namely, b &₂, ... ,b_j, ... ,b_q.

In step T3, the iterative process starts with the first link of the backup path, i.e., b_\.

In step T4, for each link, d, in L (the set that contains all the links in the communication network), S , which is the set of connections whose active path traverses link d and backup path traverses link b_j, is computed. This is done with the equations (1) and (2). In step T5, L is formed by including every link L_r , whereby S _L is not an empty set. It is

assumed that consists of m links. S can be obtained from S J ,\/d GL computed in

Step T4. The purpose of forming this set is to remove those irrelevant links from the universal set to reduce the number of times steps T6 to T8 need to be iterated later on. Steps T4 and T5 of figure 5 are similar to steps S4 and S5 of figure 4.

In step T6, a set of n links is selected from the m links in L in each iteration and there are ,..V_n } is used to represent the set

of n links selected in each iteration.

In step T7, U ,, _τ, , which is the total backup bandwidth needed for the set of connections whose active paths transverse any link L ,V ,..L and whose backup paths traverse link b_h- is calculated. This is done with the equations (1) to (9).

In step T8, it is tested if steps T6 to T8 have been repeated for all the possible sets of t . If this is the case, the process proceeds to step T9. Else, another set of L is selected and is put through steps T6 to T8.

In step T9, the backup bandwidth B_b needed on bj to cater for protection of all the connections that traverse link bj is obtained. B_b is calculated with the equation

B. = maxU^bH ,. (11) ^bt vi. eZ "^L^-^L" '

In step T10, if this iteration is not for the last link, b_q, in the backup path of the connection that is to be terminated, the iteration continues. Else, the loop is finished.

In step Til, the index is incremented such that the next link in the backup path of the connection will be used in the next iteration. In step T12, after the process has been repeated for all the links in the backup path of the connection, the bandwidth is released and set _A'_fl. ,V and set B . ,Vj are updated and stored in the memory 41 with the connection just terminated removed from the set before the process terminates and returns to step Tl to wait for another new connection termination request.

For a better understanding of the method according to the invention, in the following, two examples are given. Each uses the simple communication network shown in figure 2. The first example illustrates the process executed when the connections 1 to 5 are already established and the connection 6 is set up. The second example illustrates the process executed when the connections 1 to 6 are already established and the connection 5 is terminated.

For the examples, the communication network status is assumed to be as follows:

(a) Connection routing table: 1: active path={a}; backup path={e,f,b} ; 2: active path={a,b}; backup path={e,f} ; 3: active path={b}; backup path={f,e,a} ; 4: active path={c}; backup path={e,g,d} ; 5: active path={c,d}; backup path={e,g} ;

(b)

{4,5}

(c) Bandwidth requirement: w = {w\, w₂, W3, vv₄, vv₅}; Wi = w₂ = W3 = vv₄ = w>5 = vv₆ = 1 unit

(ά)B_a = l-,B_b =l;B_c = 0 -,B_d =l_;-3_e = 5 ,B_f = 3;B_g = 2;

In the two examples, there should be a protection against simultaneous failures of two links, i.e., n=2. In the first example, the connection number 6 as shown in figure 2 should be set up.

Figure 6 shows the communication network shown in Figure 2, wherein the active path and the backup path of connection 6 are highlighted.

The method according to the embodiment of the invention described above proceeds in the following way upon a connection request for connection 6:

1. Choose active path and backup path for the 6^th connection active path={d} backup path={g,e,c}

2. The iterative process loops over each link of the backup path, (1) For link g (a) For each link, /, in L (the set that contains all the links in the communication network), obtain S'f from A', 2?' , S'f = A'.QB' : 5^, ={};S' = {}_:5^, ={4^}_:S'«={5}_:S' ={};S^={};S^={} (b) Obtain: L={c,d} (c) Obtain: Uξ₄ =S + SJ -/*„ =2+1-1 = 2 and Cξ₄ =1 (d)C,-C -l <e)2>_f -.B_{f +}C,-3

(2) For link e (a) For each link, /, in L (the set that contains all the links in the communication network), obtain S^*f from A' ,B' , S7 = A', ]B'_e :

S'l ={1,2}; S* ={2,3}; S' = {4,5}; S ={5}; S {}; S* = {} ; S ={} (b) Obtain: L ={a,b,c,d} i)L={a,b} K_b -S_a ^e +S_b ^e-r_aιb =2 + 2-1 = 3 C^* -0 2)L={α,c} U_t' -S.'+Sc'-l c =2 + 2-0 = 4

3)Z-{ }

4)L={6,c} U: -S:+S:-L 2+2-0=4 -° 5)E={M} ^ =5₆ ^e+S_d ^e -7^=2+1-0 = 3 c^e =0 6)I={c,d} ^=5_c ^e+S -/_e ^e _,rf =2+1-1=2 C_c =0 (c) C_e =max{C₎₆, _)C, _,dC₆ ^e _jC,C_fc ^e _ιd,C_ι(/} = 0 (d)β_e=5_e+C_e=5

(3) For link c (a) For each link, /, in L (the set that contains all the links in the communication network), obtain S from A) ,B , S^* = -4,(1-9'- : 5 {};S {};S {};S {};5 {};S'^e _/-{};5^*-{} (b) Obtain: f ={} (c)C_c=w = l (d)B_c=B_c+C_c=l

4. "active path Setup" packets are sent, in the embodiment described above by the communication network controlling device 40, to all nodes of the active path, for instance from the ingress node to the nodes along the active path to establish the requested connection, which, preferably, include a unique connection identifier, the explicit route and the bandwidth requested w₆.

5. "backup path Setup" packets are sent, in the embodiment described above by the communication network controlling device 40, to all nodes of the backup path, for instance from the ingress node to the nodes along the backup path to reserve the backup bandwidth, which include the explicit route and the reserved bandwidth: B^φ_j ≡BP).

6. When the communication network 10 signals that no problems have shown up, for instance by an ACK signal received by the ingress node about the active path and backup path channels from the egress node, then the following sets are refreshed in the memory 41:

(a) Connection routing table: 1: active path={a}; backup path={e,f,b} ; 2: active path={a,b}; backup path={e,f} ; 3: active path={b}; backup path={f,e,a} ; 4: active path={c}; backup path={e,g,d} ; 5: active path={c,d}; backup path={e,g} ; 6: active path={d}; backup path={g,e,c}

(b) A = {1,2}_; Λ'₆ = {2,3}_; A {4,5}; A = {5,6}_; A^< _e = {}-, A'_f = {}; A'_g ={} B ={3}-, B => {1}-, B = {6}-, -3 {4} ; B {1,2,3,4,5,6} ; B'_f = {1,2,3}; B = {4,5,6}

(c) Bandwidth requirement: w={

w>₂, W₃, H>₄, VV_S. W>₆} (ά) B_c =l; B_e = 5 ; B_g = 3;

Then, the connection setup for connection 6 is completed. In the second example, the connection number 5 as shown in figure 2 is terminated. It is assumed that the 6^th connection has been set up according to the above example, such that the communication network comprises the set up connections 1 to 6.

The method according to the embodiment of the invention described above proceeds in the following way upon a termination request for connection 5:

1. Refresh routing table, A' B' sets (get rid of the connection 5 information from the sets): (a)

={} ={3}; B ={1}; B ={6}_; B_d ={4}_; -8'_β -{1,2,3,4,6}; B ={1,2,3}; -5', ={4,6} (b) Bandwidth requirement: M> = { ivi, w₂, u>3, w₄, M>₆}

2. The iterative process loops over each link of the 5^th backup path (1) For link g (a) For each link, /, in L (the set that contains all the links in the communication network), obtain S'f from

: 5^, ={};S^, ={};5^,f={4};S'5={6};S' ={};S'^={};S'f={} (b) Obtain: L={c,d} (c) Obtain: U_c ^g _>d =S_C* +Sj -I_t*₄ =1+1-0 = 2 (d)B_f-C^-2 (2) For link e (a) For each link, /, in L (the set that contains all the links in the communication network), obtain S * from A , B ' , S = A f]B : S'^e _a ={1,2}; S^,6 ₆ ={2,3}; S- ={4}; = {6} ; S* = {}; S'^e _; = {} ; S'^e _g = {} (b) Obtain: L={a,b,c,d} l)L={a,b}

2)E={α,c} K_c -SI +5_e ^e - = 2 + 1-0 = 4 3) E ={α,<*} U_^e _,d = S_a ^e ₊S_d ^e -i:_td = 2 + 1-0 = 3 4) E = {b,c}

5) L ={b,d}

6) E = {c,d} ^ = 5_c ^e +S_d ^e -/_c =1 + 1-0 = 2 (c) *_e = max _b,U _c,U_a ^e _td,U _c,U _d,U _d = 4

4. "active path Release" packets are sent, in the embodiment described above by the communication network controlling device 40, to all nodes of the active path, for instance from the ingress node to the nodes along the active path to establish the requested connection, which, preferably, include the released connection identifier, the explicit route and the bandwidth requested w₆.

5. "backup path Release" packets are sent, in the embodiment described above by the communication network controlling device 40, to all nodes of the backup path, for instance from the ingress node to the nodes along the backup path to reserve the backup bandwidth, which include the explicit route and the reserved bandwidth: B. b_j BP).

6. When the communication network 10 transmits a confirmation signal, the following stets are refreshed and stored in the memory 41: (a) Connection routing table: 1: active path={a}; backup path={e,f,b} 2: active path={a,b}; backup path={e,f} 3: active path={b}; backup path={f,e,a} 4: active path={c}; backup path={e,g,d} 6: active path={d}; backup path={g,e,c} (b) Bandwidth requirement: w_{ w H>₂, W₃, W> , W₆}

(c) B_e = 4,-B_g = 2;

Then, the connection termination for connection 5 is completed.

In the following, a performance analysis is given based on a simulation.

Figure 7 shows the two communication networks which are simulated. The one shown in figure 7(a) has 7 nodes and 12 links, the one shown in figure 7(b) has 15 nodes and 26 links.

In the simulation, connection requests arrive sequentially and for each connection, the source node and destination node are chosen randomly. The bandwidth requirement of each connection is assumed to be uniformly distributed between 1 and 10 units. In order to compare the bandwidth efficiency between the two cases — with sharing and without sharing of backup bandwidth — the capacity of all the links is set to infinity. When a new connection arrives, a path-disjointed active path and backup path are chosen using Dijkstra's shortest path algorithm. Ten independent simulations were carried out with different seeds and the results were averaged.

Figure 8(a) to 8(d) show the simulation results. In figures 8(a) to 8(b), NS denotes the case for which no sharing of backup bandwidth is allowed, SP/k denotes the case for which the backup bandwidth is allowed to be shared for protection against k link failures. An indicator called Bandwidth Saving Ratio (BSR) is defined to measure the extent to which bandwidth can be saved from sharing, as follows:

BSR = (Total bandwidth required for NS - Total bandwidth required for SP) / (Total bandwidth required for NS)

For the simulation, the number of connections was increased from 0 to 120. In figure 8(a), in the case in which no sharing of backup bandwidth (NS) is allowed, the backup bandwidth on each link is the sum of the backup bandwidth of all connections whose backup path traverses the link. In the case in which sharing is allowed, the bandwidth is more efficiently utilized. Also, the backup scheme that caters for only one link failure is more efficient than the backup scheme that caters for 2 link failures.

Figure 8(b) shows that bandwidth efficiency is lower when more link failures need to be guarded against. When the connection request increases, BSR increases. This is attributed to the fact that more active paths can now share a particular amount of backup bandwidth because the algorithm according to the invention does not limit the number of connections sharing their backup bandwidth. The level of sharing increases when traffic increases and thus improves the bandwidth saving ratio. The same phenomenon is observed when figure 8(b) and 8(d) are compared.

In this document, the following publications are cited:

[1] Chunming Qiao, Dahai Xu,"Distributed Partial Information Management (DPIM) Scheme for Survivable Communication networks-Part I", INFOCOM 2002. Vol 1, pp. 302 - 311, 2002. [2] Murali Kodianalm,TN.Lakshman, "Dynamic Routing of Bandwidth Guaranteed Tunnels with Restoration," IEEE, IΝFOCOM 2000, pp.902 - 910,2000.

[3] Ramamurthy S., Mukherjee B., "Survivable WDM Mesh Networks, Part I - Protection", INFOCOM '99. Eighteenth Annual Joint Conference of the IEEE Computer and Communications Societies. Proceedings. IEEE , Volume: 2 , 21-25 March 1999 Pages:744 - 751 vol.2

[4] US 2003/0065811 Al

Claims

Claims:

1. A computer-based method for determining a link capacity of a communication link of a communication network comprising a plurality of communication links and comprising a plurality of predetermined primary paths and at least one predetermined secondary path which is associated to a plurality of primary paths, wherein the primary paths and the at least one secondary path each comprises at least one link, the method comprising: - selecting a plurality of sets comprising a plurality of links, wherein each link of the selected plurality of sets of links is included in a primary path, which is associated to a secondary path comprising the link, the capacity of which link is to be determined; - determining, for each set of the plurality of selected sets, a candidate capacity of the link, such that, if the link comprises the determined candidate capacity, secondary paths comprising the link can act as backup paths for all their associated primary paths if all links in the respective set fail; - selecting the maximum candidate capacity of the determined candidate capacities of the link according to each of the at least one selected sets as the capacity.

2. The method according to claim 1, wherein the communication network is a multi- rate circuit-switched communication network or a statistically multiplexed communication network.

3. The method according to claim 1, wherein the communication network is an optical communication network.

4. A computer based method for controlling a communication network comprising a plurality of communication links and comprising at least one predetermined primary path and at least one predetermined secondary path which is associated to at least one primary path, wherein the at least one primary path and the at least one secondary path each comprises at least one link, wherein: - in case that the setup of a new connection is requested, a primary path and a secondary path for the connection, each comprising at least one link, are determined and for each link of the secondary path of the new connection, a capacity is determined by - selecting a plurality of sets comprising a plurality of links, wherein each link of the selected plurality of sets of links is included in a primary path, which is associated to a secondary path comprising the link; - determining, for each set of the plurality of selected sets, a candidate capacity of the link, such that, if the link comprises the determined candidate capacity, secondary paths comprising the link can act as backup paths for all their associated primary paths if all links in the respective set fail; - selecting the maximum candidate capacity of the determined candidate capacities of the link according to each of the at least one selected sets as the capacity.

5. A computer based method for controlling a communication network comprising a plurality of communication links and comprising at least one predetermined primary path and at least one predetermined secondary path which is associated to at least one primary path, wherein the at least one primary path and the at least one secondary path each comprises at least one link, wherein: - in case that the termination of a connection is requested, for each link of the secondary path of the connection, a capacity is determined by - selecting a plurality of sets comprising a plurality of links, wherein each link of the selected plurality of sets of links is included in a primary path, which is associated to a secondary path comprising the link; - determining, for each set of the plurality of selected sets, a candidate capacity of the link, such that, if the link comprises the determined candidate capacity, secondary paths comprising the link can act as backup paths for all their associated primary paths if all links in the respective set fail; - selecting the maximum candidate capacity of the determined candidate capacities of the link according to each of the at least one selected sets as the capacity.

6. The method according to claim 4, wherein, in case that for one of the plurality of sets selected for one of the links of the secondary path of a requested new connection, a candidate capacity cannot be determined such that, if the link comprises the determined candidate capacity, secondary paths comprising the link can act as backup paths for all their associated primary paths if all links in the respective set fail, the candidate capacity is determined to be infinite, and the requested new connection is determined to be impossible to be set up.

7. The method according to any one of the claims 4 or 6, wherein bandwidths corresponding to the determined capacities are allocated to the corresponding links of the secondary path of the requested new connection or of the connection to be terminated, respectively.

8. The method according to any one of the claims 4 to 7, wherein the communication network is a multi-rate circuit-switched commumcation network, or a statistically multiplexed communication network.

9. The method according to any one of the claims 4 to 7, wherein the communication network is an optical communication network.

10. The method according to any one of the claims 1 to 9, wherein the candidate capacity of the link for each set of the plurality of selected sets is determined by use of the equation

u^e ,_L₂,„J

wherein the following denotations are used: - -7_a,6>c,...n denotes the total backup bandwidth needed on a link e for all the connections whose primary path comprises any one of the links a, b, c, ... ,n and whose secondary path comprises the link e; - S* denotes the total bandwidth of all connections whose primary path comprises link a and whose secondary path comprises link b; - ^._C._JI i^{s me tota}l bandwidth of all connections whose primary path comprises links a, b, c, ... ,n and whose backup path comprises link e.

11. A device for determining a link capacity of a communication link of a communication network comprising a plurality of communication links and comprising a plurality of predetermined primary paths and at least one predetermined secondary path which is associated to a plurality of primary paths, wherein the primary paths and the at least one secondary path each comprises at least one link, the device comprising a determination unit which determines the capacity of a link of a communication network by:

- selecting a plurality of sets comprising a plurality of links, wherein each link of the selected plurality of sets of links is included in a primary path, which is associated to a secondary path comprising the link, the capacity of which link is to be determined; - determining, for each set of the plurality of selected sets, a candidate capacity of the link, such that, if the link comprises the determined candidate capacity, secondary paths comprising the link can act as backup paths for all their associated primary paths if all links in the respective set fail; - selecting the maximum candidate capacity of the determined candidate capacities of the link according to each of the at least one selected sets as the capacity.

12. The device according to claim 11, wherein the commumcation network is a multi- rate circuit-switched communication network or a statistically multiplexed communication network.

13. The device according to claim 11, wherein the communication network is an optical communication network.

14. The device according to any one of the claims 11 to 13, wherein the candidate capacity of the link for each set of the plurality of selected sets is determined by use of the equation

^^... , ⁼ T,^S1 ^" λ 2 χ. , ⁺ z X + - + (-l)^π ^^,..j.„) 1=1 1=1 7-1+I ι=l 7-1+M.-7+I

wherein the following denotations are used: - U_a ^e _{i) C} .._.„ denotes the total backup bandwidth needed on a link e for all the connections whose primary path comprises any one of the links a, b, c,...,n and whose secondary path comprises the link e; - S* denotes the total bandwidth of all connections whose primary path comprises link a and whose secondary path comprises link b; - 1 _{b c n} is the total bandwidth of all connections whose primary path comprises links a, b, c, ... ,/ι and whose backup path comprises link e.

15. A communication network system comprising: - a communication network comprising a plurality of communication links and comprising at least one predetermined primary path and at least one predetermined secondary path which is associated to at least one primary path, wherein the at least one primary path and the at least one secondary path each comprises at least one link; - a routing device for determining a primary path and a secondary path for a requested new connection, each comprising at least one link; - a capacity determining device for determining the capacity of each link of the secondary path of a requested new connection or a connection requested to be terminated, respectively, by - selecting a plurality of sets comprising a plurality of links, wherein each link of the selected plurality of sets of links is included in a primary path, which is associated to a secondary path comprising the link; - determining, for each set of the plurality of selected sets, a candidate capacity of the link, such that, if the link comprises the determined candidate capacity, secondary paths comprising the link can act as backup paths for all their associated primary paths if all links in the respective set fail; - selecting the maximum candidate capacity of the determined candidate capacities of the link according to each of the at least one selected sets as the capacity.

16. The communication network system according to claim 15, further comprising a capacity allocation device for allocating bandwidths corresponding to the determined capacities to the corresponding links of the secondary path of the requested new connection or of the connection to be terminated, respectively.

17. The communication network system according to claim 15 or 16, wherein the communication network is a multi-rate circuit-switched communication network or a statistically multiplexed communication network.

18. The communication network system according to claim 15 or 16, wherein the communication network is an optical communication network.

19. The communication network system according to any one of the claims 15 to 18, wherein the candidate capacity of the link for each set of the plurality of selected sets is determined by use of the equation

⁺-- -ⁱ⁾ⁿ χ.,..χ„⁾

wherein the following denotations are used: - U_a ^e _{Jtl c} „ denotes the total backup bandwidth needed on a link e for all the connections whose primary path comprises any one of the links a, b, c, ... ,n and whose secondary path comprises the link e; - S* denotes the total bandwidth of all connections whose primary path comprises link a and whose secondary path comprises link b; - l_a ^e _Jb,c...._n ^{s me tota}l bandwidth of all connections whose primary path comprises links a, b, c, ... ,n and whose backup path comprises link e.