KR100277227B1 - Multi-level recovery method of ATM network - Google Patents

Abstract

According to the present invention, when a network failure such as fiber cutting or switch failure occurs in an ATM network, a recovery path is prepared for the virtual paths affected by the failure in advance to recover the virtual path, and a bypass path is found by using a flooding algorithm. To recover (dynamic recovery). Specifically, the first step is to implement the pre-planned virtual path recovery method considering the failure situation, and if the failure is not 100% recovered, the second step is to recover the failure using the dynamic recovery method. The multi-stage recovery method of the ATM network proposed by the present invention consists of a pre-planned virtual path recovery method and a dynamic recovery protocol in consideration of a failure situation, and the pre-planned virtual path recovery method includes a step of determining and storing a recovery path, a message transmission, and a capacity allocation step. The dynamic recovery protocol consists of a path detection phase that recognizes a failure and initiates recovery, a route discovery phase that finds a new recovery path, and identifies a discovered route and finds a new virtual path affected by the failure. It consists of a verification step that turns it into a true root.

Description

Multi-step recovery method of ATM network

The present invention uses a pre-planned virtual path recovery method and a flooding algorithm that recovers virtual paths by preparing recovery paths in advance when the network failure such as fiber cutting or switch failure occurs in the ATM network. The present invention relates to a multi-stage recovery method for an ATM network that performs a recovery method (dynamic recovery) step by step by finding a bypass route.

As a result of the convergence and integration of a large amount of communication traffic into a single network in the case of an ATM-based broadband telecommunication network, even if a single network component fails, the entire network may fail. Big. Therefore, new network planning and management functions based on ATM technology to ensure the survival and reliability of services are absolutely required.

Layers that provide resiliency for ATM networks include a physical layer, a virtual path layer, and a virtual channel layer. The recovery of the physical layer mainly requires redundant investment of network resources, such as setting up a spare circuit in case of a link failure, and the recovery techniques in an STM-based network are almost identically used. The virtual path layer recovers traffic affected by failures in units of virtual paths. For example, in the event of a link failure, the new route for these virtual paths is called a recovery virtual path, not as a view of the entire traffic flowing through the link as a unit of recovery but as a unit of recovery. Find and recover. In the virtual channel, the traffic recovery unit becomes a virtual channel. There are tradeoffs in terms of cost and recovery between these three layers, and recovery in the intermediate virtual path layer is considered the most desirable.

Restoration in the virtual path hierarchy is divided into preplanned restoration and dynamic restoration, depending on how to find the recovery path. In case of preliminary plan restoration, in order to prepare for the possible failure scenario, the route is calculated in advance for each failure scenario and the relevant information is stored. When a failure occurs, the restoration is attempted by using the planned bypass route. Dynamic recovery is a way to find the bypass path based on the network conditions at the time of failure. Dynamic recovery can flexibly respond to changes in network conditions and requires less memory. Preplanning, on the other hand, is fast, has low system complexity, and can be applied to large networks. In particular, in case of dynamic recovery, as the network size increases, the messages generated during the recovery process increase exponentially and the recovery speed becomes slow. For this reason, preplanning recovery methods are considered promising. However, dynamic recovery can be useful in the case where the pre-planned recovery is significantly changed from the pre-planned time, such as when a sudden change in traffic conditions occurs. In this case, the bypass route by the preliminary plan is no longer useful. Pre-planned recovery can be divided into state dependent considering failure situation and state independent considering failure situation. A recovery method that does not consider a failure situation is always restored to the same recovery path regardless of the failure situation by recovering through a recovery path composed of nodes or links separate from the operation virtual path. The recovery path may vary depending on the situation. The recovery method considering the failure situation has an advantage of efficiently using the free capacity and the number of the virtual path identifiers (VP Identifiers) because the calculation amount is increased because the recovery path must be determined for each failure situation.

Classified by the recovery level, recovery in the virtual path hierarchy can be divided into path recovery and link recovery. For example, if a link fails, the route of the virtual path using the link is A-B-C-D-E-F. Path recovery attempts to find a new path between the start point (A) and the end point (F) of the virtual path. On the other hand, link recovery attempts to find a path between nodes C and D at both ends of the failed link (called C-D). Link recovery is easier to execute, faster recovery time, and less memory required for the node, while path recovery is more reliable, requires less free capacity, and is much more reliable. For this reason, the path recovery method is considered suitable for ATM networks. In this case, the bypass route by the preliminary plan is no longer useful. Therefore, in general, it is preferable to attempt restoration by the prior plan restoration technique, but to use the dynamic restoration technique when the route by the preplanning is not possible with the restoration virtual path.

The existing research on dynamic recovery mainly relates to link recovery techniques. This can be seen as an extension of dynamic recovery in STM networks. If both the pre-plan recovery and dynamic recovery techniques are used at the same time, the role of the dynamic recovery technique is to restore the virtual path that cannot be recovered by the pre-plan recovery technique. In this regard, the role of dynamic recovery techniques is important to ensure the maximum recovery rate. Therefore, it is preferable to use path recovery rather than link recovery.

In case of failure of ATM network, to recover service affected by failure, state-dependent preplanned VP restoration scheme and dynamic recovery protocol Technical difficulty corresponds to a technical problem to be solved by the present invention when trying to utilize a path restoration technique to improve a restoration ratio in implementing a dynamic restoration protocol step by step. That is, a distributed processing procedure of recovery messages for resolving contention among recovery messages and improving the recovery rate of the network corresponds to this.

In the present invention, a pre-planned recovery model is proposed in consideration of a failure situation, and the path recovery scheme is used for the dynamic recovery scheme to overcome the disadvantages of the existing dynamic recovery protocol.

According to the present invention, when a network failure such as fiber cutting or switch failure occurs in an ATM network, a recovery path is prepared for the virtual paths affected by the failure in advance to recover the virtual path, and a bypass path is found by using a flooding algorithm. To recover (dynamic recovery). Specifically, the first step is to implement the pre-planned virtual path recovery method considering the failure situation, and if the failure is not 100% recovered, the second step is to recover the failure using the dynamic recovery method. The multi-stage recovery method of the ATM network proposed by the present invention consists of a pre-planned virtual path recovery method and a dynamic recovery protocol in consideration of a failure situation, and the pre-planned virtual path recovery method includes a step of determining and storing a recovery path, a message transmission, and a capacity allocation step. The dynamic recovery protocol consists of a path detection phase that recognizes a failure and initiates recovery, a route discovery phase that finds a new recovery path, and identifies a discovered route and finds a new virtual path affected by the failure. It consists of a verification step that turns it into a true root.

In the present invention, the multi-stage recovery method of ATM network for recovering the failure of ATM network

(a) executing a preplanned virtual path recovery method,

(b) executing a virtual path dynamic recovery method,

Step (a) is

(a-1) After determining the recovery path for each fault situation, send information on the path and capacity.

Determining and storing recovery paths stored in

(a-2) message transmission and capacity allocation;

(a-3) When the receiving recovery node receives the message, it recovers the traffic of the failed virtual path.

It includes a route exchange step to exchange the caliber,

Step (b) is

(b-1) the failure recognition phase to recognize the failure and initiate recovery;

(b-2) route search step to find new recovery path,

(b-3) Check the discovered route and replace the virtual path affected by the failure with the newly found route.

Include a confirmation step to change.

1 is a flow chart of a multi-stage recovery procedure of an ATM network to which the present invention is applied.

Hereinafter, embodiments of the present invention will be described in more detail.

1 is a flowchart of a multi-stage recovery procedure of an ATM network for explaining the present invention. As shown, the multi-stage recovery method of the ATM network according to the present invention includes (a) setting the operation virtual path routing and spare capacity allocation information, (b) determining the pre-plan recovery method scenario, and (c) the operation virtual path. (D) determining whether to recover the virtual path, (f) dynamically recovering the virtual path, and (g) recovering the virtual path. A step of terminating the procedure is provided, wherein step (g) is executed if the operation virtual path is restored in step (c), and step (g) is executed if the virtual path is restored in step (e). Step (d) is performed by the PVRRP method, and step (f) is performed by the flooding algorithm.

In the present invention, the pre-planned recovery model considering the failure situation is a model for determining a recovery path that minimizes the recovery cost (hop count or distance) for each failure situation and sets the virtual path identification number and bandwidth to zero before the failure occurs. Rather than allocating them, we present a mathematical model and solution for determining the recovery path with the minimum recovery cost and storing the recovery path and required capacity. The proposed model captures two nodes with different operating virtual paths and recovery paths as recovery nodes in order to make the operational virtual path and the recovery path go through the same node. Step by step, it consists of the determination and storage of recovery path, message transmission and capacity allocation, and path exchange.

In determining and storing the recovery path, after determining the recovery path for each failure situation through the path based VP Restoration Routing Problem (PVRRP) method, information on the path and capacity is stored in the recovery node. In the message transmission and capacity allocation phase, if a failure occurs by OAM cell or cell header error check, the recovery node retrieves the relevant information from the database and sends the message to the recovery node. In both recovery nodes, virtual path identification numbers are assigned and capacity is allocated. In the path exchange step, when the receiving node recovers a message, traffic from the failed virtual path is exchanged for the recovery path. The path switching process is performed at every node in the recovery path until it reaches the sending node. The VR Restoration Routing Problem (VRRP), an optimization model for determining the recovery path, determines the recovery path of all virtual paths affected by the failure to minimize the recovery cost for a single link failure. Here, the recovery cost refers to the number of hops or the distance of the recovery path. The VRRP multicommodity flow problem can be thought of as a model that adds the condition that each variety must pass through a single path, and it can be extended to multilink failure or node failure. The following is an explanation of the integer planning model for VRRP. The symbols and determinants used are:

G (N, A): given network

N: VP cross-connect node set

A: link set

(m, n): failed link

K (m, n): VP set over link (m, n)

c _ij ^k : Unit cost of VP k over link ij

q ^k : capacity of VP k

d _ij : free capacity of link (i, j)

P (k): Set of candidate paths for recovery of VP k

c _p ^k : Cost of restoring VP k to candidate path p

δ _ij ^p = 1, when candidate path p _crosses link (i, j)

= 0, otherwise

Determinant: y _p ^k = 1, when the recovery path of VP k is p

= 0, otherwise

Using the symbols above, VRRP can be represented as PVRRP.

Model: Path based VRRP (PVRRP)

Minimize

st (One)

(2)

The objective function is to minimize the recovery cost, constraint (1) is a capacity constraint that cannot exceed the free capacity of each link, and constraint (2) represents a non-divisible condition that each path of each VP must be one. Many of the proposed optimization models for ATM networks are NP-hard problems, and most of them rely on heuristics. However, the present invention is a column and constraint generation approach, which is one of the optimization techniques. ) Was applied. Looking at the structure of the PVRRP model, since the number of columns is exponentially large, the solution obtained after applying the heat generation technique, which does not include all the columns in the problem but generates only the necessary columns in the process of solving the problem, preserves the integer solution if it is not an integer. We apply the constraint generation method that adds the constraints that can exclude the real solution, and if not the integer solution, we apply the branch and bound method to find the optimal solution. The process of finding a real solution using the heat generation technique is as follows. In the dual structure of the VRRP model, the generation of new columns is related to the shortest path problem. Therefore, the shortest path problem can be solved for each VP to determine the necessity of new columns.

Heat Generation Techniques for Mistakes

Step 1: Create an initial problem (IMP) by creating several paths for each VP.

Step 2: solve the initial problem (IMP)

Step 3: After solving the shortest path problem for each VP, if there is a negative path with reduced cost, add it to the IMP and go to Step 2. Otherwise we found the best real solution.

If the optimal real solution is not an integer solution, we need to find a solution. The constraint generation method is a method of obtaining an integer solution by adding a constraint that can exclude the current real solution while preserving the integer solution. For VRRP models, the cover inequality of the backpack problem satisfies these conditions. It can be seen that the constraint.

Constraint Generation Technique for Integer Solution

Step 1: Find the lifted cover inequality through the backpack problem. If more than one lifted coverinequality is present, add them to the IMP. Otherwise go to step 4.

Step 2: Unpack the added IMP

Step 3: If the solution is an integer, the optimal integer solution is found. Otherwise go to step 1.

Step 4: Find the optimal solution by the branch limit method.

Considering the above failure conditions, pre-planned virtual path recovery is less flexible in case of multiple failures. If the fault is not solved by the pre-planned virtual path recovery, the next step to solve this problem is to recover the virtual path using the dynamic recovery protocol using the flooding algorithm.

The dynamic recovery protocol proposed by the present invention includes a fault recognition step for recognizing a failure and initiating recovery, a route search step for finding a new recovery path, and a newly found route for identifying a detected route and finding a virtual path affected by the failure. It consists of a verification step that changes to.

The first stage of disability awareness is based on Recommendation I.610 of the ITU-T. If a link failure occurs, the node that recognizes the failure (downstream node in the link) forwards the VP-AIS (Alarm Indication Signal) cell to the endpoint of the virtual path affected by the failure. Upon receiving the VP-AIS cell, the endpoint node of the virtual path transmits a VP-RDI (Remote Defection Indication) cell toward the viewpoint node of the virtual path. Intermediate nodes receiving the VP-RDI cell release the working capacity of the virtual path affected by the failure.

The root search phase, which is the second stage, begins when the node of the virtual path receives the VP-RDI cell. The start node of the virtual path first creates a restoration message and forwards it to the neighbor node. The recovery message will contain enough information to find an alternative path, including the endpoint node, the identification number of the virtual path to be recovered, the required capacity (which means the operating capacity of the virtual path affected by the failure), and the hop count ( hop count), link number on alternate path, etc. Intermediate nodes that receive the recovery message first check whether the recovery message has previously been processed for the virtual path that this recovery message is intended for recovery, then delete the recovery message if it has already been processed, or add the virtual path number to the processing list. After that, look at the hop count and check if the hop count exceeds the hop count limit. If the hop count is exceeded, the recovery message is discarded. Otherwise, execute the following procedure for all adjacent links except the one where the recovery message arrived. If the unreserved spare capacity is greater than or equal to the required capacity in the recovery message, add the number of this link to the recovery message, update the link information list for this link, and link the recovery message. Transmit to the next node through. Each node has a list of link information for adjacent links, which includes unreserved spare capacity, operating capacity, reserved spare capacity, on / off status, and reserved reserve for this link. This includes a list of alternate routes with capacity. Even in the case of a single link failure, the number of virtual paths affected by the failure can increase up to 4,096, and one node must handle more than one recovery message at the same time because the discovery of alternative routes is performed by the flooding algorithm. In other words, there is a high possibility of contention between recovery messages. To solve this, if contention occurs, the processor of the node sorts the recovery messages in order of priority. The first recovery message in the sorted list is processed, and the recovery messages for the remaining list are queued. The recovery messages in the queue are processed in order after the preceding recovery message is processed. In this regard, the queues of all nodes are managed by the First In First Out (FIFO) when there is no contention, and by the priority queue when there is contention. The priority of the recovery message may be determined according to the importance of the traffic of each virtual path that the recovery message targets, or the size of the capacity required for the recovery message. In the early stages of the recovery process, there may be a large number of Invoke messages with reserved capacity for each link. This increases the likelihood that the unreserved free capacity will be exhausted and no more recovery messages can be delivered to the next node. However, if recovery messages that have reserved free capacity are later selected as alternative routes for that virtual path during the verification phase, the reserved free capacity can be converted back to the unreserved free capacity. In a few cases, it is necessary to queue recovery messages that cannot be processed. Thus, in order to improve the recovery rate, each furnace maintains a queue to hold these recovery messages separately. The recovered recovery messages are processed again when a release message arrives in a later confirmation step.

The three-step verification process begins when the first recovery message arrives at the endpoint of the virtual path affected by the failure. The node's processor updates the Virtual Path Identifier (VPI) table, sets up an alternate route that contains the recovery message that arrived as the recovery path for that virtual path, and then generates a release message and sends it to all adjacent links. The release message includes an exceptional link list containing the starting node of the virtual path, the virtual path number, the allocated capacity, and the number of links on the alternative path. After receiving the release message, the intermediate nodes check to see if they have already processed the release message for the same virtual path, and if so, delete the release message, or execute the following procedure for all adjacent links except the link on which the release message arrived: do. If the link is included in the exception link list, the node's processor adds the allocated capacity to the operating capacity, updates the VPI routing table, and sends a recovery message to the next node through this link. If the link is not included in the exception link list, but the virtual path number is included in the list of alternative routes maintained by the node, the allocated capacity is subtracted from the reserved free capacity, added to the unreserved free capacity, and the virtual Delete the route from the list of alternate routes and send a release message to the next node. Otherwise, the release message is deleted. If there is competition, the same solutions mentioned in step 2 apply. After processing for all adjacent links, the processor applies a two-step process for recovery messages in the queue. When the node in the virtual path receives the release message, the recovery path is confirmed, so after updating the VPI routing table, the traffic is switched to this alternative path.

The virtual path dynamic protocol performed by the above procedure (1), unlike the existing dynamic method, utilizes the path recovery method for dynamic recovery, (2) explicitly includes solutions to competition, and (3) In the event of a failure, the nodes in the network perform distributed processing without the intervention of the network management center.

As can be seen from the above, the multi-stage recovery algorithm of the ATM network according to the present invention implements a pre-planned virtual path recovery method and a dynamic recovery protocol step by step considering the failure situation, and the pre-plan recovery method determines and stores the recovery path. Dynamic recovery protocol consists of three phases: fault recognition, alternative route search, and verification. If the ATM network fails, the recovery process is distributed by the protocol proposed by the node affected by the failure.

It can increase the service recovery rate of ATM network and improve the network survivability and service reliability by performing the recovery step by step, which can reduce the time and cost for disaster recovery.

The present invention described above should be seen that it has been disclosed for the purpose of illustration, those skilled in the art that various modifications, changes, additions, etc. are possible within the spirit and scope of the present invention, and such modifications, changes, additions are deemed to be within the scope of the invention It should be regarded as belonging to the following claims.

Claims (7)

In the multi-stage recovery method of ATM network for recovering failure of ATM network, (a) executing a preplanned virtual path recovery method, (b) executing a virtual path dynamic recovery method, Step (a) is (a-1) determining and storing the recovery path for determining the recovery path for each failure situation and storing information on the path and capacity in the recovery node; (a-2) message transmission and capacity allocation; (a-3) a path exchange step of exchanging traffic of a failed virtual path to a recovery path when a receiving node recovers a message; Step (b) is (b-1) the failure recognition phase to recognize the failure and initiate recovery; (b-2) route search step to find new recovery path, (b-3) A multi-stage recovery method for ATM networks, comprising the step of verifying the discovered route and replacing the virtual path affected by the failure with the newly discovered route. 2. The method of claim 1, wherein the dynamic recovery method of step (b) is performed by a flooding algorithm. The method of claim 1, The method of determining and storing the recovery path in the step (a-1) is performed by the PVRRP method. The method of claim 1, In the failure recognition step of the step (b-1), when a failure occurs in the ATM network, it recognizes the failure and notifies the failure situation to the start / end node of the virtual path affected by the failure, thereby operating a recovery process. The route search step of step (b-2) searches for a route of the repair virtual path for transmitting traffic that has flowed through the virtual paths affected by the failure, and the verifying step of step (b-3) is a searched alternative. A multi-stage recovery method for ATM networks, characterized in that the route is identified and the traffic flowing through the affected virtual paths is switched to the newly found recovery virtual path. The method of claim 1, In the message transmission and capacity allocation step of step (a-2), if a failure occurs by OAM cell or cell header error check, the message is sent to the recovery node after retrieving the relevant information from the database by the recovery node. Multi-stage recovery method of ATM network, characterized in that virtual path identification number allocation and capacity allocation in recovery node. (a) setting operational virtual path routing and spare capacity allocation information; (b) determining the scenario for the preplan recovery method; (c) determining whether to recover the operational virtual path; (d) pre-planning virtual path recovery steps; (e) determining whether to recover the virtual path; (f) virtual path dynamic recovery step, (g) ending the virtual path recovery procedure; And executing step (g) if the operational virtual path is restored in step (c), and executing step (g) if the virtual path is restored in step (e). The method of claim 6, The step (d) is performed by the PVRRP method, the step (f) is performed by the flooding algorithm.