KR20030029520A - Method and system for communication of data via an optimum data path in a network - Google Patents

Abstract

Routing is one of the major concerns in data communications networks. In general, routing optimization of a particular network is realized by improving / modifying / extending the routing protocol. In order to optimize the routing of a data network without changing existing routing protocols, the present invention provides a method and system for optimizing the routing of data communicated in a network. In particular, the present invention realizes homogeneous distribution of data communication traffic in the network by minimizing the maximum link utilization and / or minimizing the average link utilization of the data link between nodes in the network.

Description

TECHNICAL AND SYSTEM FOR COMMUNICATION OF DATA VIA AN OPTIMUM DATA PATH IN A NETWORK}

In data communication within a network, the data communicated is routed through the network, i.e., data or data packets are transmitted over a data link in the network from the data transmitting node to the data receiving node. In order to correctly route data through the data link of the network and through the nodes of the network connected to the data link, so-called routing protocols representing a routing strategy or routing strategy are used. Most common routing protocols are Routing Information Protocol (RIP), Enhanced Internet Gateway Routing Protocol (EIGRP), and Open Shortest Path First (OSPF) protocol. These protocols are used as so-called internal gateway protocols for routing data within a network managed by a public network. A data network (e.g., the Internet) does not have a predetermined data path / connection between a transmitting node and a receiving node, unlike a network having a fixed communication connection (e.g., a telephone network). Internal routing protocols are generally based on how data communication from a transmitting node to a receiving node is not performed through predetermined data links and nodes in the network. Instead, the data or data package is communicated from one node to another, where each routing decision for which a data link to the next node should be selected is "make up" by each node individually. . Thus, a node called a router that transmits data to another node / router does not "know" each of the complete data paths or the data links forming that data path for data communication from the transmitting node to the receiving node. . Each node / router determines its own routing based on routing information to which the next node / router can send data or data packages. As a result, routing decisions, i.e., the determination of the next node / router to which data is transmitted, are made in part by the nodes / routers communicating the respective data.

Knowing the operating conditions of each node / router in the network, it is possible to determine the data path through which the data or data package is communicated over the network and the data link forming the data path. The operating conditions of each node / router and, in particular, the routing decision of each node / router is determined through a cost function that determines the cost of the link. The cost function depends on the protocol and may depend on the capacity of the link, the type and / or amount of data communicated, the utilization of the data link at the node where the data should be transmitted, the (physical) delay and the data communication situation of the entire network. have. In practice, the cost function depends only on predetermined and / or statistical variables such as delay and link capacity for the link once defined in the router database. Thus, if no error occurs in the network, the operating conditions of the nodes / routers of the network do not change during data communication, so that it is possible to determine the data path that data will be communicated from the transmitting node to the receiving node.

Another basic theory of the internal routing protocol described above is to always select the shortest path between the transmitting node and the receiving node. The routing protocol defines a metric in which the metric of the data path is determined and the determination of the shortest data path is distributed throughout the network. For example, the Open Shortest Path First (OSPF) protocol typically generates additional metrics based on time dependent variables such as delay for data communication over the data link and / or data communication capacity of the data link (eg, CISCO router). Used as In the OSPF metric, the cost for each data link of one node / router to another node / router branch connected to it is entered. The data communication variables of the data link between one node / router and the next node / router may depend on the direction of data communication over the data link. Thus, the cost for the data link can vary with respect to the data communication direction. The total cost of the data path is obtained by summing up all the costs of the data links included by the data path.

The metrics of the Enhanced Internet Gateway Routing Protocol (EIGRP) are more complex than the metrics of the OSPF protocol, but are mostly used in a reduced form as additional metrics.

The method of always selecting the shortest data path involves single path routing, ie the routing between two nodes / routers is the same for all data flows. Although it is possible to define multipath routing, it is rarely used for stability (e.g. loop security) and for reasons of protocol dependence (e.g. TCP degradation in case of rearranged packets). As a result of fixed cost metrics and single path routing, the burden of a data path or a single data path included in the data path may occur. This occurs, for example, if the amount of data is greater than the data communication capacity of a single data path. If the data communication capacity of the node / router is not sufficient to communicate the desired amount of data, for example if the node / router is included by the data paths of different transmitting nodes and / or different receiving nodes, Similar problems may arise.

In addition, certain data paths and / or data links may be used very often, while other data paths may only be used occasionally. Due to this non-uniform distribution of data communication over the data link, the network data link results in non-uniform link usage. In addition, the use of such non-uniform links can result in burden on different data links. Even in the case of highly used links, the network becomes inapplicable to real-time applications such as IP telephony due to the nonlinear dependence between load and latency. In addition, data routing in such a network renders the network uneconomical because no data link provided within the network requiring expensive hardware and software means is used.

Conventional methods of optimizing data routing in a network are to improve / advances existing routing protocols or to develop new routing protocols. Since modified or new routing protocols must be established across the entire network, modified or new hardware and software components are generally needed. Thus, this optimization is mainly limited due to configuration issues.

Summary of the Invention

It is an object of the present invention to solve the above-mentioned problems in the prior art. In particular, the present invention relates to the optimal routing of data in a network using existing routing protocols, i.e., without changing or relocating individual hardware and software components within the network and routing protocols already used in the network. . The present invention also provides a data routing method and system in a network that realizes the same distribution of communication / routing data or data traffic throughout the network. This method is also applicable to traffic generating in MPLS and routing in other (eg, optical / digital wrappers) optical networks.

In order to solve the above problem, the present invention provides a method for determining an optimal data path for data communication in a network defined in claim 1.

The method according to the invention is based on a method of optimizing data routing in a network based on a linear optimization problem. The linear optimization problem according to the invention is that routing / communication data in a network, also called data traffic, is distributed homogeneously / homogeneously over the network. In the present invention, the expression of "homogeneous / homogenous distribution" of data traffic should not be understood as "even distribution" of data traffic. The same distribution of data traffic across the network results in similar / same data traffic at each data link access node of the network.

Although "even distribution" of data traffic may represent an optimal solution to the linear optimization problem according to the present invention, the method according to the present invention is not particularly applicable to obtaining "even distribution" of data traffic, It relates to the optimization of data routing in a network where link utilization may vary.

The method according to the invention is used for communicating data over an optimal data path from a transmitting node to a receiving node in a network with multiple nodes connected by data links. In the following, the term "data path" is used in the data path between the transmitting node and the receiving node, where the data path is formed by each link of the data link.

In determining the optimal data path from the transmitting node to the receiving node, all loop-free data paths from the transmitting node to the receiving node through each data link and node are determined and communication between the two nodes is made. The data path is selected so that the data being communicated is over the same data path. Next, a value is determined that indicates the maximum link utilization, or the average link utilization, or a combination of the maximum link utilization and the average link utilization of the selected data path, and the minimum maximum link utilization, or the minimum average link utilization. Or a data path having a minimum value representing a combination of maximum link utilization and average link utilization is defined as the optimal data path. The data is then communicated from the transmitting node to the receiving node through its optimal data path.

In addition, the selection of the data path may include determining the data communication capacity of the data path and selecting a data path having sufficient data communication capacity to communicate data from the transmitting node to the receiving node and / or physical delay of the data path. And determining a data path with a physical delay equal to or less than the predetermined maximum physical delay.

The determination of the optimal data path from the transmitting node to the receiving node may be based on the definition of the equation system of the linear optimization problem and the solution of the equation system to define the optimal data path.

Here, the objective function may be defined to determine a value that indicates the maximum link utilization, or the average link utilization, or a combination of the maximum and average link utilization of the selected data path.

Except for objective functions, linear optimization problems require the definition of constraints. In accordance with the present invention, the transmission restriction may be defined for the determination of the non-loop data path.

In addition, routing restrictions may be defined such that all data communicated from the transmitting node to the receiving node is communicated through the same data path. This definition of routing constraints allows the selection of the shortest data path between nodes / routers to use existing routing protocols using single path routing.

In addition, capacity limits may be defined for the selection of data paths with sufficient data communication capacity, and the desired maximum data communication capacity is equal to or greater than the amount of data communicated and / or equal to or less than the predetermined maximum link utilization.

Another definition of the physical delay limit allows determining the data path with the same or less physical delay as the predetermined maximum physical delay. In the present invention, the physical delay represents the time required to communicate data between nodes of the network connected by each data link. In order to obtain the optimal data path, the system of equations for the linear optimization problem is solved by minimizing the objective function of the constraint. Solving the system of equations, all possible paths from the transmitting node to the receiving node are determined, with a minimum value representing either the minimum maximum link utilization, or the minimum average link utilization, or a combination of maximum link utilization and average link utilization. The data path is defined as the optimal data path. In particular, in a complex network including several nodes and connecting data links, it is desirable to repeatedly minimize the objective function.

For values representing a combination of average link utilization and maximum link utilization, a function representing weighted average link utilization and maximum link utilization with respect to each other may be used.

More specifically, the components of the objective function with higher influence / importance in determining the optimal data path are weighted to a higher value compared to the weight values in other components of the objective function.

For the complexity of solving the time consuming equation system, it is possible to use the final solution of the equation system determined within a predetermined time interval. For example, such a method is desirable, within a predetermined time interval, if the objective function is not feasible, if a minimum value is not determined for the objective function, or if the objective function does not converge.

Depending on the network in which the method according to the invention is used, the number of variables in the definition of the equation system is very high. In particular, the number of variables for the objective function is N ² M ² , where N represents the number of nodes / routers, and M represents the number of data links. Determined by the routing constraint, the number of variables for the constraint according to the invention is N ³ M ² size. In the case of large networks, in order to reduce the number of variables and / or the number of data links considered in the solution of the desired equation system, the method takes time for data communication from the transmitting node to the receiving node in all non-loop data paths. Includes the determination of. In addition, the data path is identified as the data path considered in the solution of the equation system in which the minimum data communication time and / or the data communication time less than the predetermined maximum data communication time is determined. Preferably, the maximum data communication time is predetermined in relation to the minimum predetermined data communication time.

For the identification of data paths that can be considered in the solution of the equation system, only the data links included in the possible data paths are used. In addition, it is not possible to consider data links in solutions of equation systems that are not included in all possible data paths.

In some networks, a preset communication time limit is used that represents the maximum time for data communication between nodes. In order to take this preset communication time limit into account, it is possible to define a physical delay limit based on the preset communication time limit. Here, the physical delay limit is determined not by the actual time required for the communication of data between nodes, but by the preset communication time limit for each data link.

Similarly, the characteristics of the network resulting from the bandwidth of the data link are taken into account by defining a physical delay limit based on the bandwidth of the data link. Instead of using the actual time it takes to communicate data over a particular data link, the bandwidth of the data link is changed with each physical delay.

Due to the fact that the amount of data communicated between the transmitting node and the receiving node may change over time, and also due to the fact that some transmitting and / or receiving nodes are communicating over the network, the optimal data path according to the invention It is desirable to continue the determination at a preset time, or at preset time intervals. Here, it is possible to determine the amount of data that is continuously communicated at a preset time, or at a predetermined time interval, where different predetermined times and for determining the optimal data path and the amount of data communicated, are determined. And / or different preset time intervals may be used.

Alternatively, it is possible to determine the optimal data path for a predetermined amount of data communicated. Preferably, the predetermined amount of data communicated is sensed or approximated.

As mentioned above, known routing protocols are based on the selection of the shortest data path for data communication from the transmitting node to the receiving node. In order to prove whether the optimal data path determined according to the invention is also the shortest data path, the data communication costs for all possible data paths over the data links forming each data path are determined. In the case where the optimal data path has the least data communication cost, the optimal data path is identified as the shortest data path.

In particular, in the case of large / complex networks, it is often possible to separate the network into at least two or partial networks, each of which determines the determination of the optimal data path separately. Due to the shortest path theory, successive solutions of smaller networks are the same as solutions obtained by optimization of one large network.

It is also possible to group at least one part of the network into a virtual node. Here, the determination of the optimal data path in the remaining part of the network including the virtual node and the determination of the optimized data path in the grouped part of the network representing the virtual node are performed separately.

The present invention also provides a system for communication of data within a network via an optimal data path. The system includes a node / router connected by a data link, at least one of which is a transmitting node, and at least one of the nodes being a receiving node, and control means connected to the network to control data communication in the network. . In particular, the control means is adapted to carry out one of the methods of the invention for communicating data via an optimal data path.

The control means may be a central control system (eg INTERNET-server) or a component of the central control system which controls the data communication of at least one node / router, preferably all nodes / routers.

The control means may also comprise at least one control unit for controlling data communication in association with one or several nodes / routers. In order to optimally control data communication and / or to control data communication of other nodes / routers, such control unit communicates with other nodes / routers, preferably all nodes / routers to obtain data communication information in the network. shall.

In particular, in a large network and / or a network representing a large amount of data communication, since the data communication control function for a single node and the data communication control function related to the entire network can be distributed to other components of the control means, The combination of the control system and the at least one control unit can improve the identification of the optimal data path.

The present invention relates to a method and system for communicating data within a network. In particular, the present invention relates to a method and system for communicating data over an optimal data path from a transmitting node to a receiving node, in a network having a plurality of nodes connected via a data link, wherein within the network The optimal data path used for the purpose of data communication is defined as the data path that will use the minimum average link and / or the minimum maximum link for the data link in the network.

1 illustrates the effect of single path routing on a data path within a network;

2 illustrates the effect of transmission restriction on a data path within a network;

3 illustrates the effect of routing limitations of the present invention on data paths within a network;

4 illustrates an example network with 6, 8, and 14 nodes / routers;

5 shows the data routing of the present invention for a six node network in minimizing average link utilization,

FIG. 6 illustrates the data routing of the present invention for a six node network in minimizing maximum link utilization, FIG.

7 is a diagram illustrating link costs for a six node network with average link utilization of minimization in accordance with the present invention;

8 illustrates link cost for a six node network with maximum link utilization of minimization in accordance with the present invention;

9 is a diagram showing a change in link utilization of the present invention in an eight-node network;

10 is a diagram showing a change in link utilization of the present invention in a 14 node network;

11 illustrates the link utilization of the present invention for selected data links of a 14 node network.

Generally, nodes in a network are computer-based systems / units (servers) capable of transmitting and receiving data over data links connected with each node. The node may act as a transmitting node from which data transmitted to the network is sent. The node can also act as a receiving node acting as a destination for receiving data communicated from each transmitting node. Also, a node can be operated as a so-called router that receives data from a transmitting node or another node / router and transmits the received data to another node / router or receiving node.

Since inter-node data links in a network may include hardware and software components necessary for the transmission of data in the data network, as well as lines for the physical transmission of electrical / optical signals representing the data on which the data links are transmitted. It is called as an interface. Such hardware and software components include conventional electric lines, data buses, buffer / memory units, processor-based units, software programs for data conversion and execution of data communication protocols (ie, routing protocols), and the like.

In order to achieve the object of the present invention, i.e. to minimize all link utilization (average link utilization), to minimize the maximum link utilization, and to maintain the physical delay of the data link between nodes in the network in a certain range. It is essential to describe in advance the network and its data traffic, i.e. the amount and flow of data communicated in the network. In this specification, three metrics are used.

First, a network with N nodes is defined by capacity metric C having size N × N, where each entry of capacity metric C represents a link capacity C _ij for each pair of nodes i and j. In the absence of a data link between two nodes, each entry in capacity metric C is set to zero.

The second metric is a traffic metric F that includes an entry f _ij for each pair of nodes i and j. Each entry f _ij represents data traffic / data flow between two nodes i and j. In the absence of data exchanged between the two nodes, each entry in the traffic metric F is set to zero. According to the method implementation of the present invention, the entry f _ij of the traffic metric F may represent a static data flow, or, for example, if the method according to the invention continues to be executed during operation of the network, then the entry f _ij of the network It can represent the actual data flow between nodes.

Third, there is a metric D that represents the physical delay of the data link. The physical delay of the data link is equal to the time required to communicate data between the two nodes over each data link. Entry d _ij of metric D represents the physical or approximate delay of the data link connecting two nodes i and j.

Based on these metrics, routing optimization according to the present invention can be defined as a linear optimization problem. As is known, the linear optimization problem is described by an equation system consisting of two parts: the objective function and the constraint. The constraint defines a multi-dimensional solution space in which the optimal solution of the objective function should be determined. The optimal solution of the objective function is, in the case of the present invention, a solution in solution space that produces a minimum value for the objective function. In order to realize the optimization of data routing in the network, other restrictions must be implemented that limit the possible solutions of the objective function.

First, for data communicated from a transmitting node to a receiving node, a data path from which data can be transmitted / routed must be determined. In the case where the transmitting node and the receiving node are not directly connected by one data link, possible data paths include not only the transmitting and receiving nodes but also other nodes and separate data links connecting the nodes. In order to avoid different data transmitted from the transmitting node to the receiving node being routed / transmitted via different data paths or data links respectively, the possible routing / data paths have been limited to so-called single path routing. Due to the single path routing constraint, exactly one data path between two nodes / routers can be defined for the possible solution of the objective function. The effect of this single path routing is shown in FIG. The shortest data path according to the routing protocol metric used between nodes / routers C and F is on node / router C or on node / router E. As a result of single path routing, both data flow AF and BF are routed first through node / router C, and then through node / router D or node / router E rather than node / router D, but the other Routed through router E.

Secondly, a capacity limit must be implemented such that the amount of data communicated over a particular data link does not exceed the capacity of each data link, i.e. the maximum amount of data that can be communicated over each data link. Since the data or data package to be transmitted may be stored, for example, in the intermediate memory of the data link and / or the node / router prior to transmission, the capacity of the data link in the network may depend on the different transmission components of the data link (ie , Bus, processor) is not limited to the physical transmission capacity. Two methods can be used to define capacity limits in terms of linear optimization problems. One method is path-oriented with respect to data links, and the other is flow-oriented. Here, the flow-oriented method was chosen to provide details because it contains fewer variables compared to the path-oriented method. After the description of the flow oriented method, a change in the representation of the path oriented method is described.

In order to apply this method to routing optimization according to the invention in combination with existing routing protocols, the routing algorithm specified by the routing protocol used has to be taken into account. As mentioned above, the routing protocols currently used, i.e., OSPF-protocol and EIGRP, are based on routing algorithms for determining the shortest data path for data routed from the transmitting node to the receiving node. These algorithms of shortest data path routing involve that all data communicated between two nodes / routers is routed over the same data link.

Objective function

The goal is to realize distribution of homogeneous / homogeneous transmission data or distribution of data traffic as much as possible throughout the entire network. In other words, all data links in the network should be used as equally as possible at the lowest possible data communication level. The flow-oriented method is applied to each data flow between nodes / routers i and j with data traffic entry f _ij ≥ 0 in data traffic metric F, and to each data link with capacity entry C _ij > 0 in capacity metric C. Boolean variable, because it is used This is introduced. This variable is set to 1 if the data flow uv from the transmitting node u to the receiving node v is routed through the data link ij between the nodes i and j, otherwise this variable is set to zero. In addition, a variable t is introduced that defines a high threshold for link utilization of all data links.

Both maximum link utilization and average link utilization should be minimized. Thus, the objective function includes two additional paths that both need to be minimized.

, C _ij > 0 (1)

The left additional portion of the objective function (1) represents the maximum link utilization, and the right additional portion of the objective function (1) represents the average link utilization. The parameter a _t of the maximum link utilization portion of the objective function (1) is used as a weighting factor for weighting the maximum link utilization portion with respect to the average link utilization portion of the objective function (1). The parameter a _t is selected in relation to the impact / importance of the small / minimized maximum link utilization for optimal data selection versus the impact / importance of the small / minimized average link utilization. For example, if the minimization of the maximum link utilization has a higher priority compared to the minimization of the average link utilization, the parameter a _t reduces the maximum link utilization by directing data traffic to the less used data link. A large value should be chosen to ensure that Comparably, the parameter a _t should be chosen to be small if the minimum impact on the optimal data path selection of average link utilization should be higher than the impact of minimization of the maximum link utilization.

Transmission limit

With the transmission restriction, for each data flow, for example, the data flow from the sending node u to the receiving node v, exactly, the non-loop data path from the sending node u to the receiving node v is the solution of the objective function (1). Variable created from It is specified by value. Here, the transmission limit includes four sub-constraints for each data flow.

1. The data flow uv from the transmitting node u to the receiving node v transmitted through the node / router is routed through only one data link starting from node / router i.

For all flow uv and all nodes / routers i (2)

2. The data flow from sending node u to receiving node v is routed through exactly one link ui from sending node u to another node / router i.

In all flow uv (3)

3. For all data flows uv from transmitting node u to receiving node v, and for each router i except source and receiving nodes u and v, of the data link to router i used by data flow uv. The number and the number of data links starting from router i are the same.

, All flow uv and all node / router (4)

4. The data flow uv from the transmitting node u to the receiving node v is routed over exactly one data link to the receiving node v.

In all flow uv (5)

The effect of the transmission restriction is shown in Figure 2 that defines exactly one non-loop data path from transmitting node u to receiving node v. According to equation (3), only data link ua connecting node u and node a carries data flow uv. One data link is directed to nodes / routers a, b, and c. Consistent with equation (4), one data link starts from each of the nodes / routers a, b, and c for the data flow uv. Conditioned by equation (5), the data path for data flow uv is directed through the data link cv between node / router c and the receiving node v and ends at the receiving node v.

Nevertheless, although this loop is inevitable by transmission limitations, it is still possible for the loop data path to be present next to the selected data path for data flow uv. However, such a loop can be avoided by minimizing the average link utilization in the objective function (1).

Capacity limit

Due to capacity limitations, the amount of data or data traffic over a data link does not exceed the capacity of each data link. The capacity limit includes two lower limits for each data link ij between nodes i and j.

The first lower constraint causes the link utilization below the fixed level defined by the parameter a _c .

For all data links ij with C _ij > 0 (6)

As set forth above, data links within a data network can coordinate data traffic in short time intervals that are greater than the physical transmission capacity of each data link. Thus, the parameter a _c is not limited to a value between 0 and 1. Setting a parameter greater than 1 is also useful for identifying a link that is the first bottleneck in the case of congestion. In the method according to the invention, the parameter a _c is set to 1 as the default value.

The second lower limit given in equation (7) defines the upper threshold for link utilization. The variable t used in the objective function (1) defines the upper bound for the utilization of all links. The parameter a _c is fixed but the variable t should be minimized. The second lower limit causes all link utilization values to be less than the value of parameter t.

For all data links ij with C _ij > 0 (7)

Routing restrictions

Routing constraints are the most important limitations because they define only this combination of data links or data paths in order to sufficiently calculate the cost in data communication / routing obtained by the solution of the equation system for linear optimization problems. Using the routing protocol described above, data is always communicated / routed over the shortest data path defined by the specific routing protocol metric for each data link. As a result, all data flows from node / router i to node / router j must be communicated / routed over the same data link between node / router i and j.

These limits are obtained by a recursive system of lower limits.

, All data flow uv, nodes / routers And in data link st (8)

, All data flow uv, nodes / routers And for data link st (9)

Considering the iterative or recursive application of equations (8) and (9), the routing between nodes / routers i and j is the same for all data flows through these nodes / routers. Thus, the solution found based on routing constraints corresponds to the implementation of a routing protocol.

3 illustrates the effect of routing restrictions. Assume that data flow ub is routed through nodes u, a, and b, data flow ac is routed through nodes a, b, and c, and data flow cv is routed through nodes c, d, and v, Due to (8) and (9), the data flow uv from transmitting node u to receiving node v must be routed through nodes u, a, b, c, d and v.

For data flow uc, two different data paths exist on node a 'or node b'. The data path comprising node a 'should not be used for data flow uc because it does not include node a which may be the opposite of equation (9). In addition, the data path including node b 'may not be used for data flow uc because it does not include node b, which may be the opposite of equation (8). In other words, data paths comprising node a 'and data paths comprising node b' are restricted in routing such that these data paths route all data flows, i.e., data flows uc, ac and bv, through the same data link. Because it does not run, it may not be used for data flow uc.

Physical delay limit

Based on the objective function (1) in combination with transmission limit, capacity limit and routing limit, data routing with shortest path routing with homogeneous, homogeneously distributed data traffic can be realized. In order to achieve the purpose of keeping the physical delay for data communication in the network within a certain range, the physical delay limit described below should be applied.

The range of possible physical delays is defined via parameter a _r , and the lowest physical limit for data flow is variable It is expressed as If P is the set of all non-loop data paths from node u to node v, and d _p is the sum of all physical delays of the data links of path p, then the minimum physical delay Is the delay of path p from node u to node v with minimum physical delay d _p .

The minimum physical delay value is obtained by calculating all physical delays of all possible non-loop data paths from node u to node v. If the parameter a _r is injected, the upper threshold May be defined to represent the maximum physical delay for each data flow.

For all data flow uv

Reduce complexity of equation system for linear optimization problem

In the solution of the objective function (1), the number of variables considered is within the size of N ² M ² , where N represents the number of nodes / routers, and M represents the number of data links. Determined by the routing constraint, the number of constraints is within N ³ M ² size. In particular, for large networks involving multiple nodes / routers and data links, the solution of the equation system to the linear optimization problem according to the invention can be very time consuming, complex or even impossible. Thus, by considering only the relevant data paths for the solution of the equation system, a so-called pre-solver is used which reduces the number of variables and constraints.

Associated data paths are identified based on each physical delay. For each data flow between two nodes / routers, all data paths that are non-loop paths are determined. For each of these non-loop data paths, the physical delay, i.e., the sum of the physical delays of the data links forming each data path, is calculated. Data path with minimum physical delay and upper physical delay threshold Data paths with lower delays are identified as related data paths. In addition, it is checked whether all data links of the related data path are included in all related data paths or none of the related data paths. The data links included in all relevant data paths can be used to route each data flow through these data links. Data links that are not included in any of the relevant data paths are not considered by the solution of the equation system of the linear optimization problem so that each data flow is not routed through these data links. Data links included in only some of the relevant data paths are considered in the solution of the equation system, and the availability of such data links is determined by the solution of the equation system.

Suboptimal routing

In some cases, for example, in a very complex network, the suboptimal (get near solution at run time) routing may be performed using a method described below, rather than using the entire equation system to calculate the routing. It may be desirable. In contrast to the above process, the maximum link utilization is not immediately minimized. Here, minimization of the maximum link utilization is realized by an iterative solution of the linear optimization problem with different objective functions of reduction and different reduced capacity constraints.

The component a _t * t of the objective function (1) for minimizing the maximum link utilization is omitted. As a result, the lower limit of the capacity limit represented by equation (7) that lowers the utilization of different data links below the upper threshold for link use is no longer necessary. On the other hand, another lower limit of the capacity limit represented by equation (6) is defined in more detail. Instead of using the default value of 1 for the parameter a _c, the parameter a _c is iteratively minimized.

Starting with a value of 1 for the parameter a _c , an equation system containing a reduced objective function is solved. The link utilization of a single data link is determined based on the known solution of the reduced equation system. In the next step, the parameter a _c is set to a value slightly smaller than the minimum determined link utilization. This process is repeated until the reduced equation system is converging. If the equation system does not converge within a predetermined time interval, or if there is no solution of the equation system, the routing determined by the final solution of the equation system is considered as a suboptimal solution of the linear optimization problem.

Specify link cost

In order to specify the data link costs for the routing determined by the equation system or the reduced equation system, the data link costs must be determined such that each of the learned and optimized data paths corresponds to the shortest path routing. This can be realized by an equation system for the linear optimization problem. In this case, the objective function is rather insignificant since the desired solution to the data link cost is obtained by an inequality system. The linear optimization problem is represented by an inequality system in combination with the objective function. Thus, the determination of the interface cost can be calculated using an equal system for the linear optimization problem. As a result, standard problem solving for the linear optimization problem can be used to determine the data link cost.

Using the above problem solving described above, all relevant / possible non-loop data paths for the data flow have been determined. For each data path except the desired / optimal data path, the sum of the data link costs for the data links forming each data path must be greater than the sum of the data link costs for the desired / optimal data path. . The inequality system that defines these limits can be very complex. Therefore, a reduction in the number of inequality equations is desirable. This can be realized because all data flows from the transmitting node to the receiving node must be routed through the same data path. Given all the data paths for the data flow, an inequality system should only be defined for data paths that contain only nodes / routers that the desired / optimal data path does not.

If the solution of such an inequality system indicates that for all data paths except the desired / optimal data path, the sum of the data link costs of the data paths is greater than the sum of the link costs of the desired / optimal data paths, the desired The optimal data path corresponds to the shortest data path.

Path-oriented method

As mentioned above, it is possible to formulate the problem as a path-oriented method. This method is identical to the flow-oriented method with the following changes.

variable

Boolean variables Is introduced. This variable is set to 1 if the data flow uv from the transmitting node u to the receiving node v is routed through the data path k, otherwise it is set to zero. k is an exponent such as u, b, ... v, where u, b, ... v are the nodes building the path.

Objective function

Similar to the flow oriented method, the objective function of the path oriented method includes two additional parts that both must be minimized.

, C _ij > 0

Transmission limit

Due to the transmission restriction, there is one non-loop data path from transmitting node u to receiving node v.

In all paths uv

Capacity limit

The sum through the flow on each link must be less than the capacity of the link.

For all data links with C _ij > 0

Routing restrictions

If X _uv defines the path from u to v and X _kv defines the path from k to v, and both paths are routed through i, then both paths are the same for paths i to k. do.

This routing restriction can be recorded as a recursive iteration through the exponential parameter k, for example the paths from u to v are routed through a, b, and c. Therefore, k is defined as k = u, a, b, c, v, the flow from a to v must use the paths a, b, c, v, and the flow from b to v is the path b, c , v, etc. This recursive iteration is also true in the opposite direction of the exponential parameter k.

example

In order to demonstrate the effectiveness of the method according to the invention, the results of the optimized routing for the exemplary network are described below. The network used is shown in FIG. The network on the right side of FIG. 4 was selected with only 6 nodes / router because its size may illustrate the final path. Due to their complex structure enabling many different data paths, networks with 8 and 14 nodes / routers were chosen. For example, in a network with six nodes / router, link capacity metrics C and data flow metrics F are shown in Tables 1 and 2.

Per optimization, the physical delay of different data links is set to 1 and the parameter a _r is set to 3. Using the above problem solving described above, for example, the number of variables for a network with 14 nodes / router has been reduced from a value of 8800 to a value of 4714. Also, the number of limits can be reduced to less than 1/2.

For all networks, optimization with some parameter settings was performed. First, optimization was performed without an effective high threshold for maximum link utilization. This is done by omitting the restriction of keeping the maximum link utilization below the value of the parameter t (starting at the "suboptimal optimization" part) and also setting the parameter a _c to a value of 10. The data link with the highest utilization of 42.9% is data link 4-3. The value of the minimized average link utilization was calculated to be 22.4%.

The parameter a _c was then set to a value of 0.4 (= 40%) which was slightly less than the calculated maximum link utilization of 42.9%. By reducing the parameter a _c that is, at a value of 0.375 (= 37.5%) and 0.36 (= 36%) repeatedly, has been found the optimum route with respect to the parameter a _c having the 0.36 value corresponding to the result of the default optimization .

In the default optimization, using the objective function (1), the parameter a _t was set to a value of 1000. The optimization result for a network with 6 nodes / routers is illustrated in FIG. 6. In addition, the most used data link is data link 4-3, but its utilization has been reduced to a value of 35.7%. As a reward, the average link utilization increased to 22.7%.

As shown in Figures 5 and 6, data flow 5-0 is routed through nodes / routers 2 and 1 away from data link 4-3. Thus, the data path for data flow 2-0 was also routed through Node / Router 1. In addition, data flows 2-4 were routed through node / router 5 instead of node / router 3 to reduce link utilization of data link 3-4 from a value of 40% to a value of 35.7%.

Based on the constant determination of the data link cost, as little data link cost as possible has been specified so that the shortest path routing is equivalent to the routing defined by the optimization described above. 7 and 8 illustrate the calculated data link costs, with each data link cost and each physical delay drawn in gray boxes. As shown in FIG. 7, the total data link cost for data flow 5-0 is three. Other possible data paths for data flows 5-9 are 5 and 4, respectively. The shortest path is the same as the desired path with minimized average link utilization.

In FIG. 8, based on the optimization regarding minimized maximum link utilization, the physical delay for the data path of data flow 5-0 is also three. The physical delays of other possible data paths through nodes / routers 4 and 3 or through nodes / routers 2 and 3 are 4 and 6, respectively. Also, the shortest data path corresponds to the desired / optimal data path.

Optimization of data routing for networks with 8 and 14 nodes / routers was performed similarly. First, average link utilization was minimized without limitation on maximum link utilization. Then, the maximum link utilization was increased repeatedly. As a result, optimization was performed to obtain the minimum possible maximum link utilization. However, a network with 14 nodes / routers is complicated to get the optimal solution. The maximum link utilization is reduced to 38.8%, but the minimum value for maximum link utilization is not known. However, the minimum link utilization should be greater than the value of 32%, for that value, since the equation system for the linear optimization problem has proved infeasible.

9 and 10, the results of routing optimization using the selected upper threshold for maximum link utilization for a network with 8 and 14 nodes / routers are shown. Maximum link utilization may be reduced, but average link utilization is rarely changed. As indicated by the bars representing the difference between link utilization of the primary and minimum utilization data links, data traffic distribution is highly homogenous with the more accurate upper threshold of maximum link utilization. In particular, the difference can be reduced to almost a quarter of a network with 8 nodes / routers (see FIG. 9). For a network with 14 nodes / routers, the optimization effect is less obvious because in such a network there is a link with approximately 1% utilization regardless of the selected upper threshold of maximum link utilization (see FIG. 10). .

Because optimization reduces the highest value for maximum link utilization, the utilization of a particular link must be increased as a reward. As shown in FIG. 11, the link utilization of the selected data link is shown for another upper threshold for maximum link utilization. The link utilization of the initially used data link is reduced against a decrease in the upper threshold for maximum link utilization. This applies to data links 9-4, 4-9 and 4-2, for example. As a result, the link utilization of data links 8-5, 5-8, and 3-1 is increased.

In a small network, it is possible to collect the necessary information to build a metric with a routing protocol (e.g. change the routing protocol or use a reserve protocol) and to calculate the optimal routing parameters within all routers. In this case, a central unit is needed.

Claims (20)

In a method of communicating data over an optimal data path in a network with a plurality of nodes (u, v, a, b, c, d), Determining an optimal data path from the transmitting node u to the receiving node v, Determining all non-loop data paths from the transmitting node u to the receiving node v through each data link and nodes u, v, a, b, c, d, Selecting a data path such that all data communicated between two nodes is communicated through the same data path, Determining a value indicating a maximum link utilization, an average link utilization, or a combination of the maximum link utilization and the average link utilization of the selected data path, A data path having a minimum value representing a minimum maximum link utilization, a minimum average link utilization, or a combination of the maximum link utilization and the average link utilization, as an optimal data path, Data communication from the transmitting node u to the receiving node v via the optimal data path Data communication method through the optimal data path comprising a. The method of claim 1, The data path selection is optimal including determining a data communication capacity of the data path and selecting a data path having a data communication capacity sufficient to communicate data from the transmitting node u to the receiving node v. Method of data communication via the data path of a PC. The method according to claim 1 or 2, The data path selection method includes determining a physical delay of the data path and selecting a data path having a physical delay equal to or less than a predetermined maximum physical delay. The method according to any one of claims 1 to 3, The optimal data path determination, Defining an equation system for a linear optimization problem to identify the optimal data path, And a solution of said equation system for defining said optimal data path. The method of claim 4, wherein The definition of the equation system includes the definition of an objective function that determines a value representing a maximum link utilization, or an average link utilization, or a combination of the maximum link utilization and the average link utilization. Data communication method. The method according to claim 4 or 5, And the definition of the equation system comprises the definition of a transmission restriction for the determination of the non-loop data path. The method according to any one of claims 4 to 6, The definition of the equation system includes a definition of routing constraints such that all data communicated between two nodes is communicated through the same data path. The method according to any one of claims 4 to 7, The definition of the equation system includes the definition of a capacity limit for determining the data path with sufficient data communication capacity. The method according to any one of claims 4 to 8, The definition of the equation system includes the definition of an optimal data path that includes the definition of a physical delay limit for determining the data path with a physical delay for each data link that is less than or equal to a predetermined maximum physical delay. Data communication method through The method according to any one of claims 5 to 9, The solution of the equation system relates to the limitation for determining all possible data paths, the minimum maximum link utilization, or the minimum average link utilization, or a combination of the maximum link utilization and the average link utilization. A method of data communication through an optimal data path comprising minimization of an objective function defining a data path with a minimum value representing the optimal data path. The method according to any one of claims 1 to 10, Wherein the combination of the average link utilization and the maximum link utilization is expressed as a function comprising the average link utilization and the maximum link utilization weighted with respect to each other. The method according to any one of claims 4 to 11, If the current equation system cannot be solved, or there is no minimization determined for the objective function, or if the objective function does not converge, The final solution of the equation system, determined within a preset time interval, identifies the optimal data path. The method according to any one of claims 4 to 12, By determining the time taken for data communication for all non-loop data paths from the transmitting node to the receiving node, Furthermore, by identifying data paths that do not take into account the solution of the equation system in which a minimum data communication time and / or a data communication time less than a preset maximum data communication time is determined, The number of variables in the equation system is reduced and / or the data link is determined in consideration of the solution of the equation system. The method of claim 13, The data link encompassed by all possible data paths is considered for the solution of the equation system, and / or A data path not covered by any one of said all possible data paths is not considered for the solution of said equation system. The method according to any one of claims 4 to 14, The physical delay limit is defined by a predetermined data communication time for the data link between the networks, and / or Wherein said physical delay limit is defined by converting the bandwidth of the data into respective physical delays. The method according to any one of claims 1 to 15, The determination of the optimal data path is carried out continuously or at a preset time or at a preset time interval and the amount of data communicated is determined continuously or at a preset time or at a preset time interval. Data communication method via data path. The method according to any one of claims 1 to 16, Determining the optimal data path is such that a predetermined or maximum amount of data is communicated. The method according to any one of claims 1 to 17, The optimal data path is proved to be the shortest data path by determining the cost for data communication over the data path, wherein the optimal data path is identified as the shortest data path if the optimal data path has the least cost. Data communication method through optimal data path. The method of claim 18, The data communication cost is determined only for data links including other nodes in the network, and / or The identification of the optimal data path is performed separately for at least two parts of the network, and / or At least one portion of the network is grouped into virtual nodes, wherein the optimal data path within the remaining portion of the network containing the virtual node and the optimal data path within the grouped portion of the network are individually identified optimal data. How to communicate data over a path. A network having a node connected by a data link, wherein at least one node is a transmitting node u and at least one node is a receiving node v; A data communication system in a network over an optimal data path, comprising control means connected to the network and controlling data communication in the network. 20. A data communication system, wherein said control means is adapted to carry out the method according to any one of claims 1 to 19.