EP1175807A2

EP1175807A2 - Method for evaluating routes in a communications network

Info

Publication number: EP1175807A2
Application number: EP00922659A
Authority: EP
Inventors: Josef Rammer; Marco Conte; Gerhard Fischer; Luigi Bella; Ferial Chummun
Original assignee: Siemens AG; Nokia Siemens Networks GmbH and Co KG
Current assignee: Nokia Solutions and Networks GmbH and Co KG
Priority date: 1999-05-05
Filing date: 2000-04-20
Publication date: 2002-01-30
Also published as: WO2000069210A2; CA2373072A1; EP1058426A1; AU781148B2; NZ513978A; WO2000069210A3; AU4297600A

Abstract

The aim of the invention is to evaluate routes R in a communications network (KN) consisting of switching nodes (K) and of transmission paths (U). To this end, modified link costs (L) are established from link costs (LK) assigned to the transmission paths (U), preferably using random numbers, and the routes (R) are evaluated according to the modified link costs (L). If the modified link costs (L) are established with each call request, connections (V), which can be set-up along a number of routes (R) with identical minimal route costs (RK), are evenly distributed on these routes (R) while retaining existing routing algorithms.

Description

description

Method for evaluating routes in a communication network

Communication networks are usually designed either as packet-oriented or as line-oriented networks. Here, packet-oriented networks are more for the transmission of information without real-time character, e.g. Data, email or files are suitable, while line-oriented networks are well suited for the transmission of real-time information such as Speech or moving images are designed. In the course of the convergence of line and packet-oriented networks, however, voice and moving image information is also increasingly being transmitted in packet-oriented networks. Examples of packet-oriented networks are the Internet or ATM (= Asynchronous Transfer Mode), the term ATM sometimes also being used as a synonym for B-ISDN (= Broadband Integrated Services Digital Network). A example of ATM is the packet-oriented network technology explained in more detail below.

The packet-oriented transmission of information is characteristic of packet-oriented networks. In ATM networks, the information is divided, for example, into packets of the same length - also called "ATM cells" - which have a 5-byte cell header and a 48-byte information part (payload). The individual cells are assigned to certain information flows - also called "virtual connections" - by the cell heads. In contrast to, for example, a line-oriented TDMA method in which time slots are assigned in advance to different types of data traffic, the information streams arriving at an ATM interface are segmented into the 53-byte cells mentioned and then these cells are sequentially arranged in the order in which they were created. The multiplexing method used at TDMA is also referred to as "static multiplexing" and that used at ATM as "statistical multiplexing". As a result of the flexibility of statistical multiplexing, the information streams in ATM can have any data rates, while in static multiplexing the data rate of the individual information streams - also called "connections" - is fixed, for example, because of the fixed assignment of the time slots to the information streams 64 kbit / s with ISDN.

As a result of this difference, the route of a requested connection in packet-oriented networks is dependent on the remaining capacity available on a route, whereas in line-oriented networks it is in principle independent of the load on the individual transmission paths. For example, on a route in a line-oriented network along which e.g. According to a TDM method, 30 connections can be made in dedicated time slots with a capacity of 64 kbit / s each, and in any case a further connection can be set up if 29 connections have already been set up because the further connection does not have a higher data rate due to its constant data rate required as the remaining capacity of 64 kbit / s still available. However, along a route in a packet-oriented network with an assumed remaining capacity of 30 Mbit / s, only connections can be set up for which a data rate of less than 30 Mbit / s has been requested. However, connections with a higher data rate are rejected. If alternative routes exist, they can alternatively be set up along an alternative route with sufficient remaining capacity. To determine an alternative route, however, a new route is required.

Various routing methods are known with which routes in networks can be determined. One possibility is the so-called "source routing", in which starting from one Initial switching node, the complete route to a destination switching node is determined. For ATM networks, for example, the ATM Forum requires source routing as part of the so-called PNNI (= Private Network Network Interface) specification. The route is determined by the initial switching node and the calculated route is then transmitted to the switching nodes along the route when the connection is established with the aid of the signaling. Another possibility is the so-called "hop-by-hop routing", in which each switching node along a route recalculates the remainder or the next section of the route. This method is used, for example, on the Internet or in ATM networks without source routing.

So-called flooding methods are proposed in order to exclude those routes that use overloaded or interrupted transmission paths during routing. The traffic values of the transmission paths connected to them are measured by all switching nodes at defined times and passed on to all other switching nodes within a group. This information transfer is called "flooding". Flooding can also be initiated if the traffic values of the transmission paths change significantly - for example, if the current load of a transmission path with a total capacity of 150 Mbit / s deviates by more than 10 Mbit / s from the load that was passed on last. For example, methods are proposed in ATM networks as part of the PNNI specification, which provide a routing algorithm with the traffic values last measured in the switching nodes of the ATM network of the transmission paths directly connected to them. In connection with PNNI, see also U. Gremmelmaier, J. Püschner, M. Winter and P. Jocher, "Performance Evaluation of the PNNI Routing Protocol using an Emulation Tool", ISS 97 XVI World Telecom Congress Proceedings, pp 401 - 408 referenced. Routing in line-oriented, public telephone networks is known. The routing usually takes place in several steps, since these networks are usually hierarchically structured due to the usually large number of switching nodes. Connections in these networks are in a first step from an initial switching node on a lower hierarchy level to a switching node on the top hierarchical level, then in a second step within the top hierarchical level to a switching node representing the connection destination and finally in a third step is routed to the destination switching node on a lower hierarchy level. In this case, the first and third steps are generally effected by fixed routes or, if these are interrupted, for example, by fixed alternative routes, while the second step often merely involves the selection of the transmission path between the two switching nodes concerned at the top hierarchical level required because the switching nodes on the top level are almost completely meshed with one another. The signaling procedure no. 7 standardized for line-oriented telephone networks, however, does not support source routing, ie the initial switching node cannot forward a route calculated by it. As a result, the switching nodes along the route also do not know the route already covered, so that when this routing method is used, loops can arise in the routes in networks which are not hierarchically structured or only partially meshed, for example the Internet.

For routing in packet-oriented networks, a dynamic routing method is disclosed in German patent DE 4441356, in which blockages of transmission paths are detected and the frequency of which determines the state of occupancy of the transmission paths. Destination traffic data can be used to calculate the probability of occupancy of transmission paths offline using a routing management processor. be net. For such a calculation, example of the "Forward-Looking Routing" algorithm is ^¬, by KR Krishnan, TJ Ott in forward-looking routing, A New State-Dependent Routing Scheme, TeleTraffic Science for New cost effective system, Networks and Services , ITC-12 (1989). However, the method only takes into account connections of the same and constant bandwidth, as are typical for conventional telephone connections in circuit-switched networks, ie the bandwidth of a connection is, for example, 64 kbits / s. For packet-oriented networks such as ATM networks (Asynchronous

Transfer mode), on the other hand, a constant bit rate is the exception, because connections can be made with different and time-variable bandwidths according to the connection requests of the participants. In addition to the desired bandwidth, e.g. B. 1 Mbit / s, connection requests from participants often also contain information regarding the required connection quality.

The object of the invention is to improve the routing for packet-oriented communication networks. The object is achieved by the features of patent claim 1.

The essential aspect of the invention consists in evaluating routes in a communication network consisting of switching nodes and transmission paths, in particular packet-oriented and possibly connection-oriented communication networks, in which changed link costs are formed from the transmission costs and the routes are evaluated depending on the changed link costs. The essential advantage of the invention is that different changes in the originally allocated link costs can result in different evaluations of the routes. It is thus advantageously possible to control the evaluation of the routes by the type of changes in the original link costs, ie without changing the evaluation itself. According to one embodiment of the method according to the invention, it is provided that the changed link costs are formed by adding randomly selected real numbers to the link costs, the absolute amount of the real numbers being smaller than a maximum number which is chosen so small that the link costs are not substantial to be changed - claim 2. This results in advantageously minimally different route costs for routes that would have identical route costs without changing the original link costs. However, a route with significantly higher route costs than the optimal route costs has an optimal route, even if the original link costs change, significantly higher route costs than the then determined optimal route costs. Thus, a minimal differentiation of the route costs is advantageously effected only within a group of routes with identical route costs with unchanged link costs, while the assignment of the routes to such groups of routes with the same route costs and the order of the groups remains unchanged.

According to a development of the method according to the invention, an optimal route defined as a function of the changed link costs is determined with the aid of a deterministic routing algorithm. This has the advantage that a deterministic routing algorithm is generally less complex than a non-deterministic one and can therefore be processed more efficiently.

According to an embodiment of the method according to the invention, the deterministic routing algorithm is designed as a Dijkstra algorithm - claim 4. Thus, proven standard software can advantageously be used, since the Dijkstra algorithm has been known since 1959 and very efficient and technically sophisticated implementations are available . The optimal route also advantageously has minimal route costs. According to a variant of the method according to the invention, routes relevant only for a requested connection through the communication network are evaluated. This advantageously reduces the number of routes to be evaluated and consequently the processing time for the evaluation of the routes.

According to a development of the method according to the invention, the routes are evaluated each time a connection is requested. The change in the link costs, in particular the random selection of the real numbers, advantageously has the effect that, with several optimal routes, without a change the link costs would have identical minimum route costs, optionally one of these routes is selected for the requested connection for each connection request, although a deterministic, ie the same optimal route-determining routing algorithm is used to select the connection-optimal route without changing the link costs. This advantageously significantly reduces the blocking probability on average, since the utilization of the transmission paths takes place more evenly than if the connections were all set up along the same route

According to an application of the method according to the invention to a method for establishing a connection in a communication network consisting of switching nodes and transmission paths, it is provided that the connection is established along an optimal route for this route. Claim 7 is thus advantageous for the selection of a route Route used. In particular, because of the randomly controlled change in link costs for several comparable routes, it is open which of the routes is optimal for the connection. This means that the connections are not automatically set up over the same route, but there is a load sharing between equivalent routes. This significantly reduces the blocking rates for connections.

According to one embodiment of the application of the method according to the invention, the route which is optimal for the connection is determined by the switching node which processes the request for the connection - claim 8. This has the advantage that the request can be processed very efficiently since there are no messages between the the request processing node and another routing node are required.

In accordance with a further development of the application of the method according to the invention, all switching nodes along the route optimal for the requested connection are informed of this when the connection is established. Claim 9 is thus advantageously used in networks with source routing.

The method according to the invention is explained in more detail below with reference to several figures. It shows

1 shows a block diagram of a communication network with switching nodes and transmission paths,

FIG. 2 shows in a table all routes starting from the switching node Ki to the other switching nodes of the communication network shown in FIG. 1,

3a shows in a table the formation according to the invention of changed link costs from the link costs assigned to the transmission paths, and

3b shows the evaluation according to the invention of the routes shown in FIG. 2 as a function of the changed link costs in a table. 1 shows a communication network KN with four switching nodes K _L , 1 <= i <= 4. The switching node Ki is connected to the switching node K ₂ by a transmission path U _i2 and to the switching node K ₃ by a transmission path U _i3 ; The switching node K ₄ is connected to the switching node K ₂ by a transmission path U ₂₄ and to the switching node K ₃ by a transmission path U ₃₄ ; A transmission path U _i4 is also provided between the switching nodes Ki and K ₄

Drawing is shown dotted. This indicates that transmission paths U - for example the transmission path Ui - can be temporarily overloaded and / or interrupted. Routing information RINF (K is assigned to each of the switching nodes K. An arrow fed to the switching node Ki also indicates that a request VA for a connection V to a connection destination VZ - for example the switching node K ₄ - is transmitted to this switching node Ki.

FIG. 2 shows the routing information RINF (Ki) assigned to the switching node Ki. It contains, for example, the routes R ₁ leading from the switching node Ki to the switching nodes K ₃ , 2 <= j <= 4 and their route costs RK (Ri-,). The routes Ri are defined here as one of possibly several different possibilities, from the switching node Ki including the switching nodes K _D , 2 <= j <= 4 and the transmission paths U to the switching destination VZ - in the example the switching node K - to get. In the example, including the optional transmission path U _i4 , three routes Ri -, _, -, 1 <= k <= 3 each lead from the switching node K _x to the switching node K--, starting from the switching node Ki the route R12- 1 directly, the route R12-2 via the switching nodes K ₃ and K ₄ and the route R12-3 via the switching node K ₄ to the switching node K ₂ ; the route is R ₁₃ - ₁ through the switching node K ₂ and K _4, the route R ₂ directly _ι3 _ and the route is R ₁₃ - ₃ via the switching node K ₄ to the switching node K _3; the route R ₄ - 1 via the switching node K ₂ , the route R ₄ - ₂ via the switching _node K ₃ and the route R ₄ - ₂ directly to the switching _node K ₄ . The route costs RK (Rii-A of the routes Rι _: _ _k each result from the

Sum of the changed link costs L of the transmission paths U used in each case by the routes. In this example, for reasons of simplicity, it is assumed that all transmission paths U are bi-directional and the link costs LK are independent of the direction of the connection.

FIG. 3a shows how changed link costs L are formed from the transmission costs U assigned link costs LK as a function of randomly selected numbers EPS. For example, for the transmission paths U _X] , ij = 12, 13, 14, 24, 34, the link costs LK (U _1D ) = 1, the number EPS (U12) = 0.003, the number EPS (U _i3 ) = 0.005, the Number EPS (Uι ₄ ) = 0.012, the number EPS (U ₂₄ ) = 0.002, the number EPS (U ₃₄ ) = 0.007 and the changed link costs L (U _ι:) = LK (UA + EPS (U ₁₃ )). It should be noted that the term “link costs” should not be interpreted literally in the sense of “costs.” Any values that are relevant for the transmission paths, such as market values or quality-of-service values, can be used to form the link costs LK If a Dijkstra algorithm is used, the choice of all link costs LK equal to 1 means that those routes have optimal route costs RK that reach their connection destination VZ via as few switching nodes K as possible - this optimization metric is also called "least hops" in the technical field preferred that their connection destination VZ with the lowest delay time Iten reach, since usually the total delay time of a route R is essentially determined by the sum of the delay times of the switching through the switching node K, as long as the transmission paths U are terrestrial and not via satellites. The highest absolute amount of the numbers EPS (U _1D ) is so small at 0.012 that the changed ones Link costs L did not significantly differ from the link costs LK under ^¬, thereby maintaining the metric Least hops even if you follow the method.

In Figure 3b, the route cost RK are the routes Ri _j shown in Figure 2 - listed _k which have been determined in accordance with the given in Figure 2 for the determination of the route cost RK formula based on the specified in Figure 3a modified link costs L. Without taking into account the optional transmission path U _i , the route R _i4 -ι is the optimal route RMIN with the lowest route costs RK of all routes R. For the requested connection V to the switching node K ₄ , the route Rι-ι is also the connection-optimal route RMIN (V) , since it has the same number of hops compared to route Rι ₄ - ₂ , but has marginally lower route costs RK. Taking into account the optional transmission path Ui4, the route R ₂ - is the optimal route RMIN with the lowest route costs RK of all routes R. For the requested connection V to the switching node K ₄ , the route R14-3 is the connection-optimal route RMIN (V ), since it has one hop less than routes R14-1 and Rι ₄ - ₂ - ie the number EPS (U _i4 ) relevant for route R14-3 has by far the greatest absolute value in comparison of all numbers EPS, however, this does not change the link costs LK substantially, which means that the optimization metric Least Hops is retained.

For the exemplary embodiment, it is assumed that the switching node Ki requests the establishment of a connection V to the connection destination VZ by the request VA. This connection destination VZ is designed as the switching node K ₄ , the connection V thus as the connection V _i4 . To limit the search space of the switching node only the Ki _i4 for this compound V relevant routes R (V _i4) are evaluated, that is, the routes Rι ₄ _ι, R14-2 and Rι ₄ - _3. For these routes, the numbers EPS and then the changed link costs L are calculated using a random number generator educated. Based on these changed link costs L, the connection-optimal route RMIN (V _i4 ) is determined, for example, by a program implementing the deterministic Dijkstra algorithm, that is, taking into account the possibly overloaded and / or interrupted transmission path U _X4, the route Rι ₄ - 3 otherwise the route Rι ₄ -ι. If the state of the transmission path U _{i4 is} known, for example by communicating the state in the network with the aid of a flooding process, the consideration is made, for example, by the transmission path U _{i4 being used} for the duration of the overload and / or interruption for the route is excluded, for example by assigning him very high link costs LK compared to the link costs LK of the not overloaded and / or interrupted transmission paths U. Following the routing, the requested connection V _{i4 is set up} along the connection-optimal route RMIN (V _i4 ).

Particularly beautiful advantages result when the invention is used in connection-oriented networks with source routing, for example ATM networks. In these networks, for example, a largely uniform distribution of requested connections over several connection-optimized routes RMIN (V) can be achieved in these networks, provided that the numbers EPS are formed again regularly, for example for each connection V requested. If the numbers EPS are formed using a random generator, for example, this results in different route costs RK for the relevant routes R (V) each time. In the embodiment _4, the routes -ι Rι and R ₁₄ - ₂ was the route cost RK (R14-1) = 2,005 and RK (Rι ₄ - ₂₎ = 2.019 in. For the next requested connection V ₁₄ , for example, the route costs RK (R14-1) = 2.023 and RK (Rι ₄ _ ₂ ) = 2.004 and consequently the route Rι ₄ - _{2 could be determined} as the connection-optimal route RMIN (V _i4 ). Without changing the link costs LK, both routes R ₁₄ - ₁ , R _ι4 - _{2 would have} identical route costs RK (R14-1) = RK (R14-2) = 2. In this case, V would be due to the deterministic behavior of each connection requested Routing algorithm connect the same optimal route RMIN (V _i4 ) can be determined, e.g. route R ₁₄ - ₁ . No connections would be established along route R _i - ₂ until route R14-1 was completely occupied. A major advantage of this largely uniform distribution can be seen in the fact that, on average, this significantly reduces the probability of rejection for several connections which generally vary in their requested data rate. The probability of rejection is advantageously further reduced if a flooding method, for example the PNNI method, is additionally used in the network in order to eliminate overloaded and / or interrupted transmission paths at least during the period of the overload and / or interruption for the route.

It should be pointed out that the invention can of course be applied to any communication networks KN, in particular connectionless communication networks KN such as, for example, the packet-oriented Internet. In the Internet, for example, each individual packet is transmitted along a packet-specific route R, ie the route of each packet of a virtual connection V is independent of the routes R of the predecessor and successor packets of the same virtual connection V; The switching nodes K, which are designed, for example, as Internet routers, only determine the next switching node K for each packet of a virtual connection V - also referred to in the technical field as "hop". In accordance with the method according to the invention, successive packets belonging to the same virtual connection V are distributed over several transmission paths U by each router. In this case, the transmission paths U connected to a router are advantageously utilized evenly on average. Different runtimes of the individual packages can change the original order of the packages, for example. In this case, the original order of the packets of the virtual connection V is restored in the receiver by a higher protocol layer. Several methods are known for this, for example the Transport Control Protocol TCP.

Claims

claims

1. Method for evaluating routes (R) in a communication network (KN) consisting of switching nodes (K) and transmission paths (U), in which

- changed link costs (L) are formed from the transmission costs (U) assigned link costs (LK), and

- The routes (R) are evaluated depending on the changed link costs (L).

2. The method according to claim 1, characterized in that the changed link costs (L) are formed by adding randomly selected real numbers (EPS) to the link costs (L), the absolute amount of the real numbers (EPS) being less than a maximum number that is chosen so small that the link costs (LK) are not changed substantially.

3. The method according to any one of claims 1 or 2, characterized in that an optimal route (RMIN) defined as a function of the changed link costs (L) is determined with the aid of a deterministic routing algorithm.

4. The method according to claim 3, characterized in that the deterministic routing algorithm is designed as a Dijkstra algorithm.

5. The method according to any one of the preceding claims, characterized in that relevant routes (R (V)) are evaluated only for a requested connection (V) through the communication network (KN).

6. The method according to claim 5, characterized in that the evaluation of the routes (R) takes place with each request for a connection (V).

7. Application of the method according to one of claims 1 to 6 to a method for establishing a connection (V) in a switching node (K) and transmission paths (U) existing communication network (KN), in which the connection (V) along one for this optimal route (RMIN (V)) is established.

8. The method according to claim 7, characterized in that the optimal route for the connection (V) (RMIN (V)) is determined by the switching node (K) which processes the request (VA) of the connection (V).

9. The method according to claim 8, characterized in that all switching nodes (K) along the optimal route for the requested connection (V) (RMIN (V)) this is communicated when the connection (V) is established.