The Transit Network Generator (TNG) Tool is used in the GSAC DSS application which takes the technical specification of a service (e.g., communication, processing and usage requirements) and suggests the most appropriate (1)network and service resource assignments (configurations) to support the service.
The TNG Tool is used to generate the most cost effective configuration of the DESSERT Service Providers overall leased transmission capacity (2). The Tool takes a set of new technical requirements and along with the existing demands on the Transit Network (TN), it determines their most cost effective configuration. Such a configuration would contain a network topology in which demands between two nodes A and B can be routed via an intermediate node C if it is cost effective to do so ( See fig. 1). Consequently in order to optimise the SP's TN usage an appropriate network topology and a routing strategy must be found. The cost function which is assumed for the TN is of the form shown in Fig. 2.
Fig. 2 - Transit Network Cost Function
Before new requirements can be configured their appropriate bandwidth must be determined. This bandwidth is dependent upon the network technology used in the Access Network (3) (e.g. ISDN or X.25) and the TNG provides support in determining feasibility of requirements for ISDN or X.25, dimensioning the requirements for each network and in selecting which network to use.
Once the bandwidths have been chosen the network is configured using an OR-type optimisation algorithm called Accelerated Greedy. The resulting network topology and routing strategy can be viewed within the TNG.
The TNG is implemented using Rule-based and OR techniques. The rule based techniques are used for feasibility checking, dimensioning and bandwidth selection, while the OR techniques are used for dimensioning and solving the network configuration problem.
The diagram in Figure 3 shows the steps taken in the TNG to determine the minimum cost solution.
Note (1): e.g. a choice of criteria to determine appropriateness at runtime is possible (for GSAC).
Note (2): In the DESSERT case study it is assumed that the Service Provider does not own a Transit Network but instead buys the necessary transmission capacity from a third party service provider.
Note(3)This is because the teletraffic characteristics may be different for different network technologies and consequently so would the offered load to the Transit Network.
