Dimitris Alevras, ISM Martin Grotschel, Konrad-Zuse-Zentrum fur lnformationstechnik Peter donas, E-Plus Mobilfunk Ulrich Paul, 0.tel.0 Roland Wessaly, Konrad-Zuse-Zentrum fur Informationstechnik ality which describes the behavior of the network. The switching level takes control of the grade of service (GoS), which normally is defined as the number of blocked (lost) calls in the network. The planning result of this level, which takes into account the number of switching nodes and the selected routing schemes, will be used as input for the transport level. Using multiplexing systems is key to optimizing the allocation of physical resources when mapping the demand from the switching level t o physical resources. This is because economies of scale are realized by multiplexing 64 kbls or even less (16 or 8 kb/s) channels for different applications and services (mobile communication, office data communication, corporate networks, etc.). Using SDH technology, the lowest multiplexing level is 2.048 Mb/s, which will be mapped into virtual containers. Quality of service (QoS), in terms of availability of used physical transmission systems (e.g., leased lines or microwave), is considered on this level. Survivability is considered during the design process by using the mentioned protection mechanisms at the transport level. Survivability in this context is the fraction of the demand that is satisfiable in a failure if a physical link or node fails). In this article, we consider the problem of designing a survivable telecommunications network, that is, the problem of selecting from a discrete set of capacities which one to install on each link of the physical network and deciding how to route each
In this article w e present a m -integer programming model f o r the problem of designing rvivable capacitated network, tion. The model and the solution methtool, DISCNET. Given a communication ods are integrated i n ou demand between each pair of switching nodes i n a region, the task i s t o determine the topology of a telecommunication network connecting the given nodes and t o select, from a given set of valid values, a capacity f o r each potential physical link such that the communication demands are satisfied, even if a network component fails. A solution consists of the chosen links and their capacity, as well as the routings for each demand, in the case of failure-free operation and the case of single component (node or link) failure. We suggest t w o alternative models t o deal w i t h failures of single network components. The first employs diversified paths t o guarantee the routing of a specified fraction of each demand w i t h o u t rerouting effort, the second allows rerouting in failure situations. A t the end we discuss alternative ways t o implement survivability using these t w o models.
he design, dimensioning, and administration of surT vivable telecommunications networks are becoming (i.e., networks that survive the failure of certain components) more and more important. This is because overall service quality has become a major competition criterion for telecommunications services. End-to-end survivability is not only a subject of broadband networks, as in the EC-sponsored IMMUNE project, which is part of the RACE program, but also for smaller telecommunications networks such as mobile networks, where overall service quality is eminent and vital in a highly competitive environment. However, the right balance between costs and quality must be determined by the design engineers. Certain protection mechanisms have been developed and applied in S D H technology and in digital cross-connect (DGC) systems, that is, synchronous digital hierarchy (SDH) or plesiochronous digitai hierarchy (PDH) technolob: Diverse routed protection, 1+ 1 automatic protection rings (APR), path protected switched routing (PPS), multiplexer s e c t i o n protected rings (MSP rings), and