Multiservice networks routing methods deal with multiple, heterogeneous, potentially conflicting QoS parameters. The associated routing problems involve the selection of a chain of network resources satisfying certain technical requirements and seeking to optimise relevant metrics. This is particularly relevant in resilient routing methods, involving the calculation of an active path and, at least, one "back-up" or protection path to be used in case of failure of the active path. In commonly used approaches the path calculation problem is formulated as a shortest path problem with a single objective function, corresponding either to a single metric or to a function encompassing different metrics, while QoS/QoE requirements are incorporated by means of additional constraints. In the case of resilient routing the calculation of pairs of disjoint paths seeking to optimise a certain metric and satisfying certain constraints, is necessary. This is the usually proposed type of approach designated as constrained-based QoS routing. As an alternative we think that multicriteria approaches are potentially advantageous by enabling a fully consistent treatment of the trade-offs among the different metrics. This desideratum can be achieved by developing multicrieria models and the associated resolution algorithms, enabling the calculation and selection of nondominated (trade-off) routing solutions acc. [Climaco et al., 05]. The main objectives of the project are the development of new multicriteria routing approaches dedicated to IP/MPLS and WDM networks and the study of model specification and model resolution (that is the resolution of associated optimisation problems in a multicriteria sense) as well as the assessment of performance issues associated with multicriteria resilient routing. Ultimately we will seek the development and evaluation of an approach to routing which is innovative in terms of the underlying methodological paradigm in the sense described in a broadest conceptual context in [Wierzbicki, 05]. The application focus is on MPLS/IP and WDM optical networks with wavelength converters [Zhang et al., 04] and is justified by the increasing relevance of these two types of network technologies for multiservice information transfer and physical transport, respectively, and their inherent capabilities. In the case of MPLS networks the impact of the establishment of an explicit route and a reservation route in terms of load balancing costs in the routes and the minimisation of delay bounds along the routes [Rosenbaum et al., 05] will be taken into account through appropriate objective functions. In the case of optical networks we will take into account the fault probabilities in the links and their impact in terms of loss of optical connections, through appropriate cost functions for the active and protection paths. The project itself is focused on the development of new multicriteria routing models and algorithms for solving the models and on network performance analysis (in a multicriteria sense) using the developed methods. This will require the implementation of analytic-numerical models for performance analysis and specialised discrete-event simulation platforms that will also enable the comparison of network performance when using other proposed routing approaches. At methodological level particular attention will be paid to the development of exact algorithms, namely multicriteria shortest path algorithms, multicriteria shortest pairs of disjoint path algorithms or meta-heuristics dedicated to specific multicriteria routing problems and to multicriteria methods for ordering and selecting routing solutions. The project will involve INESC-Coimbra as prime contractor, a participation of Nokia-Siemens Networks (Portugal) and a small participation of the Centre of Operational Research of the Univ. of Lisbon. The project will enable the gathering and articulated working of an interdisciplinary research team, including 7 PhDs with expertise in teletraffic engineering problem modelling, network reliability analysis and algorithms, and/or multicriteria analysis/optimisation. Finally if approved the project will enable (beyond its direct results in terms of publications, specialised software development and post-graduate formation) the deepening of the collaboration between two research units and a reference telecom company, namely INESC Coimbra, responsible for the execution, the Centre of Operational Research of FCUL and Nokia-Siemens Networks as well as a significant reinforcement of the already existing intervention capacity of INESC Coimbra with telecom companies/operators in the teletraffic engineering area.
Teresa Gomes 2013-12-28