Telecommunication networks support many essential services (e.g. smart grid communications, emergency services) and they are one of the critical infrastructures that our society depends on. Thus, communication networks must have high reliability and resilience, that is, high capacity to maintain acceptable levels of service in the face of failures and other challenges to its operation. Network resilience can be improved through prevention, network design and traffic management and restoration, so as to make a network fault (almost) unnoticeable to users. In natural disaster scenarios, providing critical services is of the utmost importance. Therefore, from a service provider's perspective, there is a need to support multiple quality of resilience (QoR) classes in a fashion similar to quality of service (QoS) classes.

For improving the availability of critical services, the approach to resilient networks design will explore a technique based on improving the availability of some of the elements in the physical layer of the network "designated as spine" which will allow to achieve the desired availability at a reduced cost. The impact of this approach in the network performance (e.g. network capacity and services availability) will have to be assessed by a mapping between the physical layer and the services layer of the network, taking advantage of the multilayer network management flexibility given by the Software Defined Networking (SDN). In particular, transport SDN (T-SDN) seeks to automate the transport network management, solving the cross-layer communication problem in the presence of faults.

In the aftermath of natural disasters, geodiverse routing is extremely important, so as to guarantee that faults occurring in the network due to disasters in a certain geographical area may be compensated by the use of other network elements in distinct geographical areas. In this context, a useful concept is that of Shared Risk Link Group (SRLG), a set of network links that share a common risk of failure.

This project relies on four main aspects: a) a strategy for improving the network availability at low cost, namely by considering a spine, and taking into account the multilayered network and the introduction of the SDN; b) the definition of risk groups in multilayered networks; c) the definition of regions especially vulnerable to natural disasters and their impact on the routing planning and design; d) the smart grid communications as a critical service.

The problems to be tackled (multi-layer mapping, use of T-SDN, the design of a high availability structure in the physical layer, the calculation of geodiverse SRLG-disjoint paths) are intrinsically multicriteria. Hence, these problems will be addressed first by using a single criterion approach (with constraints), followed by a multicriteria approach. Finally the performance of the proposed models for resilient network design and routing will be experimentally evaluated.