Internet traffic has been experiencing an enormous growth in the past few decades. This growth will continue with the increasing demand for and popularity of new services such as cloud computing. In particular, cloud service providers (CSP), such as Google and Facebook, transport and replicate their data between geographically distributed datacenters using inter-datacenter wide area networks. How to improve the network throughput in order to sustain growth has emerged as an important research problem.

The traditional Wavelength-Division Multiplexing (WDM) optical networks may not provide a viable solution since their frequency-grid follows the ITU-T standards and is therefore coarse and fixed, leading to inflexibility in spectrum resource allocation and low spectrum efficiency. For example, it is a waste of spectrum to allocate 50-GHz bandwidth to a 10-Gb/s connection. To mitigate these drawbacks, Elastic Optical Networks (EONs) have been proposed finer granularity and to allow for partial spectrum overlapping. EONs have many advantages, e.g., the support of high speed transmissions beyond 100 Gb/s, and new features, e.g., distance-adaptive transmission. These features make it a promising candidate for next-generation optical transport networks. This thesis focuses on efficient designs of EONs for multicast services, as could be, required by the aforementioned data replication.

Three technologies, namely, lightpath, light-tree, and light-trail, can be utilized for the provision of multicast services. As the latter two are inherent to support optical multicasting, a multicast can be provisioned either by a single light-tree/light-trail or multiple light-trees/light-trails, each connecting the source to a subset of the destinations, while for the lightpath technology, a multicast is provisioned by lightpaths, each connecting the source to a destination. We made a comprehensive comparison among these five schemes in the context of EONs taking into account their distance-adaptive transmission capability. Numerical results illustrate their performances for a range of cases.

We then adopt the light-tree scheme as is simple but efficient, and address the problem of accommodating multiple multicast demands in EONs. For this problem, we provide a Mixed Integer Linear Programming (MILP) formulation for the small-size problems. We also propose an efficient heuristic algorithm for large instances, and compare it to existing approaches. Moreover, since heuristics accommodate demands in a sequence order and different sequences yield varied performances, we investigated the impact of sequences on the heuristic performance by proposing a couple of ordering strategies for good solutions. Numerical results show that the proposed algorithm achieves better performance than the existing ones in the various cases, and approaches the optimum obtained by the MILP algorithm.

We also consider protection in the EON design in the event of single-link failures. We provision each multicast demand by a light-tree, and protect these light-trees against any single-link failure. Given an EON and a set of multicast demands, the objective is to minimize the bandwidth requirement under the condition that all the demands are accommodated by distance-adaptive spectrum allocation. For the static traffic model, we provide a MILP formulation and propose an efficient heuristic algorithm for the protection scheme. We also consider a dynamic traffic model where a Markov-chain simulation is used to evaluate the performance of the proposed algorithm. Numerical results demonstrate the effectiveness of the proposed algorithm.