The ever-increasing bandwidth demand requires constant improvement in the telecommunication networks. As a result, optical transmission has become a must in the access as well, and the requirements push fixed and mobile infrastructures to converge. Prime example is the usage of access networks as mobile backhaul. This convergence, however, poses several challenges towards the network infrastructure.
One of the most important ones is service reliability as subscribers are used to a very high level of availability for voice traffic and expect the same regardless of the transmission technology. To be able to provide this service level, providers have to take network redundancy into consideration in the access as well. So far, attempts have been cost prohibitive on this domain due to high capital expenditures.
Obtaining redundancy essentially means adding extra network resources, but the degree of the duplication is one of the most decisive factors in terms of economic viability. Investing in telecommunication network equipment is costly and, therefore, such a scheme must be flexible enough to only protect the segments in need and avoid unnecessary spending. The most feasible solution, in that sense, is to only protect the feeder path, which also provides increased uptime for the demand points connecting to the distribution unit.
The task in this thesis work, thus, is to create and implement a method that is capable of designing cost-efficient protection to next-generation access networks in order to enhance their availability. The algorithm is supposed to be selective, i.e. decide whether protection is needed for the given distribution unit at all, and it must be able to design a partial or end-to-end lowest-cost alternative route to secure the prescribed availability for the feeder network.
The nature of the problem implied the use of integer programming on a parameterised graph model to compute the protection paths in a holistic approach, that is, parallel for the full network topology while minimising the overall costs. The formulated algorithm was validated on a grid-based topology, and tested on the real-life scenario of Kőszeg.
However, this algorithm has limitations in terms of scalability due to high resource requirements. As a consequence, an auxiliary heuristic algorithm is introduced to provide scalability for large-scale scenarios. This approach outputs results approximating the optimum while running in polynomial time. Its efficacy is proved by cross-matching results for the validation scenarios.
With case studies, it is showed that both the algorithms, indeed, are capable of designing on-demand protection to real-life scenarios, and, by re-using of already existing infrastructure, are able to produce much less expensive protection than would the simple duplication of the feeder segment network resources. They demonstrated huge potential for any future application in NGA topology design to raise the availability of the telecommunication network connections in a cost-effective, financially viable way.