WDM optical networks make it possible for the bandwidth of transport networks to reach a level on which any failures would cause tremendous data loss and affect a lot of users. Thus, survivability issues of WDM optical networks have attracted a lot of research work. Within the scope of this dissertation, two categories of problems are studied, one is survivable mapping from IP topology to WDM topology, the other is p-cycle protection schemes in WDM networks.

Survivable mapping problem can be described as routing IP links on the WDM topology such that the IP topology stays connected under any single link failure in the WDM topology. This problem has been proved to be NP-complete [1]. At first, this dissertation provides a heuristic algorithm to compute approximated solutions for input IP/WDM topologies as an approach to ease the hardness of it. Then, it examines the problem with a different view, to augment the IP topology so that a survivable mapping can be easily computed. This new perspective leads to an extended survivable mapping problem that is originally proposed and analyzed in this dissertation. In addition, this dissertation also presents some interesting open problems for the survivable mapping problem as future work.

Various protection schemes in WDM networks have been explored. This dissertation focuses on methods based on the p-cycle technology. p-Cycle protection inherits the merit of fast restoration from the link-based protection technology while yielding higher efficiency on spare capacity usage [2]. In this dissertation, we first propose an efficient heuristic algorithm that generates a small subset of candidate cycles that guarantee 100% restorability and help to achieve an efficient design. Then, we adapt p-cycle design to accommodate the protection of the failure of a shared risk link group (SRLG). At last, we discuss the problem of establishing survivable connections for dynamic traffic demands using flow p-cycle.