Search

We study the benefits of optimal spectrum allocation in a wireless network with frequency agile radios and we present a cross-layer problem formulation for the joint routing and link scheduling under non-uniform spectrum allocation. We present a primal-dual decomposition to provide an exact solution for this complex optimization problem. Given the difficulty associated with such design, we propose a heuristic approach based on simulated annealing to solve the dual sub-problem of the decomposed model. Numerical results revealed that up to 44% improvement in network performance is obtained when variable-width spectrum band allocation is used, as opposed to the best fixed-width spectrum band allocation for larger networks. Numerical results also confirm that the primal-dual decomposition method using simulated annealing to solve the dual sub-problem, substantially reduces the computation time and achieves near optimal solutions. [ABSTRACT FROM AUTHOR]

Protection capacity re-provisioning mitigates the impact of double-link failures by provisioning new capacity for unprotected demands in a network designed to achieve 100% restorability under single-link failures. We study the performance of re-provisioning in networks under various resource sharing degrees. Intuitively, the lower is the sharability degree of resources the smaller is the number of unprotected connections resulting after the recovery from a failure. However, we show in this paper that limited resource sharability also implies limited flexibility for the network in finding capacity for unprotected demands after failures; which accordingly limits the capability of re-provisioning schemes in improving the network restorability. We further study the performance of different re-provisioning algorithms under distributed control and we show that contentions severely impact the restorability performance. Finally, we propose a simple mechanism to mitigate the effects of contentions. [ABSTRACT FROM AUTHOR]

Significant progress has been made towards making optical networks 100% restorable in the event of single link failures through protection schemes with preplanned spare capacity. Currently dual failures are considered not uncommon and finding shows that designs offering complete dual failures restorability require more than double the amount of spare capacity. In this paper, we study the impact of re-provisioning on improving the overall network robustness and we compare two different re-provisioning schemes under both centralized and distributed implementations. We show that under distributed implementation, network robustness degrades due to excessive contentions and accordingly we propose a solution to mitigate the impact of contentions. We evaluate the performance of our proposal through simulation experiments. [ABSTRACT FROM AUTHOR]