Embedded network design to support availability differentiation

Abstract

The problem of how to provide, in a cost efficient manner, high levels of availability and service differentiation in communication networks was investigated in [1-3]. The strategy adopted was to embed in the physical layer topology a high availability set of links and nodes (termed the spine). The spine enables through protection, routing and cross layer mapping, the provisioning of differentiated classes of resilience with varying levels of endto-end availability. Here we present an optimization model formulation of the spine design problem, considering link availability and the cost of upgrading link availability. The design problem seeks to minimize the cost while attaining a desired target flow availability. Extensive numerical results illustrate the benefits of modifying the availability of a subset of links of the network to implement quality of resilience classes.

eFRADIR: An Enhanced FRAmework for DIsaster Resilience

“…An upgrade of the availability of the edges may be accomplished by having, for instance, a more reliable equipment, backup elements, efficient repair teams that may intervene quickly, if necessary. If a downgrade of the availability of the edges is deemed acceptable, it may be accomplished by allowing an increase of its time to repair by diverting resources into other edges (i.e., by changing the location of spare parts or the efficiency of teams) [70].…”

“…The selection of the edges forming the spine and their final availability may be accomplished by solving a mixed-integer linear optimization problem (MILP) as detailed in [70] and briefly described in [7]. Note that the problem in [70] is formulated to guarantee that an edge-disjoint BP exists for all demands, which is not required here.…”

“…The selection of the edges forming the spine and their final availability may be accomplished by solving a mixed-integer linear optimization problem (MILP) as detailed in [70] and briefly described in [7]. Note that the problem in [70] is formulated to guarantee that an edge-disjoint BP exists for all demands, which is not required here. Also note that the approach in [70] accepts the downgrade of the availability of the edges that are not in the spine, which allows for a reduction of the total cost.…”

“…Note that the problem in [70] is formulated to guarantee that an edge-disjoint BP exists for all demands, which is not required here. Also note that the approach in [70] accepts the downgrade of the availability of the edges that are not in the spine, which allows for a reduction of the total cost. In our approach, the availability of those edges is not changed.…”

“…Here we also consider the possibility of guaranteeing an average availability value a WP considering all the WPs in the spine. In this case, it is necessary to remove constraint [70,Eq. (22)] and add constraint 1 − 1…”