next up previous
Next: Simulation Phase Up: Minimizing Sensitivity to Equipment Previous: Problem Definition: Minimizing Sensitivity

Solving the Sensitivity Problem with TIE

Solving the problem with our mechanism requires us to find values of $ \alpha(i,p,e)$ and $ \beta(i,p,e)$, for each $ i,e \in N$ and $ p\in P$, that lead to the desired egress-point selections over all graph transformations $ \Delta G$. Our solution has two main steps. First, a simulation phase determines the desired egress selection both at design time (under graph $ G$) and after each topology change (under graph $ \delta(G)$). The output of this phase is a set of constraints on the $ \alpha $ and $ \beta $ values for each $ (i,p)$ pair. Then, an optimization phase determines the values of $ \alpha $ and $ \beta $ that satisfy these constraints. For this problem, the egress-point selection for each $ (i,p)$ pair can be made independently.



Subsections

Maurico Resende 2005-10-14