Date of Award


Document Type


Degree Name

Master of Science in Operations Research


Department of Operational Sciences

First Advisor

Yupo Chan, PhD


Prescriptive models used to allocate resources for network improvement traditionally have used reliability or flow as Measures of Effectiveness (MOEs). Such metrics do not give value to efforts which make a component more difficult to exploit. This study developed an entirely new MOE for stochastic network improvement, flow damage utility, which uses a two person, zero-sum, non-cooperative game to optimize a probabilistic network for an estimate of expected flow minus performance degradation after a worst case component loss. A multiple criteria optimization problem that uses flow damage utility and an analogous, previously developed metric for the reliability problem is used to capture the strategic competition between the network defender and attacker and shows promise of finding 'value free' improvement defensive strategies in the context of Steuer's reverse filtering as applied to the generated efficient frontier. Irrespective of unit costs of reliability vs. bandwidth improvement, a 'value free' solution may be imputed from these game theoretic models. Examples of analysis on four different networks are presented.

AFIT Designator


DTIC Accession Number