Doctoral Thesis / Dissertation from the year 2016 in the subject Computer Science - IT-Security, grade: 4, The University of Memphis, language: English, abstract: This dissertation proposes two models for designing robust networks against intelligent attackers. In both models, three stakeholders are considered: the network manager/designer, the adversary (intelligent attacker), and the network users. The frameworks of both models and some other possible models are presented in this dissertation. The first framework is a bi-objective designer model. The designer in this model has two objectives at the top level: to reduce the total system cost and to reduce the vulnerability of the network. The Sioux Falls network consists of 24 nodes and 76 links was chosen for to evaluate this framework. The decision of the designer and attacker was improving or destroying the links. Metaheuristic algorithm was used to solve the designer and attacker problems. For the user equilibrium problem, the Frank-Wolfe algorithm was implemented. The objective of the designer of the network in the first model, consist of two goals. The two goals may conflict on the amount of amount of limited available budget to be invested on the desired project/links. Therefore, a trade off solutions between these two objectives may forms. The results proved that the proposed multi-level model is able to find the Pareto front solutions for the two objectives of the designer. The second framework is a three-level zero-sum game model. In this framework, the payoffs from the designer are assumed to have the same value to the adversary entity. Therefore, the goal of this framework is to minimize the maximum gain that the adversary can achieve. An example network with 6 nodes and 16 links was used to examine this framework. The results showed that the model could be a valuable tool to reduce the potential vulnerability of networks. Other indicators of system performance can be implemented in the upper-level of this framework, in order to examine different goals. Both frameworks were tested using a medium size network with applications to larger scale networks as a future research direction.