A Network-based Cost Comparison of Resilient and Robust System-of-Systems

Abstract

Connectivity within systems-of-systems often provides enhanced capabilities, but may also introduce new vulnerabilities. Systems engineers must therefore consider the impact of potential threats to a system, and subsequently identify methods for mitigating the effects of those threats. A common approach is to design for robustness, i.e. designing a system to be insensitive to perturbations, often through system redundancy. An alternative approach is to design for resilience, focusing on adaptation and the ability to recover lost capabilities. This paper represents systems-of-systems as networks and identifies thresholds in which a resilient network design is more cost-effective than a robust one. Resilient networks are defined with network adaptation or link rewiring. Robust networks are defined without adaptation, but instead have high network density or path redundancy. The number of initial and rewired links is used as a proxy for network cost. A command and control (C2) network of unmanned aerial vehicles performing a surveillance mission is used as an application problem. An agent-based model is used to simulate C2 network performance.