As Network-Centric Operations increasingly become the method by which the US expects to build tactical advantage in combat, the sophistication of attacks against tactical
networks is also likely to increase. Improving robustness and resilience against attacks designed to maximize damage to overall network structures can be addressed through the application of network analytic concepts. This paper demonstrates the “macro-level” implications
of targeted versus random attacks against networks, and provides a compromise strategy wherein individual nodes prioritize reconnections to maximize network-level resiliency
with minimum additional computational requirements to existing packet- and hardware-
level operations.

This paper also discusses the relevant graph theoretic/ network analytical concepts relevant to network resiliency, such as degree (the number of connections per node), dependency / betweenness (the reliance of communications on a small number of nodes), distance / pathlength (the number of intervening nodes required to connect nodes in the network), and path redundancy / walks (the number of shortest-distance alternate routes).

Empirical findings are presented based upon monte-carlo simulations of randomly generated networks, wherein different strategies of network attacks and network reconnection strategies are measured for their impacts on the integrity of the overall networks. The implications are significant primarily for self-adapting free-space optical military and satellite networks wherein network resilience is an important consideration.

Some findings include:

• Even random, distributed free-space networks have naturally occuring traffic zones and nodes that are sensitive to removal.
• Topology-targeting attacks against distributed networks can be over twice as damaging as random attacks.
• Targeted network attacks can isolate nodes, increase routing congestion, reduce routing options and increase relay distances.
• Targeted reconnections – based upon a node-level “stitching” strategy – can be significantly more effective at healing attacked networks than alternative methods.

Dowload: Graph Theoretic Strategies Against Attacks on Free-Space Optical Networks