Wireless LAN transmitter location under the threat of jamming attacks

Abstract

This paper studies the optimal placement of wireless access points in a network under the threat of jamming. We addressed this problem with a tri-level mixed-integer program. In the top level, the defender seeks to optimally place a set of capacity-limited access points to maximize total connectivity. In the middle level, an attacker seeks to optimally place a set of jammers that may be relocated between time periods to minimize total connectivity. In the bottom level, demand points seek to connect to capacitated access points such that their connections maximize their network utility. This model was examined from two viewpoints: a non-additive model in which connections were jammed if they fell within a jammer’s radius, and an additive model in which connections were jammed if enough jamming power was interfering with the connection. We proposed a solution methodology which solved a modified bi-level program efficiently via implicit enumeration and dynamic constraint generation. We showed that the addition of just one access point provided a significant increase to network connectivity, different topologies had different robustness when different utility functions were considered, and optimal jammer placement varied significantly across different topologies. Through our experiments on five topologies, we found the Spacious and Median topologies were closest to the optimal access point placement.