Jammer placement to partition wireless network

Wireless communication systems are susceptible to jamming attacks, and the use of unmanned vehicles brings new opportunities for coordinated jamming attacks. At the same time, systems of autonomous vehicles that coordinate their movements over a wireless network may be particularly vulnerable to jamming attacks that disrupt the control information. Much research has been conducted on how to efficiently jam single communication links and how to protect such links from jamming. However, less research has focused on problems of jamming attacks on the overall network. In this paper, we consider the problem of determining how to efficiently position jammers so as to partition a wireless network. The communication network is represented as a graph with the vertices representing the radios, and the edges representing the communication links. Although there has been extensive research into the problem of efficiently partitioning a graph via edge separators, the action of a jammer in a wireless network is more closely analogous to blocking reception at one or more radios, which may be modeled as partitioning a graph via node separators. We formulate several optimization problems for jammer placement. Since the optimal solution to these problems are computationally complex, we develop suboptimal solutions using spectral partitioning followed by greedy jammer placement and also a harmony search. The results show that these algorithms offer a tradeoff between complexity and performance. In the scenarios where we were able to compare performance with the optimal solution, the harmony search algorithm offered performance close to that of the optimal solution while requiring a much lower complexity.