Decentralized defence in networks
We study the problem of a planner who chooses a network of connections between node-players prior to the decentralized
Defender-Adversary game. After the links are chosen, the nodes choose simultaneously whether to protect or not after which
the adversary infects one of the nodes and the infection spreads. The objective of the designer is to maximize social welfare
while the nodes derive utility from the number of nodes they can access in the residual network, minus the cost of protection.
We study the price of stability and the price of anarchy of the whole game and show that both are equal and bounded from above by a constant. We also look at the problem with attack being random.
Joint work with Diego Cerdeiro and Sanjeev Goyal (University of Cambridge).