Intruders capture in wk-r networks

Software mobile agents are a modern computing paradigm offering new methods to deal with various problems in a distributed environment. In this report we consider the problem of capturing a malicious entity moving along a network. We devise a strategy to capture the entity in a particular network, the WK-R(3,L), using a team of collaborating agents. WK-R networks have some properties interesting "per se" and useful to understand how to deal with other networks. The malicious entity, the "intruder", has the ability to move along the network knowing the positions of the agents, the "defenders". The agents move asynchronously along the links of the network and can see the state of the neighbour nodes. They "disinfect" the nodes infected by the intruder with a monotone strategy, namely a node disinfected is not infected anymore. The disinfection strategy is fully distributed and optimal with respect to number of required agents, number of moves and time steps. We devise also a strategy where the agents have not visibility of neighbour nodes: in such a case a leader seems necessary, even if the number of agents doesn't vary.