We consider the problem of malicious intrusion in a network of a team of agents due to a mobile jammer in the vicinity. Each agent determines its control strategy based on limited information available from its neighbors in the network graph. Under this decentralized information structure, we study the performance of the entire formation when each agent runs an estimator based on the underlying information pattern in order to compute its actions. The performance measure considered in this work is the maximum time for which the network remains connected. We show that the convergence of the estimation error of one agent depends on the estimation error of the other agents, and the control law plays a dual role in such a scenario.