Robust Network Agreement on Logical Information

Abstract

Logical consensus is an approach to distributed decision making which is based on the availability of a network of agents with incomplete system knowledge. The method requires the construction of a Boolean map which defines a dynamic system allowing the entire network to consent on a unique, global decision. Previous work by the authors proved the method to be viable for applications such as intrusion detection within a structured environment, when the agent's communication topology is known in advance. The current work aims at providing a fully distributed protocol, requiring no a priori knowledge of each agent's communication neighbors. The protocol allows the construction of a robust Boolean map that is able to tolerate incorrect information spread by some agents, due to spontaneous failure or malicious intents. Effectiveness of the proposed method is shown through an implementation on a low-cost platform

Distributed motion misbehavior detection in teams of heterogeneous aerial robots

“…Although the method is general and can be applied to a wide range of applications, it has been tested with simple experiments involving real UAVs: results have been encouraging and motivate future research on this topic aiming at using the MMD as a method to monitor real systems composed of aircraft sharing the environment. Furthermore, starting from the previous experience on work [14], [34], future developments of MMD will consider the Byzantine Generals disagreement problem for the consensus approach in order to add the necessary redundancy in sensors to make the MMDS robust also to the failure of them.…”

“…The proposed idea will be formalized and proved in following section. It is worth noting that, the proposed consensus on the encoder map ϕ h (or equivalently on the events vector c h associated to agent A h ) is the novel contribution of this paper with respect to related works [12], [13], [14].…”

“…Robots need hence to combine the locally available information and reach an agreement on the reputation of the observed robot. To do this, we propose a Boolean consensus protocol that differs from those provided in [12], [13], [14]. Indeed, in this paper a consensus protocol on the events (more precisely on the encoder map defined in the following) observed by robots is proposed.…”