COLLABORATIVE DIAGNOSIS AND COMPENSATION OF MISBEHAVING NODES IN ACYCLIC CONSENSUS NETWORKS: ANALYSIS AND ALGORITHMS

摘要

In this paper the problem of intrusion diagnosis and compensation for a collaborative networked system with acyclic communication graph is considered. The main novel contributions of the paper are two: (a - monitoring set selection) provide necessary and sufficient conditions for the selection of a subset of monitoring nodes and (b - diagnostic and compensation algorithms) provide a diagnostic algorithm to select, exclude and compensate for a misbehaving node within the network based on a collaboration between the chosen monitoring nodes. The proposed solution has a collaborative multinode architecture with precise and easy instructions for each node of the monitoring network. Computational issues are investigated and the adaptation of the diagnostic algorithm in the presence of bounded uncertainties is analyzed. Simulations are made in order to validate theoretical results.