Detection and Isolation of Failures in Linear Multi-Agent Networks
For designers of multi-agent networks, this work provides a method to detect and isolate link failures using strategically placed sensors, but the results are demonstrated only through examples and simulations without quantitative benchmarks.
This paper establishes a relationship between link failures in linear multi-agent networks and jump discontinuities in the output derivatives, and proposes a sensor placement algorithm to detect and isolate any link failures based on these discontinuities.
In this paper the focus is on the relationship between the occurrence of failures in a (directed or undirected) network of linear single integrator agents and the presence of jump discontinuities in the derivatives of the network output. Based on this relationship, an algorithm for sensor placement is proposed, which enables the designer to detect and isolate any link failures across the network, based on the jump discontinuities observed by the sensor nodes. These results are explained through elaborative examples and computer experiments.