Consensus of linear time-invariant multi-agent system over multilayer network

Consensus of multi-agent system has been studied extensively for past decades. However, most literature considered a single communication network. In this paper, multilayer graph is introduced to allow individual agent to interact over multiple communication networks, each with different output information. Necessary conditions for achieving consensus under the usage of dynamic controller is proposed. Proposed conditions show the interplay between the system theoretical condition, such as detectability, with the graph theoretical condition, such as connectedness of the communication network. It is also shown that a dynamic controller can be designed for a given example when necessary conditions are satisfied.