Identification of switched Boolean networks

Abstract

This paper addresses the identification problem of switched Boolean networks (SBNs). The model considered here is more general than conventional SBNs, which allows each subnetwork to exhibit distinct state and output dynamics, simultaneously governed by switching signals. The switching mechanism influences not only the evolution of system states but also the associated output behaviors, rendering existing identification techniques for traditional Boolean (control) networks inapplicable. In this paper, the reachability and observability properties of SBNs are first analyzed, and corresponding detection conditions are proposed. Building on these properties, criteria for identifying SBNs in both single-sample and multiple-sample scenarios are derived. The exact correspondence between states and outputs is found by incorporating temporal information and then valid algorithms are developed to facilitate this identification process. A new finding is that even if some subnetworks in an SBN are unidentifiable, the entire SBN may still be identifiable. Finally, several examples are provided to demonstrate the effectiveness of the proposed methods.