Distributed Pinning Set Stabilization of Large-Scale Boolean Networks

TL;DR

This paper presents a distributed control method to stabilize large-scale Boolean networks towards a specific state subset, ensuring efficiency and applicability to biological networks.

Contribution

It introduces a novel distributed pinning control strategy for large-scale Boolean networks, addressing stabilization with time constraints and practical biological applications.

Findings

Successfully stabilizes large biological Boolean networks

Provides bounds on stabilization time based on network structure

Demonstrates effectiveness on real biological networks

Abstract

In this article, we design the distributed pinning controllers to globally stabilize a Boolean network (BN), specially a sparsely connected large-scale one, towards a preassigned subset of state space through the node-to-node message exchange. Given an appointed state set, system nodes are partitioned into two disjoint parts, which respectively gather the nodes whose states are fixed or arbitrary with respect to the given state set. With such node division, three parts of pinned nodes are selected and the state feedback controllers are accordingly designed such that the resulting BN satisfies three conditions: the states of the other nodes cannot affect the nodal dynamics of fixed-state nodes, the subgraph of network structure induced by the fixed-state nodes is acyclic, and the steady state of the subnetwork induced by the fixed-state nodes lies in the state set given beforehand. If the BN after control is acyclic, the stabilizing time is revealed to be no more than the length of the longest path in the current network structure plus one. This enables us to further design the pinning controllers with the constraint of stabilizing time. Noting that the overall procedure runs in an exponentially increasing time with respect to the largest number of functional variables in the dynamics of pinned nodes, the sparsely-connected large-scale BNs can be well addressed in a reasonable amount of time. Finally, we demonstrate the applications of our theoretical results in a T-LGL survival signal network with $29$ nodes and T-cell receptor signaling network with $90$ nodes.