Controllability backbone in multiagent networks

Obaid Ullah Ahmad

doi:10.1016/j.automatica.2026.113138

Controllability backbone in multiagent networks

2026-06-01

0

PRE

AI

DOI：10.1016/j.automatica.2026.113138

Original

Original request for help

Save

Abstract

En 中文

This paper introduces the concept of controllability backbone and studies its application in dynamical networks defined over graphs. The controllability backbone represents a small subset of edges in a given network that ensures any subnetwork containing those edges retains at least the same level of controllability as the original network, given that the same input (or leader) nodes are used. We specifically focus on strong structural controllability (SSC), which is a topology-dependent property that remains unaffected by edge weights. However, SSC is computationally challenging to evaluate. To address this, we employ two well-known lower bounds on SSC: one based on the zero forcing notion and the other on graph distances. We develop efficient algorithms to compute controllability backbones that preserve these bounds and provide a thorough analysis of their theoretical properties. Specifically, we characterize the edges that must be included in any valid backbone and analyze the edge count of backbones derived from different methods. Moreover, we compare zero forcing-based and distance-based backbones, highlighting their relative advantages. We also numerically evaluate our methods on various graphs. (c) 2026 Elsevier Ltd. All rights are reserved, including those for text and data mining, AI training, and similar technologies.

Keywords:

Strong structural controllability

Network control

Backbone

Zero forcing

Graph distances

Controllability backbone in multiagent networks

Abstract

Journal

Organization

Cited Papers