Digital Twin-assisted Topology Control for Unmanned Robotic Networks under Uncertainty

Digital twin (DT) technology can assist the predictive topology control through real-time predictions and physical-virtual interactions for unmanned robotic collaboration in hazardous and unknown environments. Nevertheless, the uncertainty of topology structures caused by prediction deviations under small-sample conditions inherently complicates the topology control process and potentially result in collaboration failures. To address this challenge, we first introduce the framework of uncertain graph to model the uncertain topology and deduce the theoretical formula of global k-connectivity index that provides resilience guarantees beyond single connectivity. Considering the availability of local and uncertain topology information in real scenarios, we explore the correlation between local and global connectivity, and then formulate the local topology optimization problem under uncertainty as an uncertain graph game, in which each node tends to create or break edges with neighbor nodes based on multilateral consensus rules. By utilizing the collaborative relationships in virtual layer, we propose a DT-assisted topology control (DTATC) algorithm to construct the local multilateral stable topology in a distributed manner. Furthermore, we prove that our proposed algorithm can converge to the global stability as long as the joint physical-virtual interaction graph is connected for each node, which reveals the positive interplay between connectivity and stability. Simulation results demonstrate the effectiveness of our proposed algorithm in adapting to uncertain and dynamic environments.