Event-Based Admission Control Over Multi-Hop Networks with Self-Interference

In this work, we investigate the application of event-triggering in a multi-hop networked control scenario with interference constraints. In particular, we consider a line network comprised of
$H$
nodes with neighboring nodes affecting the reliability of each other, hence, introducing packet loss and non-negligible end-to-end latency. Having the practical feasibility in mind, we focus on admission control mechanisms at the sensor without assuming a centralized scheduling entity that has the perfect and global knowledge of the entire network. We demonstrate that, if the limitations of the network are neglected, the event-triggering mechanism may lead to low end-to-end reliability causing a significant degradation of the control performance. As a solution, we propose two novel admission control policies that aim to find a minimum inter-event time (MIET) in order to prevent a network congestion followed by a control performance deterioration. While the first policy follows an analytical approach combining the core principles of event-triggering and congestion control, the second policy learns the MIET adaptively without the knowledge of the network model. We show through numerical evaluation that the proposed strategies improve the control performance by more than 20% if the event criterion is selected appropriately.