Control-aware scheduling over multi-hop networks

Abstract

With the proliferation of wireless networks as an indispensable component for a wide range of distributed Cyber-Physical Systems applications, the paradigm of the networking algorithms design independent from application goals abolishes. Thus, the control-aware design of the wireless resource management for wireless networked control systems (WNCSs) is shown to be more effective from the application perspective than the conventional approaches. In WNCS, the controller monitors and actuates the plant through the status updates received from the sensor over the network. This work focuses on application-aware transmission scheduling over multi-hop networks for WNCSs. As an intermediate metric, we use age of information (AoI) that captures the freshness of the data on the controller. Being a widely adopted metric for real-time applications, AoI does not consider the particular goal of control applications. Nevertheless, AoI is tightly coupled with the estimation error at the controller that, in turn, directly impacts control performance. We derive the distribution of AoI in the multi-hop network that exploits a time-varying transmission schedule. Using this distribution, we express the expected estimation error to formulate a minimization objective for the scheduling. We propose exact and heuristic methods for solving the optimization and compare different approaches to resource allocation with respect to estimation error and control costs. The proposed scheduling algorithm improves the control performance by at least 15% compared to the scheduling minimizing AoI. Introducing the schedule variability over time allows for further performance improvement by 30% in scenarios with scarce network resources.