Description
There has been a recent rise in interest in building networked control systems over a wireless network, whether they be for robot navigation, multi-robot systems, or traditional industrial automation. The wireless networks in these systems must deliver packets between the controller and the actuators/sensors reliably and with low latency. Furthermore, they should be amenable to modeling and characterization so they can be designed as part of a complete control system. Mesh networks are particularly suited for control applications because they provide greater reliability through path diversity.
This paper introduces tools for characterizing the end-to-end connectivity of two points in a wireless mesh network as a function of latency. In particular, we use tools derived from Markov chain models to compare end-to-end connectivity in two routing protocols running on the Data Link/MAC layer provided by Dust Network's Time Synchronized Mesh Protocol (TSMP): Directed Staged Flooding (DSF) and Dust Network's Unicast Path Diversity (UPD). These models also allow us to calculate the traffic load, the sensitivity of end-to-end connectivity to link estimation error, and the robustness of the network to node failure. The paper gives an example of how these tools can be used to evaluate the feasibility of running control applications over sensor networks.