GridStat is a flexible, robust, and secure real-time publish-subscribe middleware framework developed by TCIPG researchers at Washington State University. It is designed to provide extremely low delivery latencies and extremely high availability for power system data delivery. It forwards updates for a given sensor to many subscribers along multiple disjoint paths in order to provide low latency even in the face of failures along the paths. GridStat publishers include sensors (IEDs, PMUs, etc) as well as any other time-sensitive data that are useful to the subscribers, which are apps in the control center or decentralized agents.
GridStat is implemented as a graph (mesh network) of forwarding engines, that forward at the data level based on the rate required by downstream subscribers, and a management plane that handles subscription requests from subscribers, monitors the GridStat infrastructure, and manages cyber-security. The management plane is able to quickly change routing tables and adjust forwarding rates (shed communication load) during power or communication contingencies.
GridStat is, in essence, an overlay network that can be used to cleanly integrate disparate (and often siloed) legacy communications systems. Or to achieve minimum latency it can be configured to operate natively over Layer 2 links that it completely controls. This allows the management plane to manage the entire graph of connectivity as a coherent whole (for example, to account for the different stochastic performance and drop rates of the communication subsystems. It also allows provides the same API to publishers and subscribers regardless of the communications subsystems that GridStat overlays.
tGridStat on DETER: Presented by Dave Bakken, Washington State University, during the TCIPG Industry Workshop in October 2012.
GridStat in use with GridSim: Presented by Anjan Bose, Washington State University, at the TCIPG Industry Workshop in November 2011