Hierarchy aware distributed plan execution monitoring

• Elaboration of a hierarchy aware and distributed monitoring approach suitable for shared information awareness.
• Aggregation of distributed nodes into clusters and cluster heads in order to localize the information exchange at the level of the distributed nodes.
• Gossip based communication across the clusters along with asymmetric clustering to reflect hierarchical relationships among participants.
• Formalization of the information sharing technique based on communicating Markov Decision Processes and probability analysis using probabilistic model checking.

Collaborative plan execution is becoming increasingly important given its potential for operational agility and cost reduction. In this paper we propose a distributed and hierarchy aware monitoring procedure for operational plan execution taking place in a dynamic environment characterized by unreliable communication and exogenous events. The contribution of this paper consists in employing a hierarchical clustering approach supporting a multi-party and hierarchy aware information sharing mechanism that is resilient to disruptions in the execution environment. The proposed distributed monitoring procedure uses asymmetric clustering to reflect hierarchical relationships along with gossip based communication across the clusters. Of significance is the information sharing mechanism formalization which utilizes a fresh information window in conjunction with communicating Markov Decision Processes. We show the usefulness of assessing shared information awareness via probabilistic model checking for various combinations of clustering topology and disruption conditions. In this context, we assess formal specifications expressed in probabilistic temporal logic and show how the model checking results can be used to derive the best fresh window value to maximize an information awareness utility function. An illustrative case study is also presented.