Distributed global state determination via graph coloring

The authors present a novel solution to distributed global state determination. The solution is based on graph edge coloring. Edge coloring makes parallelism possible via concurrent local exchanges of state information between processors by avoiding pairing conflicts. The approach assumes that all processors are synchronized. The authors observe that this approach will continue to run in the presence of faulty links and nodes, so long as the graph remains connected. Thus, this approach, inherently, is more robust in the presence of faults than methods that depend on subgraph extraction. The authors also show how the method may be applied to detect the termination of distributed processes