vestal@SRC.Honeywell.COM (Steve Vestal) (03/15/91)
I would appreciate it if someone would email me references to papers that give protocols and results for implementing self-stabilizing finite state machines in a synchronous, distributed system subject to intermittent faults. Processors have point-to-point links; the network is connected; otherwise the topology is fixed but arbitrary. Message exchange is synchronous, but a processor may intermittently miss an exchange. Every processor knows what the state transition function is. The goal is to insure that all processors (almost) simultaneously transition to the same state with bounded delay after a transition is requested on some processor. Nearly simultaneous transition requests should be serialized (to the maximum extent possible). I'm interested in something more specific than the application of a general multi-phase concensus protocol. I'm interested in efficient protocols that have proofs of correctness and a reasonable approach for serializing transitions (including a discussion of when this is possible). Steve Vestal Mail: Honeywell S&RC MN65-2100, 3660 Technology Drive, Minneapolis MN 55418 Phone: (612) 782-7049 Internet: vestal@src.honeywell.com