The problem of detecting termination in a network which has at most one faulty node is considered and several message-efficient protocols are presented. The protocols are then extended to multiple node failures. The network is assumed to be connected in spite of the failures. The protocols are being implemented on a network of SUN workstations.
展开▼
