Abstract
Aho, Ullman, and Yannakakis have proposed a set of protocols that ensure reliable transmission of data across an error-prone channel. They have obtained lower bounds on the complexity required of the protocols to assure reliability for different classes of errors. They specify these protocols with finite-state machines. Although the protocol machines have only a small number of states, they are nontrivial to prove correct. In this paper we present proofs of one of these protocols using the finite-state-machine approach and the abstract-program approach. We also show that the abstract-program approach gives special insight into the operation of the protocol.

This publication has 28 references indexed in Scilit: