The problem of achieving communication between two processes across a network or an internetwork is considered. The notion of logical connectivity between processes in a protocol is formalized. The problem of constructing a protocol converter to achieve interoperability between processes that implement different protocols is addressed. A formal model is presented, based on the theory of protocol projection, for reasoning about the semantics of different protocols and conversions between them. Two kinds of converters are presented: memoryless converters and finite-state converters. The construction of some finite-state converters is illustrated, and examples are given. >