Reasoning about Message Passing in Finite State Environments

Abstract

We consider the problem of reasoning about message based systems in finite state environments. Two notions of finite state environments are discussed: bounded buffers and implicit buffers. The former notion is standard, whereby the sender gets blocked when the buffer is full. In the latter, the sender proceeds as if the buffer were unbounded, but the system has bounded memory and hence “forgets” some of the messages. The computations of such systems are given as communication diagrams. We present a linear time temporal logic which is interpreted on n-agent diagrams. The formulas of the logic specify local properties using standard temporal modalities and a basic communication modality. The satisfiability and model checking problems for the logic are shown to be decidable for both buffered products and implicit products. An example of system specification in the logic is discussed.

We thank Kamal Lodaya and the reviewers for helpful discussions, and Laura Semini for bringing to our attention [MS99] on which Section 5 is based.