Performance analysis of an arbiter using probabilistic timed Petri nets

Asynchronous circuits have experienced a resurgence due to potential advantages to low-power and high-speed design. The timed signal transition graph model, an interpreted Petri net, is a formal specification that is capable to characterize the temporal behavior of asynchronous circuits by utilizing interval delays. However asynchronous circuits are prone to metastability, which may subside after an arbitrarily long time. Thus an interval analysis yields a very pessimistic description of the performance bounds. In this paper we take a probabilistic approach based on a Petri net model to study the effect of metastability in the performance of asynchronous circuits. We illustrate our approach by modeling the Seitz’ arbiter as a case study.