Asynchronous real-time distributed systems emerging in many domains are distinguished by the significant run-time uncertainties that are inherent in their application environment, system resource states, and failure occurrences [7]. Consequently, upper bounds on timing variables in such systems such as duration of computational and communication steps — manifestations of application workloads and execution environment characteristics — are not known to exist at design time with sufficient accuracy. Furthermore, many of the emerging asynchronous real-time distributed systems are also safety-critical [10, 8]. Therefore, end-users of such systems require guaranteed assurance on the delivery of desired system properties, particularly safety. This defines a certification requirement.

Asynchronous real-time distributed systems thus raise fundamental issues: "How to build timely systems that operate in the presence of uncertain timeliness? Furthermore, how to certify that such systems will deliver properties including timeliness and safety?" In this paper, we discuss an approach for constructing certifiable asynchronous real-time distributed systems.