Modeling and analyzing real-time wireless sensor and actuator networks using actors and model checking

Abstract

Programmers often use informal worst-case analysis and debugging to ensure that schedulers satisfy real-time requirements. Not only can this process be tedious and error-prone, it is inherently conservative and thus likely to lead to an inefficient use of resources. We propose to use model checking to find a schedule which optimizes the use of resources while satisfying real-time requirements. Specifically, we represent a Wireless sensor and actuator network (WSAN) as a collection of actors whose behaviors are specified using a Java-based actor language extended with operators for real-time scheduling and delay representation. We show how the abstraction mechanism and the compositionality of actors in the actor model may be used to incrementally build a model of a WSAN's behavior from node-level and network models. We demonstrate the approach with a case study of a distributed real-time data acquisition system for high-frequency sensing using Timed Rebeca modeling language and the Afra model checking tool.