Symbolic Programming of Distributed Cyber-Physical Systems

Cyber-Physical Systems (CPSs) tightly integrate physical world phenomena and cyber aspects of computational units. The composition of physical, computational and communication systems demands different levels and types of abstraction as well as novel programming methodologies allowing for homogeneous programming, knowledge representation and exchange on heterogeneous devices. Current modeling approaches, frameworks and architectures result fairly inadequate to the task, especially when resource-constrained devices are involved. This work proposes symbolic computation as an effective solution to program resource constrained CPS devices with code maintaining strict ties to high-level specifications expressed in natural language while supporting interoperability among heterogeneous devices. Design, architectural, programming, and deployment aspects of CPSs are addressed through a single formalism unifying the specification of both cyber and physical parts of CPSs. In particular, programming patterns are modeled as sequences of words adhering to natural language syntax and semantics. Given a software under test (SUT), i.e. an input program expressed as a natural language sentence, formal specifications are used to generate oracles for sentence verification and to generate input test cases. The choice of natural language inspired programming supplies a mechanism for the development of the same software on different hardware platforms, ensuring interoperability among heterogeneous devices. Formal specifications also permit to generate stress tests in order to verify that program components behave as expected in repeated execution. In order to make high-level symbolic programs run on real hardware devices with no loss of expressivity during the translation of high-level specifications into an executable implementation, this work proposes a novel software architecture, Distributed Computing for Constrained Devices (DC4CD), as a supporting platform. The proposed architecture enables symbolic processing and distributed computing on devices with very limited energy, communication and processing capabilities that can be integrated into CPSs. In particular, DC4CD has been extensively used to test the symbolic distributed programming methodology on Wireless Sensor Networks (WSNs) that include nodes with actuation abilities. The platform offers networking abstractions for the exchange of symbolic code among peer devices and allows designers to change at runtime, even wirelessly on deployed nodes, not only the application code but also system code.