Systematic simplicity - Accuracy tradeoffs in parameterised contract models

Contract models underlying architecture-level verification methods must suit a range of different accuracy vs analytical complexity tradeoffs depending on domain. For example, trustworthiness in safety-critical systems is enabled by representational simplicity leading to comprehensible proofs while real-time systems require precise characterisation of execution time. A family of mutually-compatible parameterised contract models enabling such tradeoffs is needed, supporting reasoning about consistency and conformance (replaceability) which is bidirectional (from requirements to provisions and vice versa) and parametric (context-sensitive). This paper proposes a framework for such a family. The framework extends a previous formalisation of parameterised contracts. It provides more general notions of conformance, bidirectional reasoning and parameterisation, suitable for compositional architectural analyses of software products and product lines, for which software architects do not only need checking but scope for restricting or enriching service and interface contracts in predictable and compositional ways. The family of mechanisms presented here covers a range of levels of expressiveness, spanning the established four levels of component contracts, and is worked out in detail with examples for two common existing representations - tables and finite automata.