Dynamic Goal Modelling for Context-Aware IoT Systems

Abstract

Context-aware IoT systems observe actions and must interpret the underlying purpose of those actions to respond appropriately. Unlike traditional software, where user actions carry explicit meaning, a single physical action in an IoT environment can serve fundamentally different purposes. A customer picking up an item from a shelf might be buying, comparing, inspecting, or merely pondering. Current requirements engineering approaches, including goal modelling, use case modelling, and other established techniques do not account for this multiplicity of purpose, and instead treat each action as serving a single, predetermined goal and specify a single system response accordingly. As a result, the problem of determining which goal an ambiguous action actually serves is left unaddressed in the requirements specification and is instead deferred to implementation, where it manifests as ad hoc, hard-to-validate interpretation logic. We introduce Dynamic Goal Modelling (DGM), a requirements development technique that addresses this problem for context-aware IoT systems. A dynamic goal arises when a single observable action can have multiple possible intents. For each such action, DGM specifies the context-awareness evidence available to the system, multiple goal hypotheses with formal conditions, precedence rules for resolving situations where multiple hypotheses are plausible simultaneously, escalation strategies, and response bindings for each interpretation. By addressing interpretation directly at the requirements level, DGM produces specifications where every feasible runtime scenario maps to a defined, traceable system response. We formalize DGM as a domain-specific language with Xtext-based tool support for modelling dynamic goals. A preliminary evaluation using two IoT applications shows that conventional specifications produce an appropriate response for only a fraction of feasible runtime scenarios, while DGM achieves a more complete requirements specification.