Disruptions and exceptions are an important source of risks in logistics, as far as the planning of transportation services is concerned. Failing to rapidly react on and handle such events may lead to serious depreciation of the transported cargo and reputation damage. The Internet of Things seems to be the technology capable of providing the tools required to detect exceptions nearly real-time. However, currently, there is little research on how to enhance the detected exceptions with related information from internal or external sources. Furthermore, most exception detection capabilities rely on experience and not much research exist on how to improve the accuracy of using third-party knowledge. In this paper, we propose a reference architecture for situation-aware logistics. The architecture specification follows the key principles derived from an extensive requirements analysis, the state of the art literature, and the ideas promoted by the Industrial Data Space initiative. The proposed architecture has been instantiated and tested by means of a prototype designed for the case of temperature-controlled transportation services.