The Internet of Things (IoT) paradigm enables the development of new kind of services and mass market applications in multiple domains such as mobile marketing, social networks, smart cities, health care, and business processes. These applications consume high-level context data computed by processing and filtering numerous raw context data collected from the physical environment. Those applications are based upon cooperating services, which have to be designed and dynamically composed. For instance, those services manage heterogeneous context data, distribute both processing and flows of data, manage the quality of context for appropriate decision-making, enhance privacy of personal context data, and identify situations from data coming both from the IoT and from knowledge bases. Providing context-aware applications in the IoT necessitates the interaction of numerous heterogeneous services deployed in a multiscale complex distributed architecture. Research on IoT has first focused on the IoT foundations (smart objects, network of sensors). Based on those foundations, new services and applications are from now on possible. However the scalability challenges brought by the IoT in terms of network size, number of events, volatility, dynamicity and heterogeneity require new middleware solutions before context-aware applications can fully benefit from the IoT potential. With the Internet of Things (IoT), the middleware community is currently working on new paradigms and abstractions to face multi-scale Internet distribution