The architecture of the ICT infrastructure for supporting Intelligent Transportation Systems (ITS) is purely hierarchical, with sensed data flowing from the leaves (i.e., road-side or vehicle-installed sensors) to the root (i.e., the traffic management centre). The current approach does not scale adequately with the inclusion of a significant number of new elements, is not flexible in supporting an incremental growth or changes of the ITS, and exhibits latency and security issues. In ICSI we tackle all these issues by proposing a new architecture where the intelligence for sensing and actuation is distributed over some of the elements, called gateways, which host a software platform for running ITS applications, using the local storage and computation capabilities available. Communication with the remote centre happens only for the transmission of aggregated data for long-term operations, e.g., data mining, software upgrades, and logging. The approach proposed in ICSI enables scientific and technological innovations: advanced sensing algorithms will be defined, which make use of real-time availability of data; efficient distribution of context-rich data lays the foundations for novel traffic and travel management strategies. Both directions will be studied in the project. However, research challenges are associated at all levels to the realization of the system, especially for the communication among sensors, gateways, and vehicles, which are fully addressed in the project activities. Prototypes of sensors, road-side units, and communication units suitable for the cooperative operation envisaged in ICSI will be developed and integrated into an end-to-end demonstrator, which will be used in on-field experiments for the use cases of smart urban traffic management and accident recovery in highway.