For users to fully benefit from mobile services, technology should offer them personalised, seamless and transparent access to information focused on their needs, when they need it, taking into account their situation. Today, still some challenges have to be solved. This is mainly due to the difficulties to manage efficiently heterogeneous and unstable communication networks, as well as of limited performances of personalisation solutions in this context.
CLAIRVOYANT aims at contributing to the issue of contextaware mobile services through adaptation of information transmission to the enduser and her/his context, all along the transport chain. From a quality of service point of view, it focuses on the optimisation of a) information transmission in hybrid communication networks; b) gathering and aggregation of information in these networks; and c) filtering algorithms for contextaware provision of personalised information on mobile devices.
Network management will be addressed using mobility management techniques with regard to the selforganised and contextaware information transmission solution with the following objectives: 1) the best underlying access network for communication is selected at each given instant according to each context/situation and user preferences; 2) the best routing algorithms are chosen to establish communication, taking into account network topology changes and user mobility; 3) the best instant for information transmission is decided, mainly in the case of delay tolerant services.
A multi mobile agents system will be investigated for the abstraction of the communication network, and for the aggregation of contextual information out of the network. The two following objectives will be targeted: 1) the abstraction layer removes any dependency due to the network, providing a low coupling with the personalisation engine that facilitates the information filtering process; 2) the agent system is organised so as to optimise the information management and aggregation from any source including adhoc networks formed by terminal devices, targeting near realtime processing of requests.
On the filtering side, semantic web technologies coupled with fuzzy theory will be investigated. The first provides efficient means to formalise knowledge, defining its semantics, and to conduct reasoning on it. The second offers a framework to represent partial truth and imprecise valuations such as those represented with linguistic variables, and allows for approximate reasoning. It offers a natural way to deal with imprecise preferences in the user profile, as well as confident values of contextual information. The addressed question is the suitability of ontological knowledge processing, helped by fuzzy logic, to deal with complex contextual or userrelated data, and to provide an efficient mean for personalisation in hybrid networks environment. This means in particular dealing with network availability, mobile environments' technical constraints, user's situation gathered from multiple heterogeneous sources, etc. all leading to imprecise, partial or missing data.
The three identified axes will be investigated within the framework of a global system to be built. While algorithms for each will be elaborated and tested independently, they will be adjusted in the integrated system targeting the best average global performances. The tuned system will be assessed on an experimental use case of multimodal transport.