Nowadays, navigation systems rest more and more on satellite constellations (GPS, GLONASS) which provide global coverage. However, the delivered service does not match all the needs (of performances, bound to security) required by Air Transport Organisations. In order to improve this service, GNSS (Global Navigation Satellite System) program is being built, with contributions from main implied World regions (USA, Europe, Japan). The first step, GNSS1, is already under implementation : it consists in using geostationary satellites which broadcast a navigation signal (with integrity and wide area differential correction data), thus offering a better sen ice, but always complementary to and dependent on existing constellations (GPS and GLONASS);this step does not allow either to get control over the system (the two aforementioned constellations being presently under military control of single states) or to save (by withdrawal) all the classical navigation means. The next step, GNSS2, mainly aims to use signal-in-space as a navigation sole means. Definition of such a system is not yet decided I however, many studies, often driven within Europe, have led to propose new constellations for navigation satellites, which, while keeping for receivers an operational community with CPS or GLONASS, allow to be independent of the availability of both these constellations, with competitive performances. Within this framework, CNES (French Space Agency) and ALCATEL Espace (a prime manufacturer for many space telecom programs) hate considered the problem of designing orbital constellations suited for navigation needs. After a recall of economical and strategical issues, a survey is made of the main requirements (accuracy, integrity, availability, and continuity) from Air Transport Organisations. Then are given principles and methods of selection for candidate orbits matching the service : LEO (Low Earth Orbits) constellations are specially studied (with complementary geostationary satellites), relying on either dedicated satellites or additional payloads upon already available satellites, with original concepts of very light, low consumption and cheap payloads. Eventually, results from performance simulations and first components of system design are given, which allow to set a cost/efficiency; comparison for different constellations.