The timing of an array of millisecond pulsars (PTA) acts as a galactic-scale detector to observe gravitational wave (GW) sources in the nHz frequency range. The goal of this project is to detect low-frequency GWs while maximizing the scientific output of the Nançay radio telescope (NRT) and participating in operations of the South African MeerKAT radio telescope. We propose new data analysis methods to detect GWs emitted by multiple supermassive black hole binaries in eccentric orbits, while modeling pulsar and noise properties. We will develop a new state-of-the-art pulsar observing backend, to achieve coherent de-dispersion over a very large frequency range (1.5 to 3.5 GHz) and substantially increase our sensitivity at these key radio frequencies. The expected results are the potential first ever detection of GWs in the nHz domain, a much improved understanding of millisecond pulsars and of the weak perturbations that affect their timing stability, and new tests of General Relativity. This project is based on the long-term know-how of a composite team, made of radio astronomers who are specialists in pulsar timing, in Bayesian techniques and in GW data analyses. The NRT already produces high cadence pulsar data with a dedicated state-of-the-art backend enabling us to analyze data from the telescope’s L-band receiver (1.1-1.7 GHz) optimally. The new instrumentation will allow us to cover the whole band accessible with its high frequency S-band receiver, bringing a leap in sensitivity in a domain where the observed radio signal from pulsars is much less affected by interstellar medium perturbations. The support from the ANR will provide us with the resources to fully participate in the scientific exploitation of today’s best radio telescope in the Southern hemisphere, MeerKAT, and allow us to extend the sky coverage for GW searches. This involvement will also firmly install our French team in the long-term preparation of the SKA (Square Kilometer Array) project, which has just been included in the national road map. This is a unique opportunity to train future radio astronomers on one of its key science programs. We will also benefit from our engagement in the LISA project, sharing the expertise accumulated in both communities and building new GW detection algorithms at the interface between both projects, implementing more sophisticated and realistic GW models, and introducing machine learning in the trans-dimensional Bayesian analysis. The PTA technique indeed gives us access to a frequency domain complementary to those covered by Virgo-LIGO and LISA, where one expects GW emission from sources such as super massive black hole binaries (SMBHBs) formed in the long process of galaxy aggregation, and also stochastic GWs from the cosmological background generated by inflation in the very early Universe or by a network of cosmic strings. Characterizing those individual sources (parameters, rate, sky distribution, etc...) would yield unique information about the formation and evolution of SMBHBs through cosmic history and bring original constraints on the hierarchical galaxy formation. Moreover, the detection of this GW signal will allow us to refine the prognoses for SMBHB mergers in the LISA band. To reach these goals, we need a dedicated post-doc involved in both MeerKAT pulsar timing observations and in the combination of data from all radio telescopes involved in the International Pulsar Timing Array organization. Developing and implementing new data analysis techniques at the interface between PTA and LISA is a great project for a PhD student, who will take advantage of the APC environment and get unique skills for the future exploitation of LISA and SKA data. Finally, a strong involvement in MeerKAT and the availability of a wide-band pulsar instrumentation at NRT will clearly maintain the French radio telescope in the race up to the SKA era (>2025) and strengthen our position in the SKA Pulsar Science Working Group.