The consortium Thales R&T (TRT-Fr), LAAS, Institute Fresnel and IRMAR aims to go beyond the actual technological limits in the domain of phase-noise performances of microwave reference signals generated by optoelectronic oscillators (OEO) devices, in order to answer to the actual needs in telecommunication and defense. Indeed their actual evolution requires equipments that are even more embedded, compact and agile with growing bandwidth and optimized noise performances. This consortium brings together the necessary expertise for the sizing, modeling, fabrication and optimization of fibered Fabry-Perot mini-resonators with a centimeter length, delimited with innovative thin-film technology mirrors. Hence, we aim to reach with these resonators a quality factor (Q) higher than 10^8 for passive resonators and beyond by using erbium doped fiber, which will allow us to reach transparency and selective amplification regimes which are highly interesting regimes for the Q factor exaltation. Moreover, these resonators will not only answer to a compromise of finesse, but also a good reproducibility of fabrication and long term robustness of the coupling function. On top of that, the optical fibers show an excellent transparency (ranked at the second position behind crystalline material CaF2) and a large diversity of dispersive non-linear properties. Adding to this, there exists a wide scale of erbium doped amplification fibers. Hence optical fibers offer an unrivalled variety of resonant waveguides that can’t be proposed by alternative technologies developed nowadays. It is through the association of fiber technology with the worldwide excellence and know-how of Fresnel Institute in the realization of thin-film mirrors that we intend to develop an alternative technology of innovative mini-resonators showing a state of the art Q factor and a high coupling ability. We will demonstrate at TRT-Ft that the displayed qualities of such type of resonator can benefit to oscillators devices in order to synthesize 10 GHz microwave signals with state of the art phase noise performances that guarantee a simplification of oscillators systems (no need for selective RF filter, limitation of optical and microwave amplification systems). Two adaptations of the OEO device will be studied. The first one will be based on passive mini-resonator if Q factor is high enough. The other one will be based on erbium doped active resonator with an external pumping scheme that will guarantee the necessary inversion of population that is required for selective amplification. Last but not least, the state of the art know how of LAAS in the stabilization of laser sources on high Q (Q > 10^9) resonators will contribute to the qualification of the resonators made throughout the project. Then a new topology of coherent microwave frequency generation at multiples of 10 GHz will be experimentally demonstrated by the generation of Kerr combs in non-linear mini-resonators. A dedicated design of chirped thin-film mirrors adapted to the non-linear fiber properties will be studied in order to lower the Kerr comb power threshold to milliwatt range.