In this paper, we propose an analytical estimation of Power Spectral Density (PSD) re-growth of non linearly amplified Multi Carrier Modulated (MCM) signals. This work has been carried in the case of Filter Bank based Multi Carrier (FBMC) systems. This kind of MCM signal has a very good frequency localization due to the sharp filter bank used. It is shown that the Solid State Power Amplifiers (SSPA) modeling has an impact on the accuracy of the PSD estimation for FBMC signals. If polynomial models are used for the SSPA modeling, it is necessary to increase the polynomial order for FBMC or OFDM signals. Indeed, the estimation of PSD re-growth proposed in this work is based on the use of a 5th order polynomial model of AMAM characteristic. Such order provides a better PSD re-growth modeling yielding to better system performance estimation in Cognitive Radio systems.