This paper presents a resource allocation (RA) strategy for bit interleaved coded UFMC (BIC-UFMC) communications to be applied in an Internet of Things scenario characterized by short packets. The proposed RA strategy selects the best transmission parameters (TPs), i.e. code rate, bit loading (BL), power allocation (PA) and number of multicarrier symbols, by maximizing the goodput (GP) metric, defined as the number of correctly received information bits per unit of time, over a frequency-selective fading channel. The results are twofold: i) the GP performance of the BIC-UFMC system is further boosted by the best choice of the TPs compared to the case of Uniform PA and BL and a previous GP-based RA solution, named Classical RA; ii) considering the frequency offset, the combination of RA and BIC-UFMC reveals to be more robust, i.e. providing higher bit/s/Hz, than the Uniform and Classical cases.