We recently implemented the Libor Market Model and plan to implement the Piterbarg's FX model. It appears that FX spot/forward volatility correction is naturally expressed via log-normal volatilities of discount bonds, which in turn are naturally related to normal volatilities of short rates.We would like to use the calibrated log-normal volatilities of short rates from LMM. To this end, we could have 3-factor model: IR-IR-FX, using, for instance BK for IR. What is the best way to introduce FX spot/forward volatility correction in this environment - lognormal vols for IR ?Are there any published materials on the matter?Thank you